This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
contains all the multiple overlap values between MOs of A and B.
may be then expanded into terms containing an increasing number of multiple overlaps (or exchanges of orbitals): P = P1 + P2 + P3 + P4 +K is the average system dipole moment per unit volume along the direction of E0, ε can be calculated using the following expression: 4π P ε − 1 2ε rf + 1 = 9 E 0 3 2ε rf + ε
[8.59]
The exchange energy may be expanded into increasing powers of MO overlap integrals S: EX = EX (S 2 ) + EX (S 3 ) + EX (S 4 )+K
[8.60]
EX (S 2 ) = VP2 − P2
[8.61]
with
This first term is sufficient for modeling intermolecular potentials. The overlaps depend on the monomer separation roughly as exp(-αR) so the following terms give small contribution at large-medium distances. At short distances the positive exchange term rapidly increases. Reasons of uniformity with the other terms of the interaction potential sug-
448
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
gest replacing the exp(-αR) dependence with a highly negative power of R: R-12 in most cases. The exchange terms describe what is often called the steric repulsion between molecules (actually there are other short range repulsive contributions, as the electrostatic penetration components not described by the multipole expansion). The other terms Charge transfer contributions (CT) have been rarely introduced in the modeling of interaction potentials for liquids. Some attempts at modeling have been done for the study of interactions leading to chemical reactions, but in such cases, direct calculations of the interaction term are usually employed. For curious readers, we add that the tentative modeling was based on an alternative formulation of the PT (rarely used for complete studies on the intermolecular interaction, because it presents several problems) in which the promotion of electrons of a partner on the virtual orbital of the second is admitted. The formal expressions are similar to those shown in eq. [8.40]. For the second order contributions we have −∑
Ψ0a Ψ0b |VΨΚa Ψ0b
2
E Ka − E 0a
K
The sum over K is strongly reduced (often to just one term, the HOMO-LUMO interaction), with K corresponding to the replacement of the occupied MO φ0r of A with the virtual MO φvt of B (A is the donor, B the acceptor). The expression is then simplified: the numerator is reduced to a combination of two-electron Coulomb integrals multiplied by the opportune overlap. The CT contribution rapidly decays with increasing R. COUP contributions, obtained as a remainder in the variational decomposition of ∆E, are not modeled. The contributions are small and of short-range character. A conclusive view In this long analysis of the modeling of the separate components we have learned the following: • All terms of the decomposition may be partitioned into short- and long-range contributions. • The long-range contributions present problems of convergence, but these problems may be reduced by resorting to multicenter distributions based on suitable partitions of the molecular systems. • The short-range contributions are in general of repulsive character and are dominated by the EX terms. The effect of the other short-range contributions is strongly reduced when multicenter expansions are used. Adapting these remarks, one may write a tentative analytical expression for ∆E(R): ∆E ≈ ∑ ∑ ∑ Crt( n ) Rrt− n r
t
[8.62]
n
The expansion centers (called “sites”) are indicated by the indexes r and t for the molecule A and B respectively. The index n indicates the behavior of the specific term with respect to the intersite distance Rrt; each couple r-t of sites has a specific set of allowed n values.
8.4 Two-body interaction energy
449
In such a way it is possible to combine sites corresponding only to a point charge with sites having more terms in the expansion. For two sites r-t both described by a point charge -1 the only interaction term is C(1) rt R rt , while for a couple r-t described by a point charge and a dipole, the contribution to the interaction is limited to the R-2 term, and, if it is described by two dipoles, to the R-3 term. The last example shows that eq. [8.62] is cryptic or not well developed. In fact, if we examine eq. [8.12], which reports the interaction energy between two dipoles, it turns out that the dependence on the orientation angle θ apparently is missing in eq. [8.62]. The coefficients C(n) rt must be specified in more detail as given in equations [8.50] and [8.52] by adding the opportune l, l' and m indexes (or the corresponding combination of Cartesian coordinates), or must be reduced to the isotropic form. This means to replace the three components of a dipole, and the five components of a quadrupole, and so on, by an average over all the orientations (i.e., over the m values). There is a loss of accuracy, very large in the case of two dipoles, and of decreasing importance in passing to higher multipoles. The loss in accuracy is greatly decreased when a multi-site development is employed. Many analytical expressions of interaction potentials use this choice. In such cases the C(n) rt coefficients are just numbers. The coefficients C(n) rt can be drawn from experimental values (but this is limited to the cases in which there is one site only for both A and B), from ad hoc calculations, or by a fitting of ∆E values. In the past, large use has been made of “experimental” ∆E values, derived, e.g., from crystal packing energies, but now the main sources are the variationally computed QM values, followed by a fit. The use of expression [8.62] for the fitting of QM values represents a remarkable improvement with respect to the past strategy of reaching a good fitting with complete freedom of the analytical form of the expression. There are in the literature interaction potentials using, e.g., non-integer values of n and/or other analytical expressions without physical meaning. This strategy leads to potentials that cannot be extended to similar systems, and that cannot be compared with the potential obtained by others for the same system. Expression [8.62] has several merits, but also several defects. Among the latter we note that given powers of R may collect terms of different origin. For example, the term R-6 describes both the induction dipole-induced dipole interaction and the dispersion dipole-dipole interaction: two contributions with a different sensitivity with respect to changes in the molecular system. Theory and computational methods both permit getting two separate C(6) rt coefficients for the two contributions. This is done in a limited number of potentials (the most important cases are the NEMOn26 and the ASP-Wn27 families of models). The most popular choice of a unique term for contributions of different origin is motivated by the need of reducing computational efforts, both in the derivation and in the use of the potential. To fit a unique C(6) rt coefficient means to use ∆E values only: for two coefficients there is the need of separately fitting IND and DIS contributions. 8.4.5 THE RELAXATION OF THE RIGID MONOMER CONSTRAINT We have so far examined decompositions of two-body interaction potentials, keeping fixed the internal geometry of both partners. This constraint is clearly unphysical, and does not correspond to the naïve model we have considered in the introduction, because molecules always exhibit internal motions, even when isolated, and because molecular collisions in a liquid (as well as in a cluster) lead to exchanges of energy between internal as well as external degrees of freedom.
450
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
It may be convenient to introduce a loose classification of the origin of changes in the internal geometry of molecules in clusters and liquids, and to use it for the dimer case we are considering here: 1) permanent or semipermanent molecular interactions; 2) internal dynamism of the molecule; 3) molecular collisions. In the variety of liquid systems there are quite abundant cases in which relatively strong interactions among partners induce changes in the internal geometry. The formation of hydrogen-bonded adducts is an example of general occurrence (e.g., in water solutions). The interactions of a metal cations and their first solvation shells is a second outstanding example but the variety of cases is very large. It is not easy to give general rules on the distinction between permanent and semipermanent interactions of this type. In general it depends on the time scale of the phenomenon being studied, but it is clear that the long residence time of water molecule (in some cases, on the order of years28) in the first solvation shell around a cation leads us to consider this effect as permanent. In such cases, it is convenient to reconsider the definition of the cluster expansion and to introduce some extra variables in the nuclear coordinate subspace to span for the analysis. For example, in the case of a dimer Mn+·H2O, it is convenient to add three coordinates corresponding to the internal coordinates of the water molecule (the total number of degrees of freedom in such case is again 6, because of the spherical symmetry of the metal cation). There is no need of repeating, for this model, the variational decomposition of the energy (the PT approaches have more difficulties to treat changes in the internal geometry). The conclusions do not change qualitatively, but the quantitative results can be sensibly modified. Another important case of permanent interactions is related to chemical equilibria with molecular components of the liquid. The outstanding example is the prototropic equilibrium, especially the case AH + B → A- + HB+. There are water-water potentials, including the possibility of describing the ionic dissociation of H2O. The semipermanent interactions are generally neglected in the modeling of the potentials for liquids. Things are different when one looks at problems requiring a detailed local description of the interactions. Molecular docking problems are a typical example. In such cases, use is made of variational QM calculations, or especially for docking, where one molecule at least has a large size, use is made of molecular mechanics (MM) algorithms, allowing local modification of the systems. The second category corresponds to molecular vibrations; rotations of a molecule as a whole have been already considered in, or definition of, the ∆E(R) potential. The third category, molecular collisions, gives rise to exchange of energy among molecules that can be expressed as changes in the translational and rotational energies of the rigid molecule and changes in the internal vibrational energy distributions. In conclusion, all the cases of non-rigidity we have to consider here can be limited to molecular vibrations considered in a broad sense (couplings among vibrations and rotations are generally neglected in intermolecular potentials for liquids). The vibrations are generally treated at the classical level, in terms of local deformation coordinates. The local deformation functions are of the same type of those used in molecular mechanics (MM) methods. Nowadays, MM treats geometry changes for molecules of any dimension and chemical composition.
8.5 Three- and many-body interactions
451
In the MM approach, the energy of the system is decomposed as follows: Etot = [Estr + Eben + Etor + Eother] + Eelec + EvdW
[8.63]
the terms within square brackets regard contributions due to the bond stretching, to the angle bending, and to torsional interactions, supplemented by other contributions due to more specific deformations and by couplings among different internal coordinates. The Eelec terms regard Coulomb and inductions terms between fragments not chemically bound: among them there are the interactions between solute and solvent molecules. The same holds for the van der Waals interactions (i.e., repulsion and dispersion) collected in the last term of the equation. It must be remarked that Etot is actually a difference of energy with respect to a state of the system in which the internal geometry of each bonded component of the system is at equilibrium. The versions of MM potentials (generTable 8.4 ally called force fields) for solvent molecules STRETCH Estr =ks(1 - l0)2 may be simpler, and in fact limited libraries of BEND Eben = kb(θ - θ0)2 MM parameters are used to describe internal TORSION Etor = kt[1 ± cos(nω)] geometry change effects in liquid systems. We report in Table 8.4 the definitions of the basic, ks stretch force constant and simpler, expressions used for liquids. kb bend force constant To describe internal geometry effects in kt torsion force constant the dimeric interaction, these MM parameters l0 reference bond length are in general used in combination with a parreference bond angle θ0 titioning of the interaction potential into atomic sites. The coupling terms are neglected and the numerical values of the parameters of the site (e.g., the local charges) are left unchanged when there is a change of internal geometry produced by these local deformations. Only the relative position of the sites changes. We stress that interaction potential only regards interaction among sites of different molecules, while the local deformation affects the internal energy of the molecule.
8.5 THREE- AND MANY-BODY INTERACTIONS The tree-body component of the interaction energy of a trimer ABC is defined as: ∆E (ABC;RABC ) = E ABC (RABC ) −
1 EK ∑ ∑ E AB (RAB ) − ∑ 2 A B K
[8.64]
This function may be computed, point-by-point, over the appropriate RABC space, either with variational and PT methods, as the dimeric interactions. The results are again affected by BSS errors, and they can be corrected with the appropriate extension of the CP procedure. A complete span of the surface is, of course, by far more demanding than for a dimer, and actually extensive scans of the decomposition of the EABC potential energy surface have thus far been done for a very limited number of systems. Analogous remarks hold for the four-body component ∆E(ABCD; RABCD) of the cluster expansion energy, as well as for the five- and six- body components, the definition of
452
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
which can be easily extracted from the eq. [8.9]. The computational task becomes more and more exacting in increasing the number of bodies, as well as the dimensions which rise to 18 with fixed geometries of the components, and there are no reasonable perspectives of having in the future such a systematic scan as that available for dimers. The studies have been so far centered on four types of systems that represent four different (and typical) situations: 1) clusters composed of rare gas atoms or by small almost spherical molecules (e.g., methane); 2) clusters composed of polar and protic molecules (especially water); 3) clusters composed of a single non-polar molecule (e.g., an alkane) and a number of water molecules; 4) clusters composed of a single charged species (typically atomic ions) and a number of water molecules. Inside the four categories, attention has mostly been paid to clusters with a uniform chemical composition (e.g., trimers AAA more than A2B or ABC for cases 1 and 2). Clearly the types of many-body analyses are not sufficient to cover all cases of chemical relevance for liquids; for example, the mixed polar solvents (belonging to type 2) and the water solutions containing a cation in presence of some ligand (type 4) are poorly considered. In spite of these deficiencies the available data are sufficient to draw some general trends that in general confirm what physical intuition suggests. We combine here below conclusions drawn from the formal analysis and from the examination of numerical cases. ES contributions are strictly additive: there are no three- or many body terms for ES. The term “many-body correction” to ES introduced in some reviews actually regards two other effects. The first is the electron correlation effects which come out when the starting point is the HF description of the monomer. We have already considered this topic that does not belong, strictly speaking, to the many-body effects related to the cluster expansion [8.9]. The second regards a screening effect that we shall discuss later. The other contributions are all non-additive. The cluster expansion [8.9] applies to these terms too: TERM ( ABCDE K) =
1 1 ∑ ∑TERM AB (RAB ) + 6 ∑ ∑ ∑TERM ABC(RABC)+ K 2 A B A B C
[8.65]
The most important are the IND and DIS terms. CT is active for some special systems (long range electron transfer) but of scarce importance for normal liquids. EX many-body contributions rapidly fade away with the number of bodies: the three-body terms have been studied with some details,29 but they are not yet used extensively to model interaction potentials for liquids. The IND many-body contributions are dominated by the first dipole polarizability α contributions. The field F produced by the charge distribution of the other molecules and acting on the units for which α is defined, at a first (and good) approximation, as additive. The non-linearity derives from the fact that the contribution to the interaction energy is related to the square of the electric field. So in modeling many body effects for IND the attention is limited to molecular contributions of the type
8.5 Three- and many-body interactions
E ind (M) = −
453
1 M ind ∑ µ m F (rm ) 2 m =1
[8.66]
the index m runs over all the sites of the molecule M. The dipole moment of site m is decomposed into a static and an induction dipole: µ m = µ 0m + µ ind m
[8.67]
with the induction term αm depending on the polarizability and on the total electric field Ftot. The latter comes from the other permanent charges and dipoles, as well as from the induced dipoles present in the system: S
tot µ ind = α m ∑ Fs (rm ) m = α mF
[8.68]
s =1
Generally, the calculation of F is not limited to three-body contributions but includes higher order terms; for example, in simulation methods the contributions coming from all the molecules included in the simulation box, which may be of the order of hundreds. These contributions are not computed separately, but just used to have a collective value of F. The sum of three- and higher body contributions to IND thus may be written and computed in the following way:
(
IND(many − body ) ≈ ∑ E ind m;F tot m
)
[8.69]
When the system contains molecules of large size it is advisable to use a many-site description of the molecular polarizability α. This of course means to increase the computational effort, which is not negligible, especially in computer simulations. The many-body contribution to IND may be negative as well as positive. In general, it reinforces with a negative contribution minima on the PES. The non-additive DIS contributions are described in terms of the first dipole dynamic polarizability or by the corresponding approximate expression we have introduced for the dimeric case. DIS contributions are already present and are quite important for the liquid aggregation of spherical systems. Even spherical monomers exhibit dynamical dipoles and the related two- and many-body contributions to DIS. The most important term is known as the Axilrod-Teller term and describes the interaction of three instantaneous dipoles. The sign of the Axilrod-Teller term strongly depends on the geometry. It is negative for almost linear geometries and positive in triangular arrangements. Four body contributions of the same nature are generally of opposite sign and thus they reduce the effect of these terms on the PES shape. The potentials for polar liquids generally neglect many-body dispersion contributions. Screening many-body effects When the material condensed system contains mobile charges, the electrostatic interactions, which are strong and with the slowest decay with the distance, are severely damped. Every charged component of the system tends to attract mobile components bearing the op-
454
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
posite charge. Typical examples are the counterion charge cloud surrounding each ion in salt solutions, and the counterion condensation effects acting on immobile charged species as DNA or other polyelectrolytes in salt solutions. In both cases, at large distance the electrostatic effect of the charge is screened by the similar electrostatic effect, but with an opposite sign, of the counterion distribution. A similar screening effect is present for dipolar liquids: each molecule bearing a permanent dipole tends to organize around itself other dipolar molecules with the orientation which optimizes stabilization energy and screens the field produced by the singled out molecule. The organization of the orientation of molecular dipoles around a singled-out charged component is even more effective. The interaction at long distance between two point charges screened by the solvent dipoles is Q1Q2/εR12 where ε is the dielectric constant of the liquid. When the two charges are placed in a salt solution the value of the electrostatic energy is even lower, being screened by counterions as well as by oriented dipoles. For the dispersion contributions there is no screening. We arrive to the apparently paradoxical situation that for relatively large bodies, even when bearing net charges, the interactions at large distances are governed by the dispersion forces, weak in comparison with others, and having a more rapid decrease with the distance. Effective interaction potentials In the last sentences we have changed view with respect to the beginning of Section 8.4, where we introduced the cluster expansion of the interaction potentials. Still keeping a microscopic view, we passed from the examination of the properties of PES of increasing dimensions to a view that takes into consideration very limited portions of such surfaces selected on the basis of energetic effects, or, in other words, on the basis of probabilistic considerations. The screening effects we have mentioned are due to an averaged distribution of the other components around a single partner of the system, and its average is based on the relative energies of all the conformations of the system, spanning the whole pertinent PES. This view based on averaged distributions is related to the “focused” model we introduced in Section 8.2. If we suppose that the assisting part S of the system described by the Hamiltonian [8.8] with coordinates XS is in equilibrium with the main part M at every coordinate XM, we may repeat the cluster decomposition analyses given in Section 8.3, starting again from the dimer and proceeding to the trimers, etc. The potential function ∆EAB(R) we have called potential energy surface (PES) now assumes a different name, potential of mean force (MFP), emphasizing the fact that it regards no more interactions between A and B but also the mean effects of the other bodies present in the system. It must be remarked that the approach is not limited to equilibrium distributions of S: it may be extended to distributions of S displaced from the equilibrium. The non-equilibrium description has an important role, and the implementation of non-equilibrium methods today represents one of the frontiers of the theory of liquids. An important aspect of this revisited analysis of the dissection of ∆EAB into separate components is that all terms, including ES, are now described in terms of monomer distributions which feel the effects of S. The analyses we have done suggest that these changes can be expressed in terms of a monomer’s MO basis {φpi }“polarized” in a way similar to that we have used to describe IND. Actually, this “polarization” or deformation of the monomer is
8.5 Three- and many-body interactions
455
not due to classical polarization effects only, but to all the components of the interaction of S with A-B, namely dispersion, exchange, etc. The results of the decomposition of the intermolecular interaction energy of cluster M are not qualitatively different from those found for the same cluster M in vacuo. There are quantitative changes, often of not negligible entity and that correspond, in general, to a damping of the effects. The use of this approach requires us to consider the availability of efficient and accurate models and procedures to evaluate the effects of S on M, and at the same time, of M on S. The basic premises of this approach have been laid many years ago, essentially with the introduction of the concept of solvent reaction field made by Onsager in 1936,30 but only recently have they been satisfactory formulated. There are good reasons to expect that their use on the formulation of “effective” intermolecular potentials will increase in the next few years. Actually, the decomposition of the ∆EM(R) interaction energy is preceded by a step not present for isolated systems. There is the need of defining and analyzing the interactions of each monomer A, B, etc. with the solvent alone. The focused model is reduced to a single molecule alone, let us say A, which may be a solute molecule but also a component of the dominant mole fraction, the solvent, or a molecular component of the pure liquid. This subject is treated in detail in other chapters of this book and something more will be added in the remainder of this chapter. Here, we shall be concise. There are two main approaches, the first based on discrete (i.e., molecular) descriptions of S, the second on a continuous description of the assisting portion of the medium, via appropriate integral equations based on the density distribution of the medium and on appropriate integral kernels describing the various interactions (classical electrostatic, dispersion, exchange-repulsion). Both approaches aim at an equilibration of S with M: there is the need of repeating calculations until the desired convergence is reached. More details on both approaches will be given in the specific sections; here we shall limit ourselves to quoting some aspects related to interaction potentials, which constitute the main topic of this section. The approach based on discrete descriptions of the solvent makes explicit use of the molecular interaction potentials which may be those defined without consideration of S; the calculations are rather demanding of computation time, and this is the main reason explaining why much effort has been spent to have simple analytical expressions of such potentials. The second continuous approach is often called Effective Hamiltonian Approach (EHA) and more recently Implicit Solvation Method (ISM).31 It is based on the use of continuous response functions not requiring explicit solvent-solvent and solute-solvent interaction potentials, and it is by far less computer-demanding than the simulation approach. As we shall show in the section devoted to these methods, there are several possible versions; here it is worth quoting the Polarizable Continuum Model (PCM) for calculations at the ab initio QM level.32 PCM it is the only method used so far to get effective two body A-B potentials over the whole range of distances R (three-body corrections have been introduced in Ref. [33]). Until now, the EHA approach has been mostly used to study solvent effects on a solute molecule. In such studies, M is composed of just a single solute molecule (M = A) or of a solute molecule accompanied by a few solvent molecules (M = ASn). These alternative
456
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
choices make the results indirectly relevant to the question of the difference between traditional and effective intermolecular potential energies. However, some hints can be extracted from the available data. The experience thus far collected shows that when both M and S have a polar nature, the classical electrostatic contributions are dominant with an exchange-repulsion term, giving an important contribution to the energy but a small effect on the charge distribution of M and, as a consequence, on the molecular properties on which the interaction potential depends (e.g., multipole distribution of the charge density, static and dynamic polarizabilities). Dispersion effects play a role when M is not polar and they become the dominant component of the solvation effects when both M and S are not polar. The calculation of effective PCM interaction potentials has been so far limited to the cases in which the solvent effects are more sizeable, namely the interactions of metal cations with water. They have been used in computer simulations of models of very dilute ionic solutions (a single ion in water), noticeably improving the results with respect to similar studies using traditional potentials.34
8.6 THE VARIETY OF INTERACTION POTENTIALS In the preceding sections we have examined the definition of the components of the intermolecular potential, paying attention mostly to rigorous methods and to some approximations based on these methods. The beginning of qualitative and semi-quantitative computational studies on the properties of liquids dates back at least to the 1950s, and in the early years of these studies, the computational resources were not so powerful as they are today. For this reason, for many years, the potentials used have been simple, discarding what our understanding of the dimeric interactions suggested. There is no reason, however, to neglect here these simplified versions of the potential, because very simple and naïve expressions have given excellent results and are still in use. The struggle for better and more detailed potentials is justified, of course, because chemistry always tends to have more and more accurate estimates of the properties of interest, and there are many problems in which a semi-quantitative assessment is not sufficient. We shall try to give a cursory view of the variety of intermolecular potentials in use for liquid systems, paying more attention to the simplest ones. We shall almost completely neglect potentials used in other fields, such as the scattering of two isolated molecules, the determination of spectroscopic properties of dimers and trimers, and the accurate study of local chemical interactions, which all require more sophisticated potentials. The main criterion to judge the quality of a potential is a posteriori, namely, based on the examination of the performance of that potential in describing properties of the liquid system. In fact, the performances of a potential are to a good extent based on subtle factors expressing the equilibrium reached in the definition of its various components. The experience gained in using this criterion has unfortunately shown that there is no definite answer. Typically, a potential, or a family of potentials, better describe some properties of the liquid, while a second one is more proper for other properties. For this reason there are many potentials in use for the same liquid and there are in the literature continuous comparisons among different potentials. A second criterion is based on the computational cost, an aspect of direct interest for people planning to make numerical studies but of little interest for people only interested in the results.
8.6 The variety of interaction potentials
457
Our exposition will not try to classify potentials according to such criteria, but simply show the variety of potentials in use. We shall pay attention to the description adopted for a single molecule, that must then be combined with that adopted for the interaction partners. In particular, we shall consider the number of sites and the shape of the molecule used in each description. The number of sites reflects the possibility we have examined, and advocated, of using many-center expansions to improve the representation. Each expansion center will be a site. There are models with one, two, and more sites. This sequence of increasing complexity reaches the number of heavy atoms of the molecule and then the whole number of atoms, including hydrogens. It is not limited to the nuclei as expansion sites. There are potentials introducing other locations of sites, in substitution or in addition to the nuclei. For example potentials widely used in simulations adopt for water a four-site model; other potentials (rarely used in simulations) prefer to use the middle of the bonds instead of (or in addition to) nuclei. Each site of the molecule must be combined with the sites of the second (and other) molecule to give the potential. The shape of the molecule reflects the effect of the exchange-repulsion interaction. For almost all many-site models the shape is not given, but it implicitly results to be that of the union of the spheres centered on the expansion sites provided by a source of exchange-repulsion potential. There are some simple models in which the shape is explicitly stated. There will be spheres, ellipsoids, cylinders and more complex shapes, as fused spheres, spherocylinders, etc. Some typical examples are reported in Table 8.5. Table 8.5. Single site-based potentials #
Shape
Interaction
Name
1
Sphere
hard
HS
2
Sphere
soft
SS
3
Sphere
hard with charge
CHS
4
Sphere
soft with charge
CSS
5
Sphere
hard with rigid dipole
DHS
6
Sphere
soft with dipole
DSS
7
Disc
hard
HD
8
Disc
hard with dipole
SD
hard
9
Ellipsoid
10
Ellipsoid
Gay-Berne35
11
Quadrupolar shape
Zewdie36
12
Sphere
repulsion dispersion
Lennard Jones (LJ)
13
Sphere
repulsion dispersion
Buckingham
14
Sphere
LJ + charge
LJ+q
15
Sphere
LJ + dipole
Stockmayer
16
Sphere
LJ + dipole+ soft sticky
SSD37
458
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Potential 1 is extremely simple: the only parameter is the radius of the sphere. In spite of this simplicity, an impressive number of physical results have been obtained using the HS potential on the whole range of densities and aggregations. Also, mixtures of liquids have been successfully treated, introducing in the computational machinery the desired number of spheres with appropriate radii. The HS model is at the basis of the Scaled Particle Theory (SPT),38 which still constitutes a basic element of modern solvation methods (see later for more details). Of course, HS cannot give many details. A step toward realism is given by potential 2, in which the hard potential is replaced by a steep but smoothly repulsive potential. It is worth reminding readers that the shift from HS to SS in computer simulations required the availability of a new generation of computers. Simulations had in the past, and still have at present, to face the problem that an increase in the complexity of the potential leads to a large increment of the computational demand. Potentials 3 and 4 have been introduced to study ionic solutions and similar fluids containing mobile electric charges (as, for example, molten salts). In ionic solutions the appropriate mixture of charged and uncharged spheres is used. These potentials are the first examples of potentials in our list, in which the two-body characteristics become explicit. The interaction between two charged spheres is in fact described in terms of the charge values of a couple of spheres: QkQl/rlk (or QkQl/εrlk). Potentials 5 and 6 are the first examples in our lists of anisotropic potentials. The interaction here depends on the mutual orientations of the two dipoles. Other versions of the dipole-into-a-sphere potentials (not reported in Table 8.5) include induced dipoles and actually belong to a higher level of complexity in the models, because for their use there is the need of an iterative loop to fix the local value of F. There are also other similar models simulating liquids in which the location of the dipoles is held fixed at nodes of a regular 3D grid, the hard sphere potential is discarded, and the optimization only regard orientation and strength of the local dipoles. This last type of model is used in combination with solutes M described in another, more detailed way. Potentials 7-11 add more realism in the description of molecular shape. Among them, the Gay-Berne35 potential (10) has gained a large popularity in the description of liquid crystals. The quite recent potential (11)36 aims at replacing Gay-Berne potential. It has been added here to show that even in the field of simple potentials there is space for innovation. Potentials 12 and 13 are very important for chemical studies on liquids and solutions. The Lennard-Jones potential (12) includes dispersion and repulsion interactions in the form: LJ:
σ 12 σ 6 LJ E AB = ε − R R
[8.70]
The dependence of the potential on the couple of molecules is here explicit, because the parameters ε and σ depend on the couple. In the original version, ε is defined as the depth of the attractive well and σ as the distance at which the steep repulsive wall begins. More recently, the LJ expression has been adopted as an analytical template to fit numerical values of the interaction, using independently the two parameters.:
8.6 The variety of interaction potentials
459
12
LJ with 2 parameters:
E
LJ AB
1 (R) = AAB R
1 − CAB R
6
[8.71]
LJ expressions are of extensive use in MM to model non-covalent interactions among molecular sites, often with the addition of local charges (potential 14) LJ with charge:
1 E ms (rms ) = Ams rms
12
1 − Cms rms
6
q q + m s rms
[8.72]
Some potentials add an effective dielectric constant in the denominator of the Coulomb term to mimic polarization effects. The Buckhingam potential is similar to LJ with a more physical description of the repulsion term given in terms of a decaying exponential function: E
( BU ) AB
−6 6 r α r = ε exp −α − 1 − α − 6 r * α − 6 r *
[8.73]
where: ε r r* α
depth of the attractive well RAB distance of the minimum numerical parameter
The difference in the computational costs of LJ and Buckingham potentials is related to the difference between computing the square of a value already available (R-12 from R-6) and that of computing an exponential. Here again the increment in computational costs of simulations due to small changes in the potential plays a significant role. Also, the Buckingham + charge potential is used to describe liquids. The last potential of Table 8.5 is specialized for water. It has been enclosed in the table to document the progress in one-site potentials; in this case, the LJ + dipole is supplemented by a short-range “sticky” tetrahedral interaction. The listing of one-center potentials is not exhausted by the examples given in Table 8.5. We have, for example, neglected all the potentials in use for rare gases systems in which much attention is paid to using higher terms to describe the dispersion contribution. Our aim was just to show with a few examples how it is possible to define a large variety of potentials remaining within the constraints of using a single center. The problem of giving a cursory but significant enough view of potentials becomes harder when one passes to many-site potentials. In Table 8.6 we report some analogues of HS with a more complex shape. They consist of fused regular forms (spheres or combination of spheres with other regular solids). The hard version of these potentials is accompanied by soft modifications and by versions including dipole, charges, as in Table 8.5. A step further along this way is given by a potential composed by spheres linked by ‘spacers’ with a constant length but allowing changes of conformation with appropriate torsion potentials. These potentials are used for polymers or for molecules having long hydrocarbon chains. This is not, however, the main trend in the evolution of potentials.
460
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Table 8.6. Many-site simple potentials #
Shape
Interaction
Name
1
three fused spheres
hard
hard dumbbell HD
2
two half spheres + cylinder
hard
hard spherocylinder HSC
3
general convex shape
hard
hard convex core HCC
The largest number of solvents are composed by polyatomic molecules of small-medium size, exhibiting a variable (but in general not excessive) degree of flexibility. In solutions, the interactions involve these solvents and molecules having the same characteristics as solvent or with more complex chemical composition. In all cases (pure liquids, solutions with solute of variable complexity), one has to take into account interactions having a remarkable degree of specificity. The chemical approach to the problem addresses this specificity on which the whole chemistry is based. This is the real field of application of the definition and analysis of molecular interactions on which we have spent the first sections of this chapter. It is clear that we are entering here into a very complex realm, hard to summarize. A whole book would be necessary. It is possible to make a rough distinction between potential of general applicability and potentials conceived for specific couples (or collections) of molecules. In both cases, the many-site expansions are used, but for general potentials, in which transferability is asked, the description is obviously less detailed. The LJ+charge (potential 14 in Table 8.5) is a popular choice for these potentials; the sites are in general limited to the heavy atoms of the molecule. There are, however, many other versions with a variety of changes to this standard setting. The potentials elaborated for specific cases are often more detailed and they adapt a larger number of devices to increase the accuracy. Many have been developed for pure liquids, but the number of potentials regarding specific combinations of solute and solvent is not negligible. The number of solvents in chemistry is quite large. The number of available specific potentials is by far more limited, but too large to be summarized here. In addition, a simple list of names and references, not accompanied by critical remarks about the performances of such potentials in describing liquid systems would have little utility. The numerous textbooks on intermolecular potentials are of little help, because outdated or paying little attention to the analytical exposition of computational models. More useful are monographs and reviews regarding computer simulations of liquids39 or original papers making comparisons among different potentials. These last are more abundant for water. This solvent, because of its importance and its molecular simplicity, has been the benchmark for many ideas about the description and modeling of intermolecular interactions. Almost all potentials that have been thus far proposed for not extremely flexible molecular component of liquid systems may be found in the literature regarding water. Alternative strategies in the defining number and position of sites, use diffuse instead of point charges, different approaches to model the many-body corrections, flexible potentials, dissociation of the molecule are some features that can be examined by looking at the water potential literature.
8.7 Theoretical and computing modeling
461
There is a large critical literature, but two recent reviews (Floris and Tani40 and Wallqvist and Mountain,41 both published in 1999) are here recommended as an excellent guide to this subject. The two reviews consider and analyze about 100 potentials for pure water. WM review starts from historical models, FM review pays attention to recent models (about 70 models, supplemented by ion-water potentials). The two reviews partly overlap, but they are to a good extent complementary, especially in the analysis of the performances of such models. There are, to the best of our knowledge, no reviews of comparable accuracy for the potentials regarding other liquid systems.
8.7 THEORETICAL AND COMPUTING MODELING OF PURE LIQUIDS AND SOLUTIONS 8.7.1 PHYSICAL MODELS In this chapter we shall present a necessarily partial review of the main theoretical approaches so far developed to treat liquid systems in terms of physical functions. We shall restrict ourselves to two basic theories, integral equation and perturbation theories, to keep the chapter within reasonable bounds. In addition, only the basic theoretical principles underlying the original methods will be discussed, because the progress has been less rapid for theory than for numerical applications. The latter are in fact developing so fast that it is an impossible task to try to give an exhaustive view in a few pages. A fundamental approach to liquids is provided by the integral equation methods42-44 (sometimes called distribution function methods), initiated by Kirkwood and Yvon in the 1930s. As we shall show below, one starts by writing down an exact equation for the molecular distribution function of interest, usually the pair function, and then introduces one or more approximations to solve the problem. These approximations are often motivated by considerations of mathematical simplicity, so that their validity depends on a posteriori agreement with computer simulation or experiment. The theories in question, called YBG (Yvon-Born-Green), PY (Percus-Yevick), and the HNC (hypernetted chain) approximation, provide the distribution functions directly, and are thus applicable to a wide variety of properties. An alternative, and particularly successful, approach to liquids is provided by the thermodynamic perturbation theories.42,44 In this approach, the properties of the fluid of interest are related to those of a reference fluid through a suitable expansion. One attempts to choose a reference system that is in some sense close to the real system, and whose properties are well known (e.g., through computer simulation studies or an integral equation theory). As a last physical approach we mention, but do not further consider, the scaled-particle-theory (SPT)38,45 which was developed about the same time as the Percus-Yevick theory. It gives good results for the thermodynamic properties of hard molecules (spheres or convex molecules). It is not a complete theory (in contrast to the integral equation and perturbation theories) since it does not yield the molecular distribution functions (although they can be obtained for some finite range of intermolecular separations). Early work on the theory of dense fluids dealt almost exclusively with simple atomic fluids, in which the intermolecular forces are between the centers of spherical molecules and depend only on the separation distance r. However, in real fluids the intermolecular forces depend on the molecular orientations, vibrational coordinates, etc., in addition to r.
462
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
The complexity of the problem thus requires the assumption of some approximations; most of the formulated theories, for example, assume (a) a rigid molecule approximation, (b) a classical treatment of the traslational and rotational motions, and (c) a pairwise additivity of the intermolecular forces. In the rigid molecule approximation it is assumed that the intermolecular potential energy V(rN,ωN) depends only on the positions of the centers of mass rN=r1r2...rN for the N molecules and on their orientations ωN = ω1ω2K ωN . This implies that the vibrational coordinates of the molecules are dynamically and statistically independent on the center of mass and orientation coordinates, and that the internal rotations are either absent, or independent of the rN and ωN coordinates. The molecules are also assumed to be in their ground electronic states. At the bases of the second basic assumption made, e.g., that the fluids behave classically, there is the knowledge that the quantum effects in the thermodynamic properties are usually small, and can be calculated readily to the first approximation. For the structural properties (e.g., pair correlation function, structure factors), no detailed estimates are available for molecular liquids, while for atomic liquids the relevant theoretical expressions for the quantum corrections are available in the literature. The third basic approximation usually introduced is that the total intermolecular potential energy V(rN,ωN) is simply the sum of the intermolecular potentials for isolated pairs ij of molecules, i.e.,
(
V r N , ωN
) = ∑ V(r , ω , ω ) i< j
ij
i
j
[8.74]
In the sum i is kept less than j in order to avoid counting any pair interaction twice. Eq. [8.74] is exact in the low-density gas limit, since interactions involving three or more molecules can be ignored. It is not exact for dense fluids or solids, however, because the presence of additional molecules nearby distorts the electron charge distributions in molecules i and j, and thus changes the intermolecular interaction between this pair from the isolated pair value. In order to get a more reliable description, three-body (and higher multi-body) correction terms should be introduced. The influence of three-body terms on the physical properties has been studied in detail for atomic fluids,46,47 while much less is known about molecular fluids. In the latter case, the accurate potentials are few and statistical mechanical calculations are usually done with model potentials. A particular model may be purely empirical (e.g., atom-atom), or semiempirical (e.g., generalized Stockmayer), where some of the terms have a theoretical basis. The so-called generalized Stockmayer model consists of central and non-central terms. For the central part, one assumes a two-parameter central form (the classical example is the Lennard-Jones, LJ, form introduced in the previous sections). The long-range, non-central part in general contains a truncated sum of multipolar, induction and dispersion terms. In addition, a short-range, angle-dependent overlap part, representing the shape or core of the potential, is usually introduced. The multipolar interactions are pair-wise additive, but the induction, dispersion and overlap interactions contain three-body (and higher multi-body) terms. Hence, three-body interactions are strongly suspected to be of large importance also for molecular liquids.
8.7 Theoretical and computing modeling
463
This review will be mainly restricted to a discussion of small rigid molecules in their ground electronic states. However, one can generalize in a number of directions by extending the pair potential v(rij,ωi,ωj) of eq. [8.74] so as to treat: (a) non-rigid molecules, (b) molecules with internal rotation, (c) large (e.g., long chain) molecules which may be flexible, (d) electronic excited state molecules. In addition, the inclusion of covalence and charge transfer effects are also possible. The first two generalizations are straightforward in principle; one lets v depend on coordinates describing the additional degrees of freedom involved. As for (c), for very large molecules the site-site plus charge-charge model is the only viable one. Under (d) we can quote the so-called long-range ‘resonance’ interactions. As the last note, we anticipate that in the following only equilibrium properties will be considered; however, it is fundamental to recall from the beginning that dynamical (e.g., non-equilibrium) analyses are an important and active field of research in the theory of physical models.42,48,49 Before entering into more details of each theory, it is worth introducing some basic definitions of statistical mechanics. All equilibrium properties of a system can be calculated if both the intermolecular potential energy and the distribution functions are known. In considering fluids in equilibrium, we can distinguish three principal cases: (a) isotropic, homogeneous fluids, (e.g., liquid or compressed gas in the absence of an external field), (b) anisotropic, homogeneous fluids (e.g., a polyatomic fluid in the presence of a uniform electric field, nematic liquid crystals), and (c) inhomogeneous fluids (e.g., the interfacial region). These fluid states have been listed in order of increasing complexity; thus, more independent variables are involved in cases (b) and (c), and consequently the evaluation of the necessary distribution functions is more difficult. For molecular fluids, it is convenient to define different types of distribution functions, correlation functions and related quantities. In particular, in the pair-wise additive theory of homogeneous fluids (see eq. [8.74]), a central role is played by the angular pair correlation function g(r12ω1ω2) proportional to the probability density of finding two molecules with position r1 and r2 and orientations ω1 and ω2 (a schematic representation of such function is reported in Figure 8.5). In fact, one is frequently interested in some observable property , which is (experimentally) a time average of a function of the phase variables B(rN,ω N); the latter is often a sum of pair terms b(rij,ωi,ωj) so that is given by
(
)(
)
B = ∫ dr NdωNP r N ωN B r N ωN =
1 ρN ∫ dr g( rω1ω2 )b ( rω1ω2 ) 2
ω1 ω2
[8.75]
i.e., in terms of the pair correlation function g(12). Examples of such properties are the configurational contribution to energy, pressure, mean squared torque, and the mean squared force. In eq. [8.75] g ( rω1ω2 )b( rω1 ω2 ) ω ω means the unweighted average over orienta1 2 tions. In addition to g(12), it is also useful to define:
464
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Figure 8.5. Example of the radial distribution function g(r) of a typical monoatomic liquid.
Figure 8.6. Examples of h(r) and c(r) functions for a typical monoatomic liquid.
a) the site-site correlation function gαβ(rαβ) proportional to the probability density that sites α and β on different molecules are separated by distance r, regardless of molecular orientations; b) the total correlation function h(r12ω1ω2)= g(r12ω1ω2) - 1; c) the direct correlation function c(r12ω1ω2). Simple examples of h and c functions are reported in Figure 8.6. In the list above, item (a) implies definite sites within molecules; these sites may be the nuclei themselves or sites at arbitrary locations within the molecules. In addition, the total correlation, h, between molecules 1 and 2 can be separated into two parts: (i) a direct effect of 1 on 2 which is short-ranged and is characterized by c, and (ii) and indirect effect in which 1 influences other molecules, 3, 4, etc., which in turn affect 2. The indirect effect is the sum of all contributions from other molecules averaged over their configurations. For an isotropic and homogeneous fluid formed by non-spherical molecules, we have h(r12 ω1 ω2 ) = c(r12 ω1 ω2 ) + ρ ∫ dr 3 c (r13 ω1 ω3 )h(r 32 ω3 ω2 )
ω3
[8.76]
which is the generalization of the Ornstein-Zernike (OZ) equation50 to non-spherical molecules. The OZ equation is the starting point for many theories of the pair correlation function (PY, HNC, etc.); however, numerical solutions starting from [8.76] are complicated by the large number of variables involved. By expanding the direct and total correlation functions in spherical harmonics, one obtains a set of algebraically coupled equations relating the harmonic coefficients of h and c. These equations involve only one variable in the place of many in the original OZ equation. In addition, the theories we shall describe below truncate the infinite set of coupled equations into a finite set, thereby enabling a reasonably simple solution to be carried out.
8.7 Theoretical and computing modeling
465
8.7.1.1 Integral equation methods The structure of the integral equation approach for calculating the angular pair correlation function g(r12ω1ω2) starts with the OZ integral equation [8.76] between the total (h) and the direct (c) correlation function, which is here schematically rewritten as h=h[c] where h[c] denotes a functional of c. Coupled to that a second relation, the so-called closure relation c=c[h], is introduced. While the former is exact, the latter relation is approximated; the form of this approximation is the main distinction among the various integral equation theories to be described below. In the OZ equation, h depends on c, and in the closure relation, c depends on h; thus the unknown h depends on itself and must be determined self-consistently. This (self-consistency requiring, or integral equation) structure is characteristic of all many body problems. Two of the classic integral equation approximations for atomic liquids are the PY (Percus-Yevick)51 and the HNC (hypernetted chain)52 approximations that use the following closures h −c = y −1
(PY)
[8.77]
h − c = ln y
(HNC)
[8.78]
where y is the direct correlation function defined by g(12)=exp(-β v(12))y(12) with v(12) the pair potential and β=1/kT. The closures [8.77-8.78] can be also written in the form
(
c = g 1 − e βv (12 )
)
c = h − βv(12) − ln g
(PY)
[8.79]
(HNC)
[8.80]
For atomic liquids the PY theory is better for steep repulsive pair potentials, e.g., hard spheres, whereas the HNC theory is better when attractive forces are present, e.g., Lennard-Jones, and Coulomb potentials. No stated tests are available for molecular liquids; more details are given below. As said before, in practice the integral equations for molecular liquids are almost always solved using spherical harmonic expansions. This is because the basic form [8.76] of the OZ relation contains too many variables to be handled efficiently. In addition, harmonic expansions are necessarily truncated after a finite number of terms. The validity of the truncations rests on the rate of convergence of the harmonic series that depends in turn on the degree of anisotropy in the intermolecular potential. We recall that the solution to the PY approximation for the hard sphere atomic fluid is analytical and it also forms a basis for other theories, e.g., in molecular fluids the MSA and RISM theories to be discussed below. The mean spherical approximation (MSA)52 theory for fluids originated as the extension to continuum fluids of the spherical model for lattice gases. In practice it is usually applied to potentials with spherical hard cores, although extensions to soft core and non-spherical core potentials have been discussed. The MSA is based on the OZ relation together with the closure
466
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
h = −1
r <σ
[8.81]
c = −βv(12)
r <σ
[8.82]
where σ is the diameter of the spherical hard core part of the pair potential v(12). The condition [8.81] is exact for hard core potentials since g(12)=0 for r < σ, while the condition [8.82] is the approximation, being correct only asymptotically (large r). We thus expect the approximation to be worst near the hard core; we also note that the theory is not exact at low density (except for a pure hard sphere potential, where MSA=PY), but this is not too important since good theories exist for low densities, and we are therefore mainly interested in high densities. Once again the solution of the MSA problem is most easily accomplished using the spherical harmonic component of h and c. In the list of the various correlation functions we have introduced at item (b) the so-called site-site pair correlation functions gαβ(rαβ); since they depend only on the radial variables rαβ (between sites), they are naturally simpler than g(r12;ω1ω2 ) but, at the same time, they contain less information. The theories for gαβ fall into two categories based on site-site or particle-particle OZ equations, respectively. Various closures can be used with either category. As concerns the site-site approach the most important theory is the ‘reference interaction site model’, or RISM.53,54 This method applies to an intermolecular pair potential modeled by a site-site form, i.e., V(rω1ω2 ) = Σ αβ v αβ ( rαβ ) and its original intuitive derivation is based on exploring the possibility of decomposing g(r;ω1ω2 ) also in the same form, i.e., as a sum of site-site gαβ(r)s. The RISM theory consists of (i) an OZ-like relation between the set {hαβ} (where hαβ(r)=gαβ(r)-1) and corresponding set {cαβ} of direct correlation functions, and (ii) a PY-type or other type of closure. The OZ-like relation is here a matrix one. Although successful in many applications, the RISM approximation suffers from a number of major defects. First, it is not the best choice to calculate the equation of state, and the results that are obtained are thermodynamically inconsistent. Secondly, calculated structural properties show an unphysical dependence on the presence of ‘auxiliary’ sites, i.e., on sites that label points in a molecule but contribute nothing to the intermolecular potential. Thirdly, use of the RISM approximation leads to trivial and incorrect results for certain quantities descriptive of angular correlations in the fluid.48 Attempts to develop a more satisfactory theory within the interaction-site picture have developed along two different lines. The first relies on treating the RISM-OZ relation as providing the definition of the site-site direct correlation functions, cαβ. In this way the usual RISM closures can be modified by adding terms to cαβ such that the asymptotic result is satisfied. In the alternative approach, the basic observation is that the RISM-OZ relation, though plausible, does not provide an adequate basis for a complete theory of the structure of molecular fluids. Accordingly, it is there rather than in the closure relation that improvement must be sought.55 Despite the defects we have underlined, in the last years the RISM method has received much more attention than other integral equation methods; promising developments in this area are represented by the inclusion of polarizable media in RISM56 and the coupling of the RISM equations to a quantum mechanical description.57
8.7 Theoretical and computing modeling
467
8.7.1.2 Perturbation theories In perturbation theories44,58 one relates the properties (e.g., the distribution functions or free energy) of the real system, for which the intermolecular potential energy is E, to those of a reference system where the potential is E0, usually by an expansion in powers of the perturbation potential E1=E - E0. The methods used are conveniently classified according to whether the reference system potential is spherically symmetric or anisotropic. Theories of the first type are most appropriate when the anisotropy in the full potential is weak and long ranged; those of the second type have a greater physical appeal and a wider range of possible application, but they are more difficult to implement because the calculation of the reference-system properties poses greater problems. In considering perturbation theory for liquids it is convenient first to discuss the historical development for atomic liquids. In many cases the intermolecular pair potential can be separated in a natural way into a sharp, short-range repulsion and a smoothly varying, long range attraction. A separation of this type is an explicit ingredient of many empirical representations of the intermolecular forces including, for example, the Lennard-Jones potential. It is now generally accepted that the structure of simple liquids, at least of high density, is largely determined by geometric factors associated with the packing of the molecular hard cores. By contrast, the attractive interactions may, in the first approximation, be regarded as giving rise to a uniform background potential that provides the cohesive energy of the liquid but has little effect on its structure. A further plausible approximation consists of modeling the short-range forces by the infinitely steep repulsion of the hard-sphere potential. In this way, the properties of a given liquid can be related to those of a hard-sphere reference system, the attractive part of the potential being treated as a perturbation. The choice of the hard-sphere fluid as a reference system is an obvious one, since its thermodynamic and structural properties are well known. The idea of representing a liquid as a system of hard spheres moving in a uniform, attractive potential well is an old one; suffice here to recall the van der Waals equation. Roughly one can thus regard perturbation methods as attempts to improve the theory of van der Waals in a systematic fashion. The basis of all the perturbation theories we shall consider is a division of the pair potential of the form v(12) = v 0 (12) + w(12)
[8.83]
where v0 is the pair potential of the reference system and w(12) is the perturbation. The following step is to compute the effect of the perturbation on the thermodynamic properties and pair distribution function of the reference system. This can be done systematically via an expansion in powers either of inverse temperature (the “λ expansion”) or of a parameter that measures the range of the perturbation (the “γ expansion”). In spite of the fact that hard spheres are a natural choice of the reference system, for the reasons discussed above, realistic intermolecular potentials do not have an infinitely steep hard core, and there is no natural separation into a hard-sphere part and a weak perturbation. A possible improvement is that to take proper account of the “softness”’ of the repulsive part of the intermolecular potential. A method of doing this was proposed by Rowlinson,59
468
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
who used an expansion in powers of a softness parameter, n-1, about n-1=0 (corresponding to a hard sphere fluid). This method gives good results for the repulsive, but not for the attractive, part of the potential. Subsequently several researchers attempted to combine the advantages of both approaches. The first successful method was that of Barker and Henderson (BH),60 who showed that a second-order theory gave quantitative results for the thermodynamic properties of a Lennard-Jones liquid. Even more rapidly convergent results were later obtained by Weeks, Chandler and Andersen (WCA)61 using a somewhat different reference system. Given the great success of perturbation theories in treating the properties of atomic liquids, a large effort has been devoted to extending the methods to deal with molecular systems. The first rigorous application of these theories to molecular fluids seems to have been made in 1951 by Barker,62 who expanded the partition function for a polar fluid about that for a fluid of isotropic molecules. Roughly, the basic problem is the same as in the atomic case, but the practical difficulties are much more severe. A possible approach is to choose a reference-system potential v0(r) spherically symmetric so that the integrations over the orientations (absent in atomic fluids) involve only the perturbation. The real system can be thus studied by a straightforward generalization of the λ-expansion developed for atomic fluids. Perturbation theories based on spherically symmetric reference potentials, however, cannot be expected to work well when (as in the most real molecules) the short-range repulsive forces are strongly anisotropic. The natural approach in such cases is to include the strongly varying interactions in the specification of an isotropic reference potential to relate the properties of the reference system to those of hard molecules having the same shape.63 Calculation of this type have been made by Tildesley,64 exploiting a generalization of the WBA approach quoted above. In a similar way also, the BH perturbation theory has been generalized with good results.65 The main disadvantage of these methods is the large computational effort that their implementation requires. 8.7.2 COMPUTER SIMULATIONS In this section we will give a brief review of methodologies to perform computer simulations. These approaches nowadays have a major role in the study of liquid-state physics: they are developing so fast that it is impossible to give a complete view of all the different methodologies used and of the overwhelmingly large number of applications. Therefore we will only give a brief account of the basic principles of such methodologies, of how a computer simulation can be carried out, and we will briefly discuss limitations and advantages of the methods. The literature in the field is enormous: here we will refer interested readers to some of the basic textbooks on this subject.42,66,67 The microscopic state of a system may be specified in terms of the positions and momenta of the set of particles (atoms or molecules). Making the approximation that a classical description is adequate, the Hamiltonian H of a system of N particles can be written as a sum of a kinetic K and a potential V energy functions of the set of coordinates qi and momenta pi of each particle i, i.e.: H(q, p) = K(p) + V(q )
[8.84]
In the equation the coordinates q can simply be the set of Cartesian coordinates of each atom in the system, but in this case we treat the molecules as rigid bodies (as is usually done
8.7 Theoretical and computing modeling
469
in most simulations); q will consist of the Cartesian coordinates of each molecular center of mass together with a set of other parameters specifying the molecular orientation. In any case p is the set of conjugate momenta. Usually the kinetic energy K takes the form: N
K =∑∑ i =1
µ
p i2µ
[8.85]
2m i
where the index µ runs over the components (x, y, z) of the momentum of molecule i and mi is the molecular mass. The potential energy V contains the interesting information regarding intermolecular interactions. A detailed analysis on this subject has been given in the previous sections; here we shall recall only some basic aspects. It is possible to construct from H an equation of motion that governs the time-evolution of the system, as well as to state the equilibrium distribution function for positions and momenta. Thus H (or better V) is the basic input to a computer simulation program. Almost universally, in computer simulation the potential energy is broken up into terms involving pairs, triplets, etc. of molecules, i.e.:
(
)
V = ∑ v 1 (ri ) + ∑ ∑ v 2 ri , r j + ∑ ∑ i
i
j >i
i
∑ v (r , r , r )+K
j >i k > j >i
3
i
j
k
[8.86]
The notation indicates that the summation runs over all distinct pairs i and j or triplets i, j and k, without counting any pair or triplet twice. In eq. [8.86], the first term represents the effect of an external field on the system, while the remaining terms represent interactions between the particles of the system. Among them, the second one, the pair potential, is the most important. As said in the previous section, in the case of molecular systems the interaction between nuclei and electronic charge clouds of a pair of molecules i and j is generally a complicated function of the relative positions ri and rj, and orientations Ωi and Ωj (for atomic systems this term depends only on the pair separation, so that it may written as v2(rij)). In the simplest approximation the total interaction is viewed as a sum of pair-wise contributions from distinct sites a in molecule i, at position ria, and b in molecule j, at position rjb, i.e.:
(
)
v rij , Ω i , Ω j = ∑ ∑ v ab (rab )
[8.87]
In the equation, vab is the potential acting between sites a and b, whose inter-separation is rab. The pair potential shows the typical features of intermolecular interactions as shown in Figure 8.1: there is an attractive tail at large separation, due to correlation between the electron clouds surrounding the atoms (‘van der Waals’ or ‘London’ dispersion), a negative well, responsible for cohesion in condensed phases, and a steeply rising repulsive wall at short distances, due to overlap between electron clouds. Turning the attention to terms in eq. [8.86] involving triplets, they are usually significant at liquid densities, while four-body and higher are expected to be small in comparison with v2 and v3. Despite of their significance, only rarely triplet terms are included in computer simulations: that is due to the fact that the calculation of quantities related to a sum over triplets of molecules are very time-consuming. On the other hand, the average
470
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
three-body effects can be partially included in a pair-wise approximation, leading it to give a good description of liquid properties. This can be achieved by defining an “effective” pair potential, able to represent all the many-body effects. To do this, eq. [8.86] can be re-written as:
( )
V ≈ ∑ v 1 (ri ) + ∑ ∑ v 2eff rij i
i
j >i
[8.88]
The most widely used effective potential in computer simulations is the simple Lennard-Jones 12-6 potential (see eq. [8.70]), but other possibilities are also available; see Section 8.6 for discussion on these choices. 8.7.2.1 Car-Parrinello direct QM simulation Until now we have focused the attention to the most usual way of determining a potential interaction to be used in simulations. It is worth mentioning a different approach to the problem, in which the distribution of electrons is not treated by means of an “effective” interaction potential, but is treated ab initio by density functional theory (DFT). The most popular method is the Car-Parrinello (CP) approach,68,69 in which the electronic degrees of freedom are explicitly included in the description and the electrons (to which a fictitious mass is assigned) are allowed to relax during the course of the simulation by a process called “simulated-annealing”. In that way, any division of V into pair-wise and higher terms is avoided and the interatomic forces are generated in a consistent and accurate way as the simulation proceeds. This point constitutes the main difference between a CP simulation and a conventional MD simulation, which is preceded by the determination of the potential and in which the process leading to the potential is completely separated from the actual simulation. The forces are computed using electronic structure calculations based on DFT, so that the interatomic potential is parameter-free and derived from first principles, with no experimental input. Let us consider a system for which the BO approximation holds and for which the motion of the nuclei can be described by classical mechanics. The interaction potential is given by: V(R) = ψ 0 | H$ | ψ 0
[8.89]
where H is the Hamiltonian of the system at fixed R positions and ψ0 is the corresponding instantaneous ground state. Eq. [8.89] permits to define the interaction potential from first principles. In order to use eq. [8.89] in a MD simulation, calculations of ψ0 for a number of configurations of the order of 104 are needed. Obviously this is computationally very demanding, so that the use of certain very accurate QM methods (for example, the configuration interaction (CI)) is precluded. A practical alternative is the use of DFT. Following Kohn and Sham,70 the electron density ρ(r) can be written in terms of occupied single-particle orthonormal orbitals: ρ(r ) = ∑ ψ i (r ) i
2
[8.90]
8.7 Theoretical and computing modeling
471
A point on the BO potential energy surface (PES) is then given by the minimum with respect to the ψi of the energy functional: V [{ψ i }{ , RI }{ , α v }] =
∑∫ i
Ω
h2 2 d 3rψ *i (r ) − ∇ ψ i (r ) + U[ρ(r ),{RI }{ , α v }] 2m
[8.91]
where {RI} are the nuclear coordinates, {αv} are all the possible external constraints imposed on the system (e.g., the volume Ω). The functional U contains the inter-nuclear Coulomb repulsion and the effective electronic potential energy, including external nuclear, Hartree, and exchange and correlation contributions. In the conventional approach, the minimization of the energy functional (eq. [8.91]) with respect to the orbitals ψi subject to the orthonormalization constraint leads to a set of self-consistent equations (the Kohn-Sham equations), i.e.: h2 2 ∂U ∇ + − ψ i (r ) = ε i ψ i (r ) 2 ∂ρ m r ( )
[8.92]
whose solution involves repeated matrix diagonalizations (and rapidly growing computational effort as the size of the system increases). It is possible to use an alternative approach, regarding the minimization of the functional as an optimization problem, which can be solved by means of the simulated annealing procedure.71 A simulated annealing technique based on MD can be efficiently applied to minimize the KS functional: the resulting technique, called “dynamical simulated annealing” allows the study of finite temperature properties. In the “dynamical simulated annealing,” the {Ri}, {αv} and {ψi} parameters can be considered as dependent on time; then the Lagrangian is introduced, i.e.: 1 &i L = ∑ µ ∫ d 3r ψ i 2 Ω
2
1 1 &2 + ∑ M I R& I + ∑ µ v α , RI }{ , α v }] [8.93] v − V[{ψ i }{ I 2 v 2
with:
∫d
3
rψ *i (r, t)ψ j (r, t) = δ ij
[8.94]
Ω
In eq. [8.94] the dot indicates time derivative, MI are the nuclear masses and µ are arbitrary parameters having the dimension of mass. Using eq. [8.94] it is possible to generate a dynamics for {Ri}, {αv} and {ψi} through the following equations: && = −∇ V MIR I RI && v = − ∂V µv α ∂α v
[8.95]
[8.96]
472
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
&& (r, t) = − µψ
∂V + ∑ Λ ik ψ k (r, t) ∂ψ *i (r, t) k
[8.97]
In eq. [8.97] the Λ are the Lagrange multipliers introduced to satisfy eq. [8.94]. It is worth noticing that, while the nuclear dynamics [8.95] can have a physical meaning, that is not true for the dynamics associated with the {αv} and {ψi}: this dynamics is fictitious, like the associated “masses” µ. If µ and the initial conditions {ψi}0 and {dψi/dt}0 are chosen such that the two classical sets of degrees of freedom (nuclear and electronic) are only weakly coupled, the transfer of energy between them is small enough to allow the electrons to adiabatically follow the nuclear motion, then remaining close to their instantaneous BO surface. In such a metastable situation, meaningful temporal averages can be computed. The mentioned dynamics is meant to reproduce what actually occurs in real matter, that is, electrons adiabatically following the nuclear motion. QM potentials have been widely used in molecular dynamics simulation of liquid water using the CP DFT algorithm. See, for example, refs. [72,73]. 8.7.2.2 Semi-classical simulations Computer simulations are methods addressed to perform “computer experimentation”. The importance of computer simulations rests on the fact that they provide quasi-experimental data on well-defined models. As there is no uncertainty about the form of the interaction potential, theoretical results can be tested in a way that is usually impossible with results obtained by experiments on real liquids. In addition, it is possible to get information on quantities of no direct access to experimental measures. There are basically two ways of simulating a many-body system: through a stochastic process, such as the Monte Carlo (MC) simulation,74 or through a deterministic process, such as a Molecular Dynamics (MD) simulation.75,76 Numerical simulations are also performed in a hybridized form, like the Langevin dynamics42 which is similar to MD except for the presence of a random dissipative force, or the Brownian dynamics,42 which is based on the condition that the acceleration is balanced out by drifting and random dissipative forces. Both the MC and the MD methodologies are used to obtain information on the system via a classical statistical analysis but, whereas MC is limited to the treatment of static properties, MD is more general and can be used to take into account the time dependence of the system states, allowing one to calculate time fluctuations and dynamic properties. In the following, we shall briefly describe the main features of MD and MC methodologies, focusing the attention to their use in the treatment of liquid systems. Molecular dynamics Molecular Dynamics is the term used to refer to a technique based on the solution of the classical equation of motion for a classical many-body system described by a many-body Hamiltonian H. In a MD simulation, the system is placed within a cell of fixed volume, usually of cubic shape. A set of velocities is assigned, usually drawn from a Maxwell-Boltzmann distribution suitable for the temperature of interest and selected to make the linear momentum equal to zero. Then the trajectories of the particles are calculated by integration of the classical equation of motion. It is also assumed that the particles interact through some forces,
8.7 Theoretical and computing modeling
473
whose calculation at each step of the simulation constitutes the bulk of the computational demand of the calculation. The first formulation of the method, due to Alder,75 referred to a system of hard spheres: in this case, the particles move at constant velocity between perfectly elastic collisions, so that it is possible to solve the problem without making any approximations. More difficult is the solution of the equations of motion for a set of Lennard-Jones particles: in fact, in this case an approximated step-by-step procedure is needed, since the forces between the particles change continuously as they move. Let us consider a point Γ in the phase space and suppose that it is possible to write the instantaneous value of some property A as a function A(Γ). As the system evolves in time, Γ and hence A(Γ) will change. It is reasonable to assume that the experimentally observable “macroscopic” property Aobs is really the time average of A(Γ) over a long time interval: Aobs = A
time
= A( Γ(t))
time
t
1 A( Γ(t))dt t→∞ t ∫ 0
= lim
[8.98]
The time evolution is governed by the well-known Newton equations, a system of differential equations whose solution is practical. Obviously it is not possible to extend the integration to infinite time, but the average can be reached by integrating over a long finite time, at least as long as possible as determined by the available computer resources. This is exactly what is done in a MD simulation, in which the equations of motion are solved step-by-step, taking a large finite number τ of steps, so that: Aobs =
1 ∑ A( Γ(τ)) τ τ
[8.99]
It is worth stressing that a different choice in the time step is generally required to describe different properties, as molecular vibrations (when flexible potentials are used), translations and rotations. A problem arising from the methodology outlined above is whether or not a suitable region of the phase space is explored by the trajectory to yield good time averages (in a relatively short computing time) and whether consistency can be obtained with simulations with identical macroscopic parameters but different initial conditions. Generally, thermodynamically consistent results for liquids can be obtained provided that careful attention is paid to the selection of initial conditions. Monte Carlo As we have seen, apart from the choice of the initial conditions, a MD simulation is entirely deterministic. By contrast, a probabilistic (stochastic) element is an essential part of any Monte Carlo (MC) simulation. The time-average approach outlined above is not the only possible: it is in fact practical to replace the time average by the ensemble average, being the ensemble a collection of points Γ distributed according to a probability density ρ(Γ) in the phase space. The density is determined by the macroscopic parameters, NPT, NVT, etc., and generally will evolve in time. Making the assumption that the system is “ergodic”, the time average in eq. [8.98] can be replaced by an average taken over all the members of the ensemble at a particular time: Aobs = A
ensemble
= ∑ A(Γ)ρ ensemble (Γ) Γ
[8.100]
474
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Sometimes, instead of using the ρ(Γ), a “weight” function w(Γ) is used: ρ(Γ) =
w(Γ)
[8.101]
Q
Q = ∑ w(Γ) Γ
A
ensemble
=
[8.102]
∑ w(Γ)A(Γ) ∑ w(Γ) Γ
[8.103]
Γ
Q is the partition function of the system. From Eq. [8.102], it is possible to derive an approach to the calculation of thermodynamics properties by direct evaluation of Q for a particular ensemble. Q is not directly estimated, but the idea of generating a set of states in phase space sampled in accordance with the probability density ρ(Γ) is the central idea of MC technique. Proceeding exactly as done for MD, replacing an ensemble average as in Eq. [8.103] with a trajectory average as in Eq. [8.99], a succession of states is generated in accordance with the distribution function ρNVE for the microcanonical NVE ensemble. The basic aim of the MC method (so-called because of the role that random numbers play in the method), which is basically a technique for performing numerical integration, is to generate a trajectory in phase space that samples from a chosen statistical ensemble. It is possible to use ensembles different from the microcanonical: the only request is to have a way (physical, entirely deterministic or stochastic) of generating from a state Γ(τ) a next state Γ(τ + 1). The important point to be stressed is that some conditions have to be fulfilled: these are that the probability density ρ( Γ ) for the ensemble should not change as the system evolves, any starting distribution ρ( Γ ) should tend to a stationary solution as the simulation proceeds, and the ergodicity of the systems should hold. With these recommendations, we should be able to generate from an initial state a succession of points that in the long term are sampled with the desired probability density ρ( Γ ). In this case, the ensemble average will be the same as a “time average” (see Eq. [8.99]). In a practical simulation, τ runs over the succession of states generated following the previously mentioned rules, and is a large finite number. This approach is exactly what is done in an MC simulation, in which a trajectory is generated through phase space with different recipes for the different ensembles. In other words, in a MC simulation a system of particles interacting through some known potential is assigned a set of initial coordinates (arbitrarily chosen) and a sequence of configurations of the particles is then generated by successive random displacements (also called ‘moves’). If f(R) is an arbitrary function of all the coordinates of the molecule, its average value in the canonical ensemble is: f
∫ f (r ) e = ∫e
− βV ( R )
− βV ( R )
dR
dR
where β = 1/kT and V(R) is the potential energy.
[8.104]
8.7 Theoretical and computing modeling
475
In the MC methods the integration is done by re-writing
[8.105]
e − βV ( R k ) ∑ e − βV (R k )
[8.106]
k
Pk =
k
The search for the most probable configuration Rk is done using Markov’s chain theory, so that after setting the system at a given configuration Ri, another configuration Rj is selected, making a random move in the coordinates. Not all configurations are accepted; the decision on whether or not to accept a particular configuration is made in such a way as to ensure that asymptotically the configuration space is sampled according to the probability density appropriate to a chosen ensemble. In particular: if
( )
V R j < V(Ri )
the move is accepted in the chain. Otherwise, if
( )
V R j > V(Ri )
the move is subjected to a second screening and accepted if
e
− β (V j −V i )
<γ
where γ is a random number between 0 and 1. The original models used in MC were highly idealized representations of molecules, such as hard spheres and disks, but nowadays MC simulations are carried out on the basis of more reliable interaction potential. The use of realistic molecule-molecule interaction potential makes it possible to compare data obtained from experiments with the computer generated thermodynamic data derived from a model. The particle momenta do not enter the calculation, there is no scale time involved, and the order in which the configurations occur has no special significance. It is worth noticing that, because only a finite number of states can be generated in a simulation, the results obtained from a MC simulation are affected by the choice of initial conditions, exactly as previously said for the MD results. As we have already said, it is possible to extend the MC method, originally formulated in the microcanonical ensemble, to other ensembles. Particularly important in the study of liquid systems is the extension of the Monte Carlo method to the grand canonical µVT ensemble:76 that is due to the fact that such an extension permits to calculate the free energy of the system, a quantity of particular significance in the chemistry and physics of liquids. The grand canonical MC involves a two-stage process: the first stage is exactly identical to what is done in the conventional MC, with the particles moving and the move accepted or rejected as previously said. The second stage involves an attempt either to insert a particle at a randomly chosen point in the cell or to remove a particle that is already present. The deci-
476
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
sion whether to make an insertion or a removal is made randomly with equal probabilities. The trial configuration is accepted if the pseudo-Boltzmann factor: WN = e
[Nβµ −log N ! − βV
N
]
(r N )
[8.107]
increases. If it decreases, the change is accepted when equal to: W N +1 1 [ βµ − β (VN +1 −VN ] e = WN N+1
[8.108]
for the insertion, or to: W N −1 − βµ − β (V N +1 −V N ] = Ne [ WN
[8.109]
for the removal. This method works very well at low and intermediate densities. As the density increases, it becomes difficult to apply, because the probability of inserting or removing a particle is very small. As we have already said, the grand canonical Monte Carlo provides a mean to determining the chemical potential, and hence, the free energy of the system. In other MC and MD calculations a numerical value for the free energy can always be obtained by means of an integration of thermodynamic relations along a path which links the state of interest to one for which the free energy is already known, for example, the dilute gas or the low-temperature solid. Such a procedure requires considerable computational effort, and it has a low numerical stability. Several methods have been proposed and tested. The difficulties in using MC and MD arise from the heavy computational cost, due to the need of examining a large number of configurations of the system usually consisting of a large number of particles. The size of the system one can study is limited by the available storage on the computer and by the speed of execution of the program. The time taken to evaluate the forces or the potential energy is proportional to N2. Other difficulties in molecular simulations arise from the so-called quasi-ergodic problem,77 i.e., the possibility that the system becomes trapped in a small region of the phase space. To avoid it, whatever the initial conditions, the system should be allowed to equilibrate before starting the simulation, and during the calculation, the bulk properties should be carefully monitored to detect any long-time drift. As already said, apart from the initial conditions, the only input information in a computer simulation are the details of the inter-particle potential, almost always assumed to be pair-wise additive. Usually in practical simulations, in order to economize the computing time, the interaction potential is truncated at a separations rc (the cut-off radius), typically of the order of three molecular diameters. Obviously, the use of a cut-off sphere of small radius is not acceptable when the inter-particle forces are very long ranged. The truncation of the potential differently affects the calculation of bulk properties, but the effect can be recovered by using appropriate “tail corrections”. For energy and pressure for monoatomic fluids, for example, these “tail corrections” are obtained by evaluating
8.7 Theoretical and computing modeling
477
the following integrals [8.110] and [8.111] between the limits r=rc and r=∞, and assuming that g(r)=1 for r>rc: E = 2πρ ∫ v(r ) g(r ) r 2 dr N
[8.110]
2πβρ βP = 1− v(r ) g(r ) r 3 dr 3 ∫ ρ
[8.111]
As already mentioned above, the extent of “tail corrections” strongly depends on the property under study. In the case of pressure, this correction can be very large; for example, in the case of a Lennard-Jones potential truncated at rc=2.5σ, at conditions close to the triple point, E/Nε=-6.12, to which the tail correction contributes -0.48, but βP/ρ=0.22, of which -1.24 comes to the tail. The neglecting of the tail correction would lead to a strongly positive value for pressure.42 Usually, when a cut-off radius is exploited, for the calculation of the interaction between a particle and the others, the “nearest-neighbor” convention is used. It means that, assuming the particles of interest lie in a cell and that this basic unit is surrounded by periodically repeated images of itself, a particle i lying on the central cell will interact only with the nearest image of any other particle j. The interaction is then set to zero if the distance between i and the nearest image is greater than rc. The usual assumption rc
478
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
(non-Euclidean geometry has to be considered) are expected to decrease as the system increases, so that in the case of solutes of large size, “spherical boundary conditions” are expected to be a noticeable method for simulating bulk liquids. In this case, boundary conditions can be obtained by inserting the spherical cell into a large sphere and then by performing a simulation of a lower computational level in the extended region. However, recent versions of the method tend to replace the second portion of the liquid, described at a molecular level, with a description by means of continuum models (see Section 8.7.3), which can be easily extended up to r=∞. Turning attention to the accuracy gained in the calculation of liquids properties by molecular simulations, it turns out that certain properties can be calculated more accurately than others. For example, the mean potential energy can be obtained with an uncertainty less than 1%; larger errors are associated with the calculation of thermodynamic properties, such as, the specific heat, that are linked to fluctuations in a microscopic variable. Typically, an uncertainty of the order of 10% is expected.42 The dynamical properties of liquids can be generally computed by calculating the time-correlation functions. They provide a quantitative description of the microscopic dynamics in liquids. Here, computer simulations play a key role, since they give access to a large variety of correlation functions, many of which are not measurable by laboratory experiments. In a MD simulation, the value of the correlation function CAB(t) at time t is calculated by averaging the product A(t+s)B(s) of two dynamical variables, A(t) and B(t) over many choices of the time origin s. It has been shown that the uncertainty in time correlations between events separated by an interval τ increases as τ1/2. In addition, the correlated motion of large numbers of particles determines many dynamical properties. These collective properties are usually subjected to larger errors than in the case of single particle properties, and their calculation is very demanding from a computational point of view.42 QM/MM As final remark of this section, we would like to briefly mention the so-called QM/MM method.78,79 In this approach, a part of the system (the solute M) is treated explicitly by a quantum mechanical (QM) method, while another part (the bulk solvent S) is approximated by a standard Molecular Mechanics (MM) force field. It is clear that such a method takes advantage of the accuracy and generality of the QM treatment, and of the computational efficiency of the MM calculation. Using the BO approximation and assuming that no charge transfer occurs between S and M, the Hamiltonian of the system can be separated into three terms: H$ = H$ M + H$ MS + H$ SS
[8.112]
where the first term is the Hamiltonian of the solute in vacuo, the second one represents the QM/MM solute-solvent interaction and couples solvent effects into QM calculations. It can be separated in an electrostatic term, a van der Waals terms and a polarization term, i.e.: el $ pol + H$ vdW H$ MS = H$ MS MS + H MS
[8.113]
8.7 Theoretical and computing modeling
479
The third term in eq. [8.112] is the interaction energy between solvent molecules. It is usually approximated by the molecular mechanics force field, which in general contains bond stretching, angle bending, dihedral torsion and non-bonded terms (see Section 8.4.5). Once the Hamiltonian of the system is defined, the total energy of an instantaneous configuration sampled during an MC or MD simulation is determined as: E = Φ H$ Φ = E M + E MS + E SS
[8.114]
More details on the procedure can be found in literature.78 The forces used to integrate Newton’s equation of motion in MD simulations are determined by differentiating eq. [8.114] with respect to nuclear coordinates. In the practical QM/MM calculation, the solvent S is equilibrated first using simulations (or a RISM version of the integral equation approach, see Section 8.7.1.1), and then the electronic structure of the solute M is modified via an iterative QM procedure in the presence of a fixed potential of the S component. The procedure is repeated iteratively; at the next step the S distributions is determined again, now taking into account the modified description of M. This sequence of steps is repeated until convergence. A problem of such a methodology is that changes in the internal geometry of M must be treated apart, scanning point-by point the relevant PES, not being available analytical expressions for first and second derivatives of the energy with respect to nuclear coordinates, which would be necessary to take into account any modification in the solute geometry. To overcome problems arising from the finite system size used in MC or MD simulation, boundary conditions are imposed using periodic-stochastic approximations or continuum models.75 In particular, in stochastic boundary conditions the finite system is not duplicated but a boundary force is applied to interact with atoms of the system. This force is set as to reproduce the solvent regions that have been neglected. Anyway, in general any of the methods used to impose boundary conditions in MC or MD can be used in the QM/MM approach. 8.7.3 CONTINUUM MODELS Originally, continuum models of solvent were formulated as dielectric models for electrostatic effects. In a dielectric model the solvent is modeled as a continuous medium, usually assumed homogeneous and isotropic, characterized by a scalar, static dielectric constant ε. This model of the solvent, that can be referred to the original work by Born, Onsager and Kirkwood 60-80 years ago, assumes linear response of the solvent to a perturbing electric field due to the presence of the molecular solute. This simple definition then has been largely extended to treat more complex phenomena, including not only electrostatic effects; and nowadays continuum solvation models represent very articulate methodologies able to describe different systems of increasing complexity. The history of, and the theory behind, continuum solvation models have been described exhaustively in many reviews31a,80,81 and articles82-84 in the past, so we prefer not to repeat them here. In addition, so large and continuously increasing is the amount of examples of theoretical developments on one hand, and of numerical applications on the other, that we shall limit our attention to a brief review of the basic characteristics of these models which have gained wide acceptance and are in use by various research groups.
480
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Trying to give a more theoretically-stated definition of continuum models, but still keeping the same conciseness of the original one reported above, we have to introduce some basic notions of classical electrostatics. The electrostatic problem of a charge distribution ρM (representing the molecular solute) contained in a finite volume (the molecular cavity) within a continuum dielectric can be expressed in terms of the Poisson equation: r r r r r → ∇ ε(r )∇φ(r ) = −4πρ M (r )
[
]
[8.115]
where φ is the total electrostatic potential. Eq. [8.115] is subject to the constraint that the dielectric constant inside the cavity is εin=1, while at the asymptotic boundary we have: r r [8.116] lim rφ(r ) = α lim r 2 φ(r ) = β r →∞
r →∞
with α and β finite quantities. At the cavity boundary the following conditions hold: r r φin (r ) = φout (r )
r ∂φ ∂φin = ε out (r ) out ∂n$ ∂n$
[8.117]
where subscripts in and out indicate functions defined inside and outside the cavity, and n$ is the unit vector perpendicular to the cavity surface and pointing outwards. Further conditions can be added to the problem, and/or those shown above (e.g., the last one) can be modified; we shall consider some of these special cases later. As said above, in several models an important simplification is usually introduced, r i.e., the function ε out (r) is replaced by a constant ε (from now on we skip the redundant subscript out). With this simplification we may rewrite the electrostatic problem by the following equations: r r r inside the cavity: ∇ 2 φ(r ) = −4πρ M (r ), outside the cavity: ∇ 2 φ(r ) = 0 [8.118] Many alternative approaches have been formulated to solve this problem. In biochemistry, methods that discretize the Poisson differential operator over finite elements (FE) have long been in use. In general, FEM approaches do not directly use the molecular cavity surface. Nevertheless, as the whole space filled by the continuous medium is partitioned into locally homogeneous regions, a careful consideration of the portion of space occupied by the molecular solute has still to be performed. There is another family of methods, known as FD (finite difference) methods,85 which exploit point grids covering the whole space, but conversely to the FEM point, in FD methods the points are used to replace differential equations by algebraic ones. These methods can be applied both to the “simplified” model with constant ε, and to r models with space dependent ε( r ) or real charges dispersed in the whole dielectric medium. They aim at solving the Poisson equation (1) expressed as a set of finite difference equations for each point of the grid. The linear system to be solved has elements depending both on g
8.7 Theoretical and computing modeling
481
and ρM, and its solution is represented by a set of φ values in the grid points. These values have to be reached iteratively. Many technical details of the methods have been changed in the years. We signal, among others, the strategy used to define the portion of the grid, especially near the molecular surface where there is a sharp change in ε, and the mathematical algorithms adopted to improve convergence in the iterative processes. Almost all FD methods have been conceived for, and applied to, classical descriptions of solutes, generally represented by atomic charges. The latter are usually drawn from previous QM calculations in vacuo or from Molecular Mechanics effective potentials, as in the study of large molecules, usually proteins. The deficiencies due to the use of rigid charges are partially alleviated by introducing a dielectric constant greater than one for the inner space (ε between 2 and 4), and/or by including atomic polarization functions (akin to atomic dipole polarizabilities). The DelPhi program85a is one of these proposals. This method, which has been extensively improved during the years, exists now in a QM version86 exploiting a recent high level QM method, called PS-GVB,87 based on a synthesis of grid point integration techniques and standard Gaussian orbital methods. From a PS-GVB calculation in vacuo, a set of atomic charges, the potential derived (PD) charges, is obtained by a fitting of the molecular electrostatic potential (MEP). These charges are used in DelPhi to solve the Poisson equation in the medium, and then to get, with the aid of a second application of DelPhi in vacuo, the reaction field (RF). The latter is represented by a set of point charges on the cavity surface, which define the solute-solvent interaction potential to be inserted in the PS-GVB Hamiltonian. Then a new computational cycle starts, and the whole process is repeated until convergence is reached. In parallel, the classical FD procedure of Bashford and Karplus88 has been coupled with the Amsterdam density fuctional theory (DFT)89 code to give another version of QM-FD methods. Starting from a DFT calculation in vacuo, the RF potential is obtained as the difference between the solutions of the Poisson equation obtained with a FD method in the medium and in vacuo (in both cases, the electrostatic potential φM is expressed in terms of PD atomic charges). The RF potential is then re-computed to solve a modified Kohn-Sham equation. The calculations are repeated until convergence is achieved. An alternative to the use of finite differences or finite elements to discretize the differential operator is to use boundary element methods (BEM).90 One of the most popular of these is the polarizable continuum model (PCM) developed originally by the Pisa group of Tomasi and co-workers.32 The main aspect of PCM is to reduce the electrostatic Poisson equation (1) into a boundary element problem with apparent charges (ASCs) on the solute cavity surface. A method that is very similar to PCM is the conductor-like screening model (COSMO) developed by Klamt and coworker91 and used later by Truong and coworkers,92 and by Barone and Cossi.93 This model assumes that the surrounding medium is well modeled as a conductor (thus simplifying the electrostatics computations) and corrections are made a posteriori for dielectric behavior. Electrostatic BE methods can be supported by nonelectrostatic terms, such as dispersion and exchange, in different ways, and their basic theory can be extended to treat both classical and quantum mechanical solutes; in addition, many features, including analytical gradients with respect to various parameters, have been added to the original models so as to
482
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
make BE methods one of the most powerful tools to describe solvated systems of increasing complexity. More details on their main aspects will be given in the following. Further alternative methods for solving the Poisson equation are to represent the solute and the solvent reaction field by single-center multipole expansions (MPE), as first pioneered by the Nancy group of Rivail and coworkers.94 Originally, this formalism was adopted because it simplified computation of solvation free energies and analytic derivatives for solute cavities having certain ideal shapes, e.g., spheres and ellipsoids. The formalism has been extended to arbitrary cavity shapes and multipole expansions at multiple centers. An alternative method still adopting the same MPE approach is that of Mikkelsen and coworkers;95 this method has been implemented in the Dalton computational package.96 The idea of a distributed multipole expansion to represent the charge distribution is also employed by the so-called generalized Born (GB) approach to continuum solvation. In this case, however, only monopoles (i.e., atomic partial charges) are employed and instead of solving the Poisson equation with this charge distribution, one uses the generalized Born approximation. The most widely used solvation models adopting this procedure are the original generalized Born/surface area (GBSA) approach of Still and co-workers,97 including the subsequent extensions and adaptations of several groups in which the GB electrostatics are combined with a classical MM treatment of the solute, and the SMx models of Cramer, Truhlar and coworkers98 in which GB electrostatics are employed in a quantum mechanical treatment of the solute. In both of these approaches, semiempirical atomic surface tensions are used to account for nonelectrostatic solvation effects. For the SMx models, the value of x ranges from 1 to 5.42, corresponding to successively improved versions of the prescriptions for atomic charges and radii, the dielectric descreening algorithm, and the functional forms for the atomic surface tensions. 8.7.3.1 QM-BE methods: the effective Hamiltonian As anticipated above, in BE methods for solvation the basic feature is to solve the Poisson equation [8.115] with the related boundary conditions [8.117] (or in a simpler isotropic case the system [8.118]) by defining an apparent charge distribution σ on the solute cavity surface (such charge is usually indicated as ASC).31a This continuous charge which depends on both the solvent dielectric constant and the solute charge and geometry, is then discretized in terms of point charges each placed on a different portion of surface identified by a collocation point sk and an area ak. This partition of the cavity surface into patches, or tesserae, represents one of the main features of the method as well as the shape of the cavity itself, which here has not been limited to analytical simple shapes (as in the original MPE methods), but it can be well modeled according to the actual solute geometry. The particular aspect we shall treat here is the generalization of the basic theory of these methods to QM descriptions of the solute. To do that, one has to introduce the concept of the Effective Hamiltonian (EH); for the isolated system, the fundamental equation to solve is the standard Schrödinger equation H 0Ψ 0 = E 0Ψ 0
[8.119]
where the superscript 0 indicates the absence of any interactions with other molecules. In the presence of a solvent treated in the ASC-BE framework, an external perturbation has to be taken into account: the reaction field due to the apparent charges placed on the cavity surface. Such perturbation can be represented in terms of an electrostatic potential VR
8.7 Theoretical and computing modeling
483
to be added to the Hamiltonian H0 so to obtain an effective Hamiltonian H which will satisfy a different Schrödinger equation with respect to [8.119], namely, we have HΨ = EΨ
[8.120]
Both the wave function and the eigenvalue E will be modified with respect to [8.119], due to the presence of the solvent perturbation. An important aspect, until now not introduced, is that the solvent apparent charges, and consequently the reaction operator VR, depend on the solute charge, i.e., in the present QM framework, on the wave function they contribute to define. This mutual interactions between Ψ and VR induces a complexity in the problem which can be solved through the standard iterative procedures characterizing the self-consistent (SC) methods. Only for an aspect the calculation in solution has to be distinguished from that in vacuo way; the energy functional to be minimized in a variational solution of [8.120] is not the standard functional E but the new free energy functional G G = Ψ H 0 + 1 / 2V R Ψ
[8.121]
The 1/2 factor in front of VR accounts for the linear dependence of the operator on the solute charge (i.e., the quadratic dependence on Ψ). In a more physical description, the same factor is introduced when one considers that half of the interaction energy has been spent in polarizing the solvent and it has not been included in G. In the standard original model the perturbation VR is limited to the electrostatic effects (i.e., the electrostatic interaction between the apparent point charges and the solute charge distribution); however, extensions to include dispersion and repulsion effects have been formulated. In this more general context the operator VR can be thus partitioned in three terms (electrostatic, repulsive and dispersive), which all together contribute to modify the solute wave function. For clarity’s sake in the following we shall limit the analysis to the electrostatic part only, referring the reader to refs. [31,83], and to the original papers quoted therein, for details on the other terms. For a more complete report we recall that nonelectrostatic effects can be also included a posteriori, i.e., independently of the QM calculation, through approximated or semiempirical models still exploiting the definition of the molecular cavity. In PCM and related methods, the term corresponding to the energy spent in forming the solute cavity in the bulk liquid, is usually computed using the SPT integral equation method (see Section 8.7.1.1). Until now no reference to the specific form of the operator VR has been given, but here it becomes compulsory. Trying to remain at the most general level possible, we start to define the apparent charges q(sk) through a matrix equation, i.e., q = −M −1f (ρ)
[8.122]
where M is a square matrix depending only on the geometry of the system (e.g., the cavity and its partition in tesserae) and the solvent dielectric constant and f(ρ) is a column vector containing the values of a function f, which depends on the solute distribution charge ρ, computed on the cavity tesserae. The form of this function is given by the specific solvation
484
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
model one uses; thus, for example, in the original version of PCM it coincides with the normal component of the electrostatic field to the cavity surface, while for both the COSMO model91-93 and the revised PCM version, known as IEF (integral equation formalism),99 it is represented by the electrostatic potential. In any case, however, this term induces a dependence of the solvent reaction to the solute charge, and finally to its wave function; in a standard SC procedure it has to be recomputed (and thus also the apparent charges q) at each iteration. Also, the form of M depends on the solvation model; contrary to f, it can be computed once, and then stored, if the geometry of the system is not modified during the calculation. Once the apparent charges have been defined in an analytical form through eq. [8.122], it is possible to define the perturbation VR as VR =∫
q (sk) r − sk
ρ(ρ)dr
[8.123]
In particular, by distinguishing the source of ρ, i.e., the solute nuclei and electrons, we can always define two equations [8.122], one for each source of f, and thus compute two sets of apparent charges, qN and qe, depending on the solute nuclei and electrons, respectively. This allows one to partition the reaction operators in two terms, one depending on the geometry of the system and the solute nuclear charge, and the other still depending on the geometry and on the solute wave function. Such partition is usually introduced to get an effective Hamiltonian whose form resembles that of the standard Hamiltonian for the isolated system. In fact, if we limit our exposition to the Hartree-Fock approximation in which the molecular orbitals are expressed as an expansion over a finite atomic orbital basis set, the Fock operators one defines for the two systems, the isolated and the solvated one, become (over the atomic basis set) F0 = h + G(P)
F = h + j + G(P) +X(P)
[8.124]
where h and G(P) are the standard one and two-electron interaction matrices in vacuo (P represents the one-electron density matrix). In [8.124] the solvent terms are expressed by two matrices, j and X(P), whose form follows directly from the nuclei and electron-induced equivalents of expression [8.122], respectively; for the electronic part, ρ has to be substituted by P. The parallel form of F0 and F in [8.124] shows that the two calculations can be performed exactly in the same way, i.e., that solvent does not introduce any complication or basic modification to the procedure originally formulated for the isolated system. This is a very important characteristic of continuum BE solvation methods which can be generalized to almost any quantum mechanical level of theory. In other words, the definition of ‘pseudo’ one and two-electron solvent operators assures that all the theoretical bases and the formal issues of the quantum mechanical problem remain unchanged, thus allowing a stated solution of the new system exactly as in vacuo. Above, we have limited the exposition to the basic features of single point HF calculations, and to the thermodynamic functions (the solvation free energy is immediately given by eqs. [8.121-8.124]). However, extensions to other QM procedures as well as to other types of analysis (the evaluation of molecular response functions, the study of chemical re-
8.7 Theoretical and computing modeling
485
activity, the reproduction of spectra of various nature, just to quote the most common applications) have been already successfully performed.83 Many of these extensions have been made possible by the formulation and the following implementation of efficient procedures to get analytical derivatives of the free energy in solution with respect to various parameters.100 To understand the role of this methodological extension in the actual development of the solvation models we just recall the importance of an efficient computation of analytical gradients with respect to nuclear coordinates to get reliable geometry optimizations and dynamics.101 It is here important to recall that such improvements are not limited to BE solvation methods; for example, Rivail and the Nancy group102 have recently extended their multipole-expansion formalism to permit the analytic computation of first and second derivatives of the solvation free energy for arbitrary cavity shapes, thereby facilitating the assignment of stationary points on a solvated potential energy surface. Analytic gradients for SMx models at ab initio theory have been recently described103 (even if they have been available longer at the semiempirical level104), and they have been presented also for finite difference solutions of the Poisson equation and for finite element solutions. Many other extensions of different importance have been added to the original BE models in the last few years. Here it may be worth recalling a specific generalization which also has made BE models preferable in particular applications until now almost completely restricted to FE and FD methods. At the beginning of the present section we indicated the simplified system (4), valid only in the limit of an homogeneous and isotropic dielectric, the commonly studied problem; actually, this has been the only affordable system until recently, with some exceptions represented by systems constituted by two isotropic systems with a definite interface. The situation changed only a few years ago (1997-1998) when BE-ASC methods were extended to treat macroscopically anisotropic dielectrics (e.g., liquid crystals),99 ionic solutions (e.g., isotropic solutions with nonzero ionic strength),99 and supercritical liquids.105 Another important feature included in BE methods, as well in other continuum methods, are the dynamical effects. In the previous section devoted to Physical Models, we have recalled that an important issue of all the solvation methods is their ability to treat nonequilibrium, or dynamical, aspects. There, for clarity’s sake, we preferred to skip the analysis of this feature, limiting the exposition to static, or equilibrium problems; here, on the contrary, some comments will be added. In continuum models no explicit reference to the discrete microscopical structure of the solvent is introduced; on one hand, this largely simplifies the problem, but on the other hand it represents a limit which can effect the quality of the calculations and their extensibility to more complex problems such as those involving dynamical phenomena. However, since the first formulations of continuum models, different attempts to overcome this limit have been proposed. One of the theories which has gained more attention identifies the solvent response function with a polarization vector depending on time, P(t), which varies according to the variations of the field from which it is originated (i.e., a solute and/or an external field). This vector accounts for many phenomena related to different physical processes taking place inside and among the solvent molecules. In practice, strong approximations are necessary in order to formulate a feasible model to deal with such complex quantity. A very simple but
486
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
also effective way of seeing this problem is to partition the polarization vector in terms accounting for the main degrees of freedom of the solvent molecules, each one determined by a proper relaxation time (or by the related frequency). In most applications, this spectral decomposition is limited to two terms representing “fast” and “slow” phenomena, respectively106,107 P(t) = Pfast + Pslow
[8.125]
In particular, the fast term is easily associated with the polarization due to the bound electrons of the solvent molecules which can instantaneously adjust themselves to any change of the inducing field. The slow term, even if it is often referred to as a general orientational polarization, has a less definite nature. Roughly speaking, it collects many different nuclear and molecular motions (vibrational relaxations, rotational and translational diffusion, etc.) related to generally much longer times. Only when the fast and the slow terms are adjusted to the actual description of the solute and/or the external field, on one hand, and to each other on the other hand, do we have solvation systems corresponding to full equilibrium. Nonequilibrium solvation systems, on the contrary, are characterized by only partial response of the solvent. A very explicative example of this condition, but not the only one possible, is represented by a vertical electronic excitation in the solute molecule. In this case, in fact, the immediate solvent response will be limited to its electronic polarization only, as the slower terms will not able to follow such fast change but will remain frozen in the equilibrium status existing immediately before the transition. This kind of analysis is easily shifted to BE-ASC models in which the polarization vector is substituted by the apparent surface charge in the representation of the solvent reaction field.107 In this framework, the previous partition of the polarization vector into fast and slow components leads to two corresponding surface apparent charges, σf and σs, the sum of which gives the total apparent charge σ. The definition of these charges is in turn related to the static dielectric constant ε(0) (for the full equilibrium total charge σ), to the frequency dependent dielectric constant ε(ω) (for the fast component σf), and to a combination of them (for the remaining slow component σs). The analytical form of the frequency dependence of ε(ω) in general is not known, but different reliable approximations can be exploited. For example, if we assume that the solvent is polar and follows a Debye-like behavior, we have the general relation: ε(ω) = ε(∞) +
ε(0) − ε(∞) 1 − iωτ D
[8.126]
with ε(0)=78.5, ε(∞)=1,7756, and τD=0.85×10-11s in the specific case of water. It is clear that at the optical frequencies usually involved in the main physical processes, the value of ε(ω) is practically equal to ε(∞); variations with respect to this, in fact, become important only at ωτD <<1, where ε(ω) is close to the static value. In practice, this means that for a polar solvent like water, where ε(∞) is much smaller than the static value, the dynamical solvent effects are small with respect to the static ones. Numerically, this means that all processes described in terms of a static or a dynamic solvent approach, as well as solute response properties computed in the full solute-solvent equilibrium or, instead in a condition of nonequilibrium, will present quite different values.
8.8 Practical applications of modeling
487
As a last note we underline that, while it is easy to see that nonequilibrium effects have to be taken into account in the presence of fast changes in the electronic distribution of the solute, or of an oscillating external field (as that exploited to measure molecular optical properties), nonequilibrium approaches for nuclear vibrational analyses are still open.
8.8 PRACTICAL APPLICATIONS OF MODELING In this section we will present some examples of how to determine properties of pure liquids by means of the previously-shown computational procedures. In spite of the fact that most literature in this field is concerned with the study of the properties of water, we will rather focus the attention on the determination of the properties of a few organic solvents of common use (for a review of studies on water, we address interested readers to the already quoted paper of Floris and Tani).40 Obviously, the following discussion cannot be considered as a complete review of all the literature in the field, but it is rather intended to give the reader a few suggestions on how the previously-shown computational methodologies can be used to obtain data about the most usual properties of pure liquids. From what we said in the previous section, it should be clear to the reader that continuum models cannot properly be used in order to determine liquid properties: they are in fact concerned in the treatment of a solute in a solvent, and so not of direct use for the study of the solvent properties. For this reason, we will present in the following some examples concerning the application of integral equations and computer simulations. The analysis will be limited to the examination of the most usual static and dynamic properties of liquids. Dielectric constant Let us begin the discussion by considering a couple of examples regarding the calculation of the dielectric constant of pure liquids. The critical role of the dielectric constant in determining the properties of a liquid and in particular in moderating the intermolecular interactions cannot be overemphasized. It is then extremely important that potential energy functions used, for example, in computer simulations are able to reproduce the experimental value of the dielectric constant and that such a calculation can be used to estimate the validity of any empirical potential to be used in simulations. The problem of the calculation of the dielectric response is complex:108 many theories and methods are available, giving sometimes different and contrasting results. It is accepted that for the calculation of the dielectric constant (as well as other properties dependent on intermolecular angular correlations) via computer simulations, it is compulsory to include long-range electrostatic interactions in the treatment. Such interactions can be taken into account by using either the Ewald summation approach or using the reaction field approach. Without going into details, we simply recall here that in the Ewald approach, electrostatic interactions are evaluated using an infinite lattice sum over all periodic images, i.e., by imposing a pseudo-crystalline order on the liquid. In the reaction field approach, however, electrostatic interaction beyond the spherical cutoff (see Section 8.7.2 for details) are approximated by treating the part of the system outside the cutoff radius as a polarizable continuum of a given dielectric constant εrf. Several methods exist for the calculation of the dielectric constant by means of computer simulations. The most widely used is the calculation of the average of the square of the total dipole moment of the system, <M2> (fluctuation method). Using reaction field bound-
488
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
ary conditions, the dielectric constant ε of the liquid within the cutoff radius is calculated by applying the following expression, where V is the volume of the system, k the Boltzmann constant and T the temperature: 2 4π M ε − 1 2ε rf + 1 = 9 kVT 3 2ε rf + ε
[8.127]
In the application of eq. [8.127] careful attention has to be paid to the choice of εrf, the dielectric constant to use for the continuum part of the system. Usually, rapid convergence in calculations is obtained by choosing εrf ≈ ε. A different approach to the calculation of the dielectric constant via computer simulations is given by the polarization response method, i.e., to determine the polarization response of a liquid to an applied electric field E0. If
[8.128]
Even if this second approach is more efficient than the fluctuation method, it requires the perturbation of the liquid structure following the application of the electric field. As an example, such a methodology has been applied by Essex and Jorgensen109 to the calculation of the dielectric constant of formamide and dimethylformamide using Monte Carlo statistical mechanics simulations (see Section 8.7.2 for details). The simulation result for dimethylformamide, 32±2, is reasonably in agreement with respect to the experimental value, 37. However, in the case of formamide, the obtained value, 56±2, underestimates the experimental value of 109.3. The poor performance here addresses the fact that force field models with fixed charges underestimate the dielectric constant for hydrogen-bonded liquids. Other methodologies exist for the calculation of the static dielectric constant of pure liquids by means of computer simulations. We would like to recall here the use of ion-ion potentials of mean force,110 and an umbrella sampling approach whereby the complete probability distribution of the net dipole moment is calculated.111 Computer simulations are not the only methods which can be used to calculate the dielectric constant of pure liquids. Other approaches are given by the use of integral equations, in particular, the hypernetted chain (HNC) molecular integral equation and the molecular Ornstein-Zernike (OZ) theory (see Section 8.7.1 for details on such methodologies). The molecular HNC theory gives the molecular pair distribution function g(12) from which all the equilibrium properties of a liquid can be evaluated. It is worth recalling here that the HNC theory consists for a pure liquid in solving a set of two equations, namely, the OZ equation [8.76] and the HNC closure relation [8.78] and [8.80]. In order to solve the HNC equations, the correlation functions are written in terms of a series of rotational invariants Φ mnl µv (12 ), which are defined in terms of cosines and sines of the angles defining the relative orientation of two interacting particles 1 and 2:
8.8 Practical applications of modeling
489
mnl c(12) = ∑ c µmnl v (R)Φ µv (12)
[8.129]
mnl µv
where the coefficients c mnl µν (12 ) depend only on the distance between the mass centers of the particles, and the indexes m, n, l, µ, and v are integers satisfying the following inequalities: m − n ≤ l ≤ m + n,
− m ≤ µ ≤ m,
−n ≤v ≤n
[8.130]
The dielectric constant can be obtained from the rotational invariant coefficients of the pair correlation function. For example, following Fries et al.,112 it can be evaluated using the Kirkwood g factor: 4π
∞
ρ ∫ R 2 h 110 00 (R)dR
[8.131]
(ε − 1)(2ε + 1) 4π βρm e2 g = 9 9ε
[8.132]
g = 1−
3
0
In the previous relations, ρ is the number density of molecules, β=1/kT, and me is the effective dipole of molecules along their symmetry axes. Such an approach has been used to evaluate the dielectric constant of liquid acetonitrile, acetone and chloroform by means of the generalized SCMF-HNC (self-consistent mean field hypernetted chain), and the values obtained are 30.8 for acetonitrile, 19.9 for acetone, and 5.66 for chloroform. The agreement with the experimental data, 35.9 for acetonitrile, 20.7 for acetone, and 4.8 for chloroform (under the same conditions), is good. Another way of evaluating the dielectric constant is that used by Richardi et al.113 who applied molecular Ornstein-Zernike theory and eq. [8.131] and [8.132]. The obtained values for both acetone and chloroform are low with respect to experimental data, but in excellent agreement with Monte Carlo simulation calculations. In this case, the dielectric constant is evaluated by using the following expressions: N
(ε − 1)(2ε rf + 1) ρµ 2 = 3(2ε rf + ε) 9kTε 0
N
∑ ∑µ µ i =1 j =1
Nµ 2
i
j
[8.133]
where εrf is the dielectric constant of the continuum part of the system, i.e., the part outside the cutoff sphere, N the number of molecules of the system and µ the dipole moment. A further improvement in the mentioned approach has been achieved by Richardi et al.114 by coupling the molecular Ornstein-Zernike theory with a self-consistent mean-field approximation in order to take the polarizability into account. For the previously-mentioned solvents (acetone, acetonitrile and chloroform), the calculated values are in excellent agreement with experimental data, showing the crucial role of taking into account polarizability contributions for polar polarizable aprotic solvents.
490
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Thermodynamical properties Paying attention again to calculations for acetone and chloroform, we would like to mention two papers by the Regensburg group113,114 exploiting molecular Ornstein-Zernike (MOZ) theory and site-site Ornstein-Zernike (SSOZ) theory for the calculation of excess internal energies. In the MOZ framework, the internal excess energy per molecule is computed by a configurational average of the pair potential V acting between molecules: E ex =
ρ 2Ω 2
∫ ∫ ∫ V(R
12
, Ω 1, Ω 2 ) g (R12 , Ω 1, Ω 2 ) dR12 dΩ 1dΩ 2
[8.134]
with ρ the density of molecules. Using instead SSOZ, Eex can be directly calculated from the site-site distribution function, gαβ, as: ∞
k
k
E ex = 2πρ ∫ ∑ ∑ g αβ (r )Vαβ (r ) r 2 dr
[8.135]
0 α=1 β =1
where k is the number of sites per molecule. In both cases the calculated values are in good agreement with experimental data and with calculations done exploiting the Monte Carlo computer simulation. In this case, Eex can be obtained as: E ex =
2 2 ε rf − 1 z αz β e 0 rαβ, ij 1 N N k k V r + ∑ ∑ ∑ ∑ αβ αβ, ij 2ε + 1 4πε r 3 N i =1 j >1 α=1 β =1 rf 0 cutoff
(
)
[8.136]
where k is the number of sites per molecule, rαβ,ij is the distance between site α of molecule i and site β of molecule j. Only pairs i,j of molecules for which rij
8.8 Practical applications of modeling ∞
491
(
)
000 ρkTκ T = 1 + 4πρ ∫ g 00 (R12 ) − 1 R122 dR12
[8.137]
0
and exploiting SSOZ can be calculated from any site-site distribution function gαβ as: ∞
(
)
ρkTκ T = 1 + 4πρ ∫ g αβ (r ) − 1 r 2 dr
[8.138]
0
For acetone and chloroform, the obtained results are too large. Relaxation times and diffusion coefficients Let us now consider an example of how to calculate dynamic properties of pure liquids by computational methodologies. As already said in Section 8.7.2, molecular dynamics simulations are able to take into account the time-dependence in the calculation of liquid properties. Paying attention to pure acetone, Brodka and Zerda116 have calculated rotational relaxation times and translational diffusion coefficients by molecular dynamics simulations. In particular, the calculated rotational times of the dipole moment can be compared with a molecular relaxation time τM obtained from the experimentally determined τD by using the following expression, which considers a local field factor: τM =
2εs + ε ∞ τD 3εs
[8.139]
where εs and ε ∞ are the static and optical dielectric constants, respectively. The calculated data are in good agreement with experimental values. Diffusion coefficients can be calculated directly from the velocity correlation functions or from mean square displacements, as: Dvcf =
∞
1 v(t) v(0) dt 3 ∫0 2 1 r (t) r (0) i →∞ 6t
D msd = lim
[8.140]
[8.141]
The values obtained by using the [8.140] or the [8.141] are almost the same, and properly describe the experimental temperature and density dependencies of the diffusion coefficients, even if about 30% smaller than obtained using NMR spin-echo techniques. Among the pure solvents we have treated so far, other available data regard the calculation of the self diffusion coefficient (D) for liquid ethanol at different temperatures.115d The D parameter was obtained from the long-time slope of the mean-square displacements of the center of mass: experimental changes of D over the 285-320K range of temperatures were acceptably reproduced by molecular dynamics simulations.
492
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
Shear viscosity To end this section, we would like to report an example of calculation of the shear viscosity again taking as an example pure acetone. The viscosity can be computed by using molecular dynamics simulations from the virial form of the molecular pressure tensor P, which can be represented as a sum of four contributions: N
PV = ∑ i =1
pi pi + PSV + PR V + PC V mi
[8.142]
where V is the volume of the cell, N the number of molecules, pi and mi the momentum and mass of molecule i. The first term in eq. [8.142] is the kinetic contribution, while PS is the short-range potential interaction contribution, PR the reciprocal-space portion of the pressure tensor, and PC a correction tensor term. From the xy elements of the time-averaged molecular pressure tensor [8.142] and the applied shear (or strain) rate γ = ∂v x / ∂y, the viscosity η can be computed using the constitutive relation: η=−
Pxy − Pyx 2γ
[8.143]
the calculated values for acetone (extrapolated data) are in reasonable agreement with experimental data, as well as results for acetone/methanol and acetone/water mixtures.117
8.9 LIQUID SURFACES Until now our task has been relatively easy, as we have only considered the general methodological aspects of the computational description of liquids. To proceed further, we have to be more specific, entering into details of specific problems, all requiring the introduction and the characterization of different properties, and of the opportune methodological (computational) tools to treat them. We also note that, as additional complication, it can happen that things change considerably when the full range of macroscopic variables (temperature, pressure, volume) is considered, and when other components are added to the liquid system. Not much effort is needed to convince readers of the overwhelming complexity of the task we are here considering. Let us take a simple pure liquid: to span the range of the thermodynamical variables means to consider systems in which the liquid is in presence of the solid, of the vapor, including supercritical as well as super-cooled states, and all the related phenomena, as ebullition, vaporization, condensation and freezing. All states and all dynamical processes connecting states are of interest for chemistry, chemical engineering, and physical chemistry. Some properties can be defined for all the states (such as the thermodynamic basic quantities); others are specific for some situations, such as the surface tension. But also for properties of general definition, the techniques to use have often to be very different. It is not possible, for example, to use the same technique
8.9 Liquid surfaces
493
to study the temperature dependence of the Helmholtz free energy in a bulk liquid or in the same liquid under the form of a fog near condensation. Things become more complex when a solute is added to the liquid, and more and more complex when the system contains three or more components. We are not sure whether any case occurring in this field has been subjected to theoretical analysis and modeling; however, the number of problems considered by theoretical research is very large, and there are models, generally based on the use of molecular potentials, that can give hints, and often useful computational recipes. These remarks, which are trivial for every reader, with even a minimal experience in working with solvents, have been added here to justify our choice of not trying to give a schematic overview, and also to justify the extremely schematic presentation of the unique specific more advanced subject we shall here consider, the interfacial properties of liquids. 8.9.1 THE BASIC TYPES OF LIQUID SURFACES The surface of a liquid is often represented as a very small fraction of the liquid specimen, and it may be neglected. This statement is not of general validity, however, and often the surface plays a role; in some cases, the dominant role. To be more specific, it is convenient to consider a classification of liquid surfaces, starting from a crude classification, and introducing more details later. The basic classification can be expressed as follows: • Liquid/gas surfaces; • Liquid/liquid surfaces; • Liquid/solid surfaces. Liquid/gas surfaces may regard a single substance present in both phases, liquid and vapor at equilibrium or out of equilibrium, but they may also regard a liquid in the presence of a different gas, for example, air (in principle, there is always a given amount of solvent in the gas phase). In the presence of a different gas, one has to consider it in the description of the liquid, both in the bulk and in the interface. In many cases, however, the role of the gas is marginal, and this is the reason why in many theoretical studies on this interface the gas is replaced by the vacuum. This is a very unrealistic description for liquids at equilibrium, but it simply reflects the fact that in simplifying the model, solvent-solute interactions may be sacrificed without losing too much in the description of the interfacial region. The potentials used to describe these interfaces are those used to describe bulk liquids. The differences stay in the choice of the thermodynamical ensemble (grand canonical ensembles are often necessary), in the boundary conditions to be used in calculations, and in the explicit introduction in the model of some properties and concepts not used for bulk liquids, like the surface tension. Much could be said in this preliminary presentation of liquid/gas interfaces, but we postpone the few aspects we have decided to mention, because they may be treated in comparison with the other kind of surfaces. Liquid/liquid surfaces regard, in general, a couple of liquids with a low miscibility coefficient, but there are also cases of one-component systems presenting two distinct liquid phases, for which there could be interest in examining the interfacial region at the coexistence point, or for situations out of the thermodynamical equilibrium. If the second component of the liquid system has a low miscibility with the first, and its molar ratio is low, it is easy to find situations in which this second component is arranged as a thin layer on the surface of the main component with a depth reducing, in the opportune conditions, to the level of a single molecule. This is a situation of great practical importance
494
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
to what we experience in our everyday life (petrol on the surface of water); however, it has to be related to a finer classification of liquid interfaces we shall consider later. Once again the potentials used for liquid/liquid surfaces are in general those used for bulk liquids eventually with the introduction of a change in the numerical values of the parameters.118 The difference from the bulk again regards the thermodynamical ensemble, the boundary conditions for the calculations and the use of some additional concepts. The liquid/solid surfaces may be of very different types. They can be classified according to the electrical nature of the solid as: • Liquid/conductor • Liquid/dielectric The presence of mobile electrons in metals and other solid substances (to which some liquids may be added, like mercury119), introduces in the modelling of the liquid interface new features not present in the previously-quoted interfaces nor in the bulk liquids. For many phenomena occurring at such interfaces there is the need for taking into explicit account electrons flowing from one phase to the other, and of related electron transfers occurring via redox mechanisms. The interaction potentials have to be modified and extended accordingly.120 The good electrical conductivity of the solid makes more sizeable and evident the occurrence of phenomena related to the presence of electric potentials at the interface (similar phenomena also occur at interfaces of different type, however121). A well-known example is the double layer at the liquid side of an electrolyte/electrode surface. For the double layer, actually there is no need of interaction potentials of special type: the changes in the modelling mainly regard the boundary conditions in the simulation or in the application of other models, of continuum or integral equation type. It is clear that attention must be paid to the electrostatic part of the interaction. Models for the solid part similar to those used for bulk liquids play a marginal role. There are models of LJ type, but they are mainly used to introduce in the solid component the counterpart of the molecular motions of molecules, already present in the description of the liquid (vibrations of the nuclei: phonons, etc.). The electrostatic problem may be treated with techniques similar to those we have shown for the continuum electrostatic model for solutions but with a new boundary condition on the surface, e.g., V=constant. In this case, a larger use is made (especially for planar surfaces) of the image method,122 but more general and powerful methods (like the BEM) are gaining importance. An alternative description of the metal that can be profitably coupled to the continuum electrostatic approach is given by the jellium model. In this model, the valence electrons of the conductor are treated as an interacting electron gas in the neutralizing background of the averaged distribution of the positive cores (some discreteness in this core distribution may be introduced to describe atomic motions). The jellium is described at the quantum level by the Density Functional Theory (DFT) and it represents a solid bridge between classical and full QM descriptions of such interfaces:123 further improvements of the jellium which introduce a discrete nature of the lattice124 are now often used.125 The liquid/dielectric surfaces exhibit an even larger variety of types. The dielectric may be a regular crystal, a microcrystalline solid, a glass, a polymer, an assembly of molecules held together by forces of different type.
8.9 Liquid surfaces
495
The potentials in use for crystals and those used for covalently bound systems are cognate with those used for liquids, so it is not compulsory to open here a digression on this theme. We have already quoted the MM methods generally used for solids of molecular type; for more general classes of compounds, a variety of ‘atom-atom’ potentials have been given.126 The difference with respect to the bulk is mainly due, again, to different ways of handling the opportune boundary conditions. It is evident that the techniques used to describe a water-protein interface greatly differ, in microscopic detail, time scale, etc., to those used to describe the surfaces between a liquid and a regular crystal surface. For many problems, the solid surface in contact with the liquid may be held fixed: this characteristic has been amply exploited in simulation studies. We quote here two problems for which this simplification is not possible: problems in which the structure of the solid (or solid-like) component is subjected to conformational changes, and problems in which a progressive expansion (or reduction) of the solid phase plays the main role. To the first case belong problems addressing the structure of the interfacial zone in the case of biopolymers (e.g., proteins) or similar covalently bound massive molecules, and ordered assemblies of molecules held together by strong non-covalent interactions (as membranes or similar structures). In both cases, the liquid and the solid component of the systems are generally treated on the same footing: interaction force fields are used for both components, introducing, where appropriate, different time scales in the molecular dynamics calculations. The second case mainly regards the problem of crystal growth or dissolution in a liquid system (pure liquid or solution). A mobile surface of this type is not easy to treat with standard methods: it is better to pass to more specialistic approaches, for which there is the need of introducing a different mathematical treatment of the (mobile) boundary surface conditions. This is a subject that we cannot properly treat here. 8.9.2 SYSTEMS WITH A LARGE SURFACE/BULK RATIO To complete this quick presentation of the three main types of liquid surfaces, it is convenient to consider systems in which the interfacial portion of the liquid is noticeably larger than in normal liquids. Some among them are denoted in literature as constrained liquids, others as dispersed systems. This classification is not universal, and it does not take in account other cases. A typical example of constrained liquid is that present in pores or capillaries. The surface is of liquid/solid type, and the same remarks about interaction potentials and computational methods we have already offered may be applied here. More attention must be paid in computations of interactions among separate portions of the liquid interface. In the slit pore case (i.e., in a system composed of two parallel plane solid surfaces, with a thin amount of liquid between them), there may be an interference between the two distinct surfaces, more evident when the thickness of the liquid is small, or when there is an electrical potential between the two faces. The slit pore model is extensively used to describe capillarity problems: in the most usual cylindrical capillary types there is a radial interaction similar in some sense to the interaction with a single solid surface. Pores of different shape are present in many solid materials: zeolites and other aluminosilicates are outstanding examples. When the pore is very small there is no more distinction between superficial and bulk molecules of the liquid. Nature (and human ingenuity) offers examples over the complete range of situations. For all the systems we have here introduced (quite important in principle and for practical reasons), there is no need of
496
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
introducing important changes in the standard theoretical approach with respect to that used for other liquid systems. Computer simulations are the methods more largely used to describe pores and micro-pores in molecular sieves of various nature. A different type of confined liquid is given by a thin layer spread on a surface. The difference with respect to the micro-pore is that this is a three-component system: the surface may be a solid or a liquid, while the second boundary of the layer may regard another liquid or a gas. An example is a thin layer liquid on a metal, in the presence of air (the prodrome of many phenomena of corrosion); a second example is a layer on a second liquid (for example, oil on water). The standard approach may be used again, but other phenomena occurring in these systems require the introduction of other concepts and tools: we quote, for example, the wetting phenomena, which require a more accurate study of surface tension. The thin layer in some case exhibits a considerable self-organization, giving rise to organized films, like the Blodgett-Langmuir films, which may be composed by several layers, and a variety of more complex structures ending with cellular membranes. We are here at the borderline between liquid films and covalently held laminar structures. According to the degree of rigidity, and to the inter-unit permeation interaction, the computational model may tune different potentials and different ways of performing the simulation. Another type of confined liquid regards drops or micro-drops dispersed in a medium. We are here passing from confined liquids to the large realm of dispersed systems, some among which do not regard liquids. For people interested in liquids, the next important cases are those of a liquid dispersed into a second liquid (emulsions), or a liquid dispersed in a gas (fogs), and the opposite cases of a solid dispersed in a liquid (colloids) or of a gas dispersed in a liquid (bubbles). Surface effects on those dispersed systems can be described, using again the standard approach developed in the preceding sections: intermolecular potentials, computer simulations (accompanied where convenient by integral equation or continuum approaches). There is not much difference with respect to the other cases. A remark of general validity but particular for these more complex liquid systems must be expressed, and strongly underlined. We have examined here a given line of research on liquids and of calculations of pertinent physico-chemical properties. This line starts from QM formal description of the systems, then specializes the approach by defining intermolecular potentials (two- or many-body), and then it uses such potentials to describe the system (we have paid more attention to systems at equilibrium, with a moderate extension to non-equilibrium problems). This is not the only research line conceivable and used, and, in addition, it is not sufficient to describe some processes, among them, those regarding the dispersed systems we have just considered. We dedicate here a limited space to these aspects of theoretical and computational description of liquids because this chapter specifically addresses interaction potentials and because other approaches will be used and described in other chapters of the Handbook. Several other approaches have the QM formulation more in the background, often never mentioned. Such models are of a more classical nature, with a larger phenomenological character. We quote as examples the models to describe light diffraction in disordered systems, the classical models for evaporation, condensation and dissolution, the transport of the matter in the liquid. The number is fairly large, especially in passing to dynamical and
8.9 Liquid surfaces
497
out of equilibrium properties. Some models are old (dating before the introduction of QM), but still in use. A more detailed description of molecular interactions may lead to refinements of such models (this is currently being done), but there is no reasonable prospect of replacing them with what we have here called the standard approach based on potentials. The standard approach, even when used, may be not sufficient to satisfactorily describe some phenomena. We quote two examples drawn from the presentation of liquid surfaces in dispersed systems: bubbles and solid small particles. The description of the energetics of a single bubble, and of the density inside and around it, is not sufficient to describe collective phenomena, such as ebullition or separation of a bubble layer. The description of a single bubble is at a low level in a ladder of models, each introducing new concepts and enlarging the space (and time) scale of the model. The same holds for small solid particles in a liquid. There is a whole section of chemistry, colloidal chemistry, in which our standard approach gives more basic elements, but there is here, again, a ladder of models, as in the preceding example. Our standard model extends its range of applicability towards these more complex levels of theory. This is a general trend in physics and chemistry: microscopic approaches, starting from the study of the interaction of few elementary particles, are progressively gaining more and more importance over the whole range of sciences, from biology to cosmology, passing through engineering and, in particular, chemistry. In the case of liquids, this extension of microscopical models is far from covering all the phenomena of interest. 8.9.3 STUDIES ON INTERFACES USING INTERACTION POTENTIALS To close this section on liquid surfaces, limited to a rapid examination of the several types of surface of more frequent occurrence, we report some general comments about the use of the standard approach on liquid surfaces. The first information coming from the application of the method regards the density profile across the interface. Density may be assumed to be constant in bulk liquids at the equilibrium, with local deviations around some solutes (these deviations belong to the family of cybotactic effects, on which something will be said later). At each type of liquid surface there will be some deviations in the density, of extent and nature depending on the system. A particular case is that of mobile surfaces, i.e., liquid/gas and liquid/liquid surfaces. Having liquids a molecularly grained structure, there will always be at a high level of spatial deviation a local deviation of the surface from planarity. If there are no other effects, this deviation (sometimes called corrugation, but the term is more convenient for other cases, like liquid/solid surfaces, where corrugation has a more permanent status) averages to zero. Entropic forces are responsible for other effects on the density profile, giving origin at a large scale to a smooth behavior of the density profile, connecting the constant values of the two bulk densities. The most natural length unit in liquids is the bulk correlation length, ξ, i.e., the distance at which the pair correlation gAB(r) has decayed to 1. For bulk water ξ is around 5-6 Å. This large scale behavior hides different local behaviors, that range among two extremes, from capillary waves to van der Waals regime.127 Capillary waves correspond to a sharp boundary (corrugation apart) with “fingers” of a liquid protruding out of the bulk. Capillary waves may happen both at liquid and gas surfaces, in the former, in a more stable status (i.e., with a large mean life) than in the latter.
498
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
The van der Waals regime locally corresponds to the large scale energy profile, with a smooth shape. The behavior of the density profile largely depends on the chemical nature of the composite system. Some phenomenological parameters, like mutual solubility, may give a good guess about the probable energy profile, but “surprises” are not rare events. It may be added that the density profile is sensitive to the quality of the potential and the method using the potentials. It is not clear, in our opinion, if some claims of the occurrence of a thin void (i.e., at very low density) layer separating two liquid surfaces is a real finding or an artifact due to the potentials. Old results must be considered with caution, because computations done with old and less powerful computers tend to reduce the size of the specimen, thus reducing the possibility of finding capillary waves. The dimension of the simulation box, or other similar limiting constraints, is a critical parameter for all phenomena with a large length scale (this is true in all condensed systems, not only for surfaces). The density profile for liquid/solid surfaces has been firstly studied for hard spheres on hard walls, and then using similar but a bit more detailed models. It is easy to modify the potential parameters to have an increment of the liquid density near the wall (adhesion) or a decrement. Both cases are physically possible, and this effect plays an important role in capillarity studies. A corrugation at the solid surface can be achieved by using potentials for the solid phase based on the atomic and molecular constitution. This is almost compulsory for studies on systems having a large internal mobility (biomolecules are the outstanding example), but it is also used for more compact solid surfaces, such as metals. In the last quoted example the discreteness of the potential is not exploited to address corrugation effects but rather to introduce mobility (i.e., vibrations. or phonons) in the solid part of the system, in parallel to the mobility explicitly considered by the models for the liquid phase. Another type of mobility is that of electrons in the solid phase: this is partly described by the continuum electrostatic approach for dielectrics and conductors, but for metals (the most important conductors) it is better to resort to the jellium model. A QM treatment of jellium models permits us to describe electric polarization waves (polarons, solitons) and their mutual interplay with the liquid across the surface. Jellium, and the other continuum descriptions of the solid, have the problem of exactly defining the boundary surface. This is the analog of the problem of the definition of the cavity boundary for solutes in bulk solvents, occurring in continuum solvation methods (see Section 8.7.3). The only difference is that there are more experimental data for solutes than for liquid/solid surfaces to have hints about the most convenient modeling. The few accurate ab-initio calculations on liquid/metal systems are of little help, because in order to reach an acceptable accuracy, one is compelled to reduce the solid to a small cluster, too small to describe effects with a large length scale. The disturbances to the bulk density profile are often limited to a few solvent diameters (typically 2-3), but the effect may be larger, especially when external electric fields are applied. The second information derived from the application of our methods to surfaces regards the preferred orientation of liquid molecules near the surface. To treat this point there is the need of a short digression. In bulk homogeneous liquids at the equilibrium orientation, the location of a specific molecule is immaterial. We need to know the orientational modes (often reduced to libra-
8.9 Liquid surfaces
499
tions) of the components of the liquid system to write the appropriate statistical partition function. In the case of solvation energies, the rotational component of the factorized solute partition function gives a not negligible contribution to ∆Gsolv . The translational component of the solute partition function can be factorized apart, as first suggested by Ben-Naim,128 and included in the cratic component of ∆Gsolv with a small correction (the cratic component is the numerical factor taking into account the relationships between standard states in a change of phase). In the homogeneous bulk liquid we have to explicitly consider orientations for dynamical relaxation effects. The same holds for translation in other dynamical problems (diffusion, transport, etc.). Coming back to the static problems, it turns out that near the limiting surface, the free energy of the components of the liquid also depends on positions and orientations. A quantity like ∆Gsolv for the transfer of M in the solvent S in the bulk must be replaced by ∆Gsolv (r, Ω) specifying position and orientation of M with respect to the surface of S. This effect must not be confused with the cybotactic effects we have mentioned, nor with the hole in the solute-solvent correlation function gMS(r) (see Figure 8.5). The hole in the radial correlation function is a consequence of its definition, corresponding to a conditional property, namely that it gives the radial probability distribution of the solvent S, when the solute M is kept at the origin of the coordinate system. Cybotactic effects are related to changes in the correlation function gMS(r) (or better gMS(r, Ω)) with respect to a reference situation. Surface proximity effects can be derived by the analysis of the gMS(r, Ω) functions, or directly computed with continuum solvation methods. It must be remarked that the obtention of gMS(r) functions near the surface is more difficult than for bulk homogeneous liquids. Reliable descriptions of gMS(r, Ω) are even harder to reach. Orientational preferences are the origin of many phenomena, especially when at the surface there are electric charges, mobile or fixed. Outstanding examples occur in biological systems, but there is a large literature covering other fields. A relatively simple case that has been abundantly studied is the electric potential difference across the surface.129 We have not considered yet the behavior of solutes near the interface. The special, but quite important, case of the charged components of a salt solution near a charged boundary is widely known, and it is simply mentioned here. More generally, the behavior of solutes largely depends on the chemical composition of the system. Small ions are repelled at the air/gas interface, hydrophobic molecules are attracted to the interface. Even the presence of a simple CH3 group has some effect: the air interface of dilute solutions of methanol is saturated in CH3OH over a sizeable molar fraction range.130 At the liquid/liquid interface between water and hydrocarbon solvents, solutes with a polar head and an alkylic chain tend to find an equilibrium position, with the polar head in water, at a distance from the surface depending on the chain length.131 All the quoted results have been obtained at the static level, employing a mean force potential, obtained in some cases with inexpensive continuum models; in other cases, with computer demanding simulations of the free energy profile. Simulations can give more details than continuum methods, which exploit a simplified and averaged model of the medium. Among the simulation results we quote the recognition that small cations, exhibiting strong interactions with water molecules, tend to keep bound the first solvation layer when forced to pass from water to another solvent. Analogous effects are under examination for neutral polar solutes at water/liquid interfaces in case
500
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
the bulk solubility of the two partners is not symmetric (a typical example is the water/octanol system).132 We have here mentioned the process of transport of a solute from a liquid phase to another. This is a typical dynamic process for which static descriptions based on mean forces simply represent the starting-point, to be completed with an explicit dynamical treatment. Molecular dynamics is the chosen tool for these studies, which are quite active and promise illuminating results. MD simulations also are the main approach to study another important topic: chemical reactions at the surface.133 The study of the various aspects of chemical reactions in bulk liquids, from the thermodynamical balance of the reaction, to the mechanistic aspects, to the reaction rates, represents nowadays a well explored field, in which several approaches may be used. The complexity of chemical systems does not permit us to consider this problem completely settled under the methodological and computational point of view, but important advances have been achieved. Reactions at the surface are a frontier field, and the same remarks hold for reactions in dispersed liquid systems. Interaction potentials, simulations and other methods will find here renewed stimuli for more precise and powerful formulations.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
M. Eyring, J. Walter, and G.E. Kimball, Quantum Chemistry, Wiley, New York, 1944. R. Bonaccorsi, R. Cimiraglia, P. Palla, and J. Tomasi, Int. J. Quant. Chem., 29, 307 (1983). J. Tomasi, B. Mennucci, and R. Cammi in Molecular Electrostatic Potentials. Concepts and Applications, J.S. Murray, K. Sen, Eds., Elsevier, Amsterdam, 1996, pp.1-103. K. Kitaura, K. Morokuma, Int. J. Quant. Chem., 10, 325 (1976). F. London, Z. Phys., 60, 245 (1930). R. G. Parr, and W. Yang, Density Functional Theory of Atoms and Molecules, Oxford University Press, Oxford, 1989. (a) R. Mc Weeny, Croat. Chem. Acta, 57, 865 (1984); (b) P. Claverie, in Molecules in Physics, Chemistry and Biology, J. Maruani Ed., Kluwer, Dordrecht, 1988, vol.2, pp.393. (a) R. Cammi, and J. Tomasi, Theor. Chim. Acta, 69, 11 (1986); (b) S. Nagase, T. Fueno, S. Yamabe, and K. Kitaura, Theor. Chim. Acta, 49, 309 (1978). S.F. Boys, and F. Bernardi, Mol. Phys., 19, 953 (1970). R. Cammi, R. Bonaccorsi, and J. Tomasi, Theor. Chim. Acta, 68, 271 (1985). G. Alagona, C. Ghio, R. Cammi, and J. Tomasi, in Molecules in Physics, Chemistry and Biology, J. Maruani Ed., Kluwer, Dordrecht, 1988, vol.2, pp.507. F.B. van Duijneveldt, J.G.C.M. van Duijneveldt-van de Rijdt, and J.H. van Lenthe, Chem. Rev., 94, 1873 (1994). (a) B. J. Jeziorski, R. Moszynski, and K. Szalewicz, Chem. Rev., 94, 1887 (1994); (b) K. Szalewicz, B. Jeziorski, Mol. Phys., 38, 191 (1979). R. Eisenschitz, and F. London, Z. Phys., 60, 491 (1930) G.P. Arrighini, Intermolecular Forces and Their Evaluation by Perturbation Theory, Springer, Berlin 1981. P. Claverie in Intermolecular Interactions: from Diatomics to Biopolimers, B. Pullmann, Ed., Wiley, New York, 1978. B. J. Jeziorski, and K. Kolos in Molecular Interactions, H. Rateiczak, and W. J. Orville-Thomas, Eds., Wiley, Chichester, 1978, pp. 29. (a) E. Scrocco, and J. Tomasi, Topics Curr. Chem., 42, 97 (1973); (b) E. Scrocco, and J. Tomasi, Adv. Quant. Chem., 11, 115 (1978). J. Tomasi in Chemical Applications of Atomic and Molecular Electrostatic Potentials, P. Politzer, and D. G. Truhlar, Eds., Plenum, New York, 1981. J. Tomasi, R. Bonaccorsi, and R. Cammi in Theoretical Models of Chemical Bonding, Z. Maksic, Ed., Springer, Berlin, 1991, part IV. G. Alagona, R. Cimiraglia, E. Scrocco, and J. Tomasi, Theor Chim. Acta, 25, 103 (1972).
8 Interactions in solvents and solutions
22 23 24 25 26
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
50 51 52 53 54 55 56 57 58 59 60
501
F. A. Momany, J. Phys. Chem., 82, 598 (1978). J. Mulliken, J. Chem. Phys., 23, 1833 (1955). F.W. Bader, Atoms in Molecules : A Quantum Theory, Oxford University Press, Oxford, 1994. (a) R. Bonaccorsi, E. Scrocco, and J. Tomasi, J. Am. Chem. Soc., 98, 4049 (1976); (b) M.M. Francl, R.F. Hout, and W.J. Hehre, J. Am. Chem. Soc., 106, 563 (1984). (a) A. Wallqvist, P. Ahlström, and G. Karlström, J. Phys. Chem., 94, 1649 (1990); (b) P. -O Åstrand, A. Wallqvist, and G. Karlström, J. Chem. Phys., 100, 1262 (1994); (c) P. -O Åstrand, P. Linse, and G. Karlström, Chem. Phys., 191, 195 (1995). (a) M.P. Hodges, A.J. Stone, and S.S. Xantheas, J. Phys. Chem. A, 101, 9163 (1997); (b) C. Millot, J-C. Soetens, M.T.C. Martins Costa, M.P. Hodges, and A.J. Stone, J. Phys. Chem. A, 102, 754 (1998). (a) L. Helm, and A.E. Merbach, Coord. Chem. Rev., 187, 151 (1999); (b) H. Ontaki, and T. Radnai, Chem. Rev., 93, 1157 (1993). G. Chalasinski, J. Rak, M.M. Sczesniak, S.M. Cybulski, J. Chem. Phys., 106, 3301 (1997). L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). (a) J. Tomasi, and M. Persico, Chem. Rev., 94, 2027 (1994); (b) C. J. Cramer, and D. G. Truhlar, Chem. Rev., 99, 2161 (1999). (a) S. Miertuš, E. Scrocco, and J. Tomasi, Chem. Phys., 55, 117 (1981); (b) R. Cammi, and J. Tomasi, J. Comp. Chem., 16, 1449 (1995). (a) F.M. Floris, M. Persico, A. Tani, and J. Tomasi, Chem. Phys., 195, 207 (1995); (b) F.M. Floris, A. Tani, and J. Tomasi, Chem. Phys., 169, 11 (1993). F.M. Floris, M. Persico, A. Tani, and J. Tomasi, Chem. Phys. Lett., (a) 199, 518 (1992); (b) 227, 126 (1994). J.G. Gay, and B.J. Berne, J. Chem. Phys., 74, 3316 (1981). H. Zewdie, J. Chem. Phys., 108, 2117 (1998). Y. Liu, and T. Ichiye J. Phys. Chem., 100, 2723 (1996). (a) H. Reiss, H. L. Frisch, J.L. Lebowitz, J. Chem. Phys., 31, 369 (1959); (b) H. Reiss, Adv. Chem. Phys., 9, 1 (1965). M. Sprik in Computer Simulation in Chemical Physics, M.P. Allen and D.J. Tildesley, Eds., Kluwer, Dordrecht, 1993. F.M. Floris, and A. Tani in Molecular Dynamics: from Classical to Quantum Methods, P.B. Balbuena, and J.M. Seminario, Eds., Elsevier, Amsterdam, 1999; pp. 363-429. A. Wallqvist, and R.D. Mountain, Rev. Comp. Chem., 13, 183 (1999). J. -P. Hansen, and I. R. McDonald, Theory of Simple Liquids, Academic Press, London, 1986. C. G. Gray, and K. E. Gubbins, Theory of Molecular Fluids. Volume 1: Fundamentals, Clarendon Press, Oxford, 1984. J. A. Barker, and D. Henderson, Rev. Mod. Phys., 48, 587 (1976). R.M. Gibbons, Mol. Phys., 17, 81 (1969); 18, 809 (1970). J.O. Hirschfelder (Ed.), Intermolecular Forces, Adv. Chem. Phys., 12 (1967). H. Margenau, and N. R. Kestner, Theory of Intermolecular Forces, Pergamon, Oxford, 1971; 2nd Edition, Chapt. 5. D.A. McQuarrie, Statistical Mechanics, Harper and Row, New York, 1976. (a) B.J. Berne, and, R. Pecora, Dynamic Light Scattering, Wiley, New York, 1976; (b) J.T. Hynes, Ann. Rev. Phys. Chem., 28, 301 (1977); (c)D. R. Bauer, J.I. Brauman, and R. Pecora, Ann. Rev. Phys. Chem., 27, 443 (1976). H. L. Frisch, and J.L. Lebowitz, Eds., The equilibrium theory of classical fluids, Benjamin, New York, 1964. J.K. Percus, and G. J. Yevick, Phys. Rev., 110, 1 (1958). J.L. Lebovitz, and J.K. Percus, Phys. Rev., 144, 251 (1966). (a) D. Chandler, and H.C. Andersen, J. Chem. Phys., 57, 1930 (1972); (b) L. R. Pratt, and D. C. Chandler, J. Chem. Phys., 67, 3683 (1977). F. Hirata, P.J. Rossky, J. Chem. Phys., 78, 4133 (1983). D. Chandler, R. Silbey, and B.M. Ladanyi, Mol. Phys., 46, 1335, (1982). B.C. Perng, M.D. Newton, F.O. Raineri, H.L. Friedman, J. Chem. Phys., 104, 7153 (1996); 7177 (1996). S. Ten-no, F. Hirata, and S. Kato, Chem. Phys. Lett., 214, 391 (1993); (b) H. Sato, F. Hirata, and S. Kato, J. Chem. Phys., 105, 1546 (1996); (c) L. Shao, H.A. Yu, J.L. Gao, J. Phys. Chem. A, 102, 10366 (1998). (a) H.C. Andersen, D. Chandler, and J.D. Weeks, Adv. Chem. Phys., 34, 105 (1976); (b) T. Boublik, Fluid Phase Equilibria, 1, 37 (1977); (c) M.S. Wertheim, Ann. Rev. Phys. Chem., 30, 471 (1979). J.S. Rowlinson, Mol. Phys., 8, 107 (1964). J. A. Barker, and D. Henderson, J. Chem. Phys., 47, 4714 (1967).
502
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
80 81 82 83 84 85
86 87
88 89 90 91 92 93 94
95
96
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli J.D. Weeks, D. Chandler, and H.C. Andersen, J. Chem. Phys., 54, 5237 (1971). J.A. Barker, J. Chem. Phys., 19, 1430 (1951). (a) H.C. Andersen, and D. Chandler, J. Chem. Phys., 57, 1918 (1972); (b) B.M. Ladanyi, and D. Chandler, J. Chem. Phys., 62, 4308 (1975). D.J. Tildesley, Mol. Phys., 41, 341 (1980) M. Lombardero, J.L.F. Abascal, and S. Lago, Mol. Phys., 42, 999 (1981). M. P. Allen, and D. J. Tildesley, Computer Simulation of Liquids, Clarendon Press, Oxford, 1987. D. Frenkel, and B. Smit, Understanding Molecular Simulation, Academic Press, San Diego, 1996. R. Car, and M. Parrinello, Phys. Rev. Lett., 55, 2471 (1985). G. Galli, and M. Parrinello in Computer Simulation in Materials Science, M. Meyer, and V. Pontikis, Eds., Kluwer, Dordrecht, 1991. W. Kohn, and L. J. Sham, Phys. Rev., 140, A1133 (1965). S. Kirkpatrik, C. D. Gelatt, Jr., and M. P. Vecchi, Science, 220, 671 (1983). K. Laasonen, M. Sprik, M. Parrinello, and R. Car, J. Chem. Phys., 99, 9080 (1993). E. S. Fois, M. Sprik, and M. Parrinello, Chem. Phys. Lett., 223, 411 (1994). M. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. N. Teller, and E. Teller, J. Chem. Phys., 21, 1087 (1953). B. J. Alder, and T. E. Wainwright, J. Chem. Phys., (a) 27, 1208 (1957); (b) 31, 459 (1959). D. Nicholson, and N. G. Parsonage, Computer Simulation and the Statistical Mechanics of Adsorption, Academic Press, New York, 1982. W. W. Wood, and F. R. Parker, J. Chem. Phys., 27, 720 (1957). J. Gao, Rev. Comp. Chem., 7, 119 (1996). C. S. Pomelli, and J. Tomasi in The Encyclopedia of Computational Chemistry, P. V. R. Schleyer, N. L. Allinger, T. Clark, J. Gasteiger, P. A. Kollman, H. F. Schaefer III, and P. R. Schreiner, Eds., Wiley, Chichester, 1998, vol. 4, p. 2343-2350. J.-L. Rivail, and D. Rinaldi in Computational Chemistry, Review of Current Trends, J. Leszczynski, Ed., World Scientific, New York, 1995; pp. 139. O. Tapia, in Quantum Theory of Chemical Reactions, R. Daudel, A. Pullman, L. Salem, and A. Viellard, Eds., Reidel, Dordrecht, 1980; Vol. 2, pp. 25. M. Orozco, C. Alhambra, X. Barril, J.M. Lopez, M. Busquest, and F.J. Luque, J. Mol. Model., 2, 1 (1996); C.Amovilli, V. Barone, R. Cammi, E. Cancès, M. Cossi, B. Mennucci, C.S. Pomelli, J. Tomasi, Adv. Quant. Chem., 32, 227 (1998). J.G. Angyan, J. Math. Chem., 10, 93 (1992). (a) B. Honig, K. Sharp, and A.S. Yang, J. Phys. Chem., 97, 1101 (1993); (b) V.K. Misra, K.A. Sharp, R.A. Friedman, and B. Honig, J. Mol. Biol., 238, 245 (1994); (c) S.W. Chen, and B. Honig, J. Phys. Chem. B, 101, 9113 (1997). R.A. Friesner, Ann. Rev. Phys. Chem., 42, 341 (1991) M. Ringnalda, J.M. Langlois, B. Greeley, R. Murphy, T. Russo, C. Cortis, R. Muller, B. Marten, R. Donnelly, D. Mainz, J. Wright, W.T. Polland, Y. Cao, Y. Won, G. Miller, W.A. Goddard III, and R.A. Friesner, PS-GVB Schrodinger Inc. (1994). D. Bashford, and M. Karplus, Biochemistry, 29, 10219 (1990). G. Te Velde, and E.J. Baerends, J. Comp. Phys., 99, 84 (1992). D. E. Beskos, Boundary Element Methods in Mechanics, North Holland, Amsterdam, 1997. (a) A. Klamt, and G. Schuurmann, J. Chem. Soc., Perkin Trans., 2, 799 (1993); (b) J. Andzelm, C. Kolmel, and A. Klamt, J. Chem. Phys., 103, 9312 (1995). (a) T.N. Truong and E.V. Stefanovich, J. Chem. Phys., 103, 3709 (1995); (b) E.V. Stefanovich and T.N. Truong, J. Chem. Phys., 105, 2961 (1996). V. Barone, and M. Cossi, J. Phys. Chem. A, 102, 1995 (1998). D. Rinaldi, and J.-L. Rivail, Theor. Chim. Acta, 32, 57 (1973); (b) J.-L. Rivail, and D. Rinaldi, Chem. Phys., 18, 233 (1976); (c) J.-L. Rivail, B. Terryn, and M.F. Ruiz-Lopez, J. Mol. Struct. (Theochem), 120, 387 (1985). (a) K.V. Mikkelsen, E. Dalgaard and P. Swanstrøm, J. Phys. Chem., 91, 3081 (1987); (b) K.V. Mikkelsen, H. Ågren, H.J.Aa. Jensen and T. Helgaker, J. Chem. Phys., 89, 3086 (1988); (c) K.V. Mikkelsen, P. Jørgensen and H.J.Aa. Jensen, J. Chem. Phys., 100, 6597 (1994); (d) K.V. Mikkelsen, Y. Luo, H. Ågren and P. Jørgensen, J. Chem. Phys., 100, 8240 (1994). DALTON, Release 1.0 (1997), T.Helgaker, H.J.Aa. Jensen, P. Jørgensen, J. Olsen, K. Ruud, H. Ågren, T. Andersen, K.L. Bak, V. Bakken, O. Christiansen, P. Dahle, E.K. Dalskov, T. Enevoldsen, B. Fernandez,
8 Interactions in solvents and solutions
97
98 99
100 101 102 103 104 105 106
107
108 109 110 111 112 113 114 115
116 117 118 119 120 121 122 123 124 125 126 127 128
503
H. Heiberg, H. Hettema, D. Jonsson, S. Kirpekar, R. Kobayashi, H. Koch, K.V. Mikkelsen, P. Norman, M.J.Packer, T. Saue, P.R. Taylor, and O. Vahtras. W.C. Still, A. Tempczyk, R.C. Hawley, and P. Hendrickson, J. Am. Chem. Soc., 102, 6127 (1990); (b) D. Qiu, P.S. Shemkin, F.P. Hollinger, and W.C. Still, J. Phys. Chem. A, 101, 3005 (1997); (c) M.R. Reddy, M.D. Erion, A. Agarwal, V.N. Viswanadhan, D.Q. McDonald, and W.C. Still, J. Comput. Chem., 19, 769 (1998). C.J. Cramer and D.G. Truhlar, (a) J. Am. Chem. Soc., 113, 8305 (1991); (b) Science, 256, 213 (1992); (c) J. Comp. Chem., 13, 1089 (1992); (d) J. Comput.-Aid. Mol. Des., 6, 629 (1992); and references in (29b). (a) E. Cancès, B. Mennucci and J. Tomasi, J. Chem. Phys., 107, 3032 (1997); (b) B. Mennucci, E Cancès, and J. Tomasi, J. Phys. Chem. B, 101, 10506 (1997); (c) E. Cancès, and B. Mennucci, J. Math. Chem., 23, 309 (1998). (a) R. Cammi, and J. Tomasi, J. Chem. Phys., 100, 7495 (1994); (b) R. Cammi, and J. Tomasi, J. Chem. Phys., 101, 3888 (1994). (a) E. Cancès, and B. Mennucci, J. Chem. Phys., 109, 249 (1998); (b) E. Cancès, B. Mennucci, and J. Tomasi, J. Chem. Phys., 109, 260 (1998); (d) B. Mennucci, R. Cammi, J. Tomasi, J. Chem. Phys., 110, 6858 (1999). V. Dillet, D. Rinaldi, J. Bertran, and J.-L. Rivail, J. Chem. Phys., 104, 9437 (1996). T. Zhu, J. Li, D.A. Liotard, C.J. Cramer, and D.G. Truhlar, J. Chem. Phys., 110, 5503 (1999). Y.-Y. Chuang, C.J. Cramer, and D.G. Truhlar, Int. J. Quantum Chem., 70, 887 (1998). C.S. Pomelli, and J. Tomasi, J. Phys. Chem. A, 101, 3561 (1997). (a) Y. Ooshika, J. Phys. Soc. Jpn., 9, 594 (1954); (b) R. Marcus, J. Chem. Phys., 24, 966 (1956); (c) E.G. McRae, J. Phys. Chem., 61, 562 (1957); (d) S. Basu, Adv. Quantum. Chem., 1, 145 (1964); (e) V.G. Levich, Adv. Electrochem. Eng., 4, 249 (1966); (f) H.J. Kim, and J.T. Hynes, J. Chem. Phys., 93, 5194 and 5211 (1990); (g) A.M. Berezhkosvskii, Chem. Phys., 164, 331 (1992); (h) M.V. Basilevski, and G.E. Chudinov, Chem. Phys., 144, 155 (1990); (i) K.V. Mikkelsen, A. Cesar, H. Ågren, and H. J. Aa. Jensen, J. Chem. Phys., 103, 9010 (1995). (a) M.A. Aguilar, F.J. Olivares del Valle, and J. Tomasi, J. Chem. Phys., 98, 7375 (1993); (b) R. Cammi, J. Tomasi, Int. J. Quantum Chem: Quantum Chem. Symp., 29, 465 (1995); (d) M.L. Sanchez, M.A. Aguilar, and F.J. Olivares del Valle, J. Phys. Chem., 99, 15758 (1995); (e) B. Mennucci, R. Cammi, and J. Tomasi, J. Chem. Phys., 109, 2798 (1998); (f) B. Mennucci, A. Toniolo, and C. Cappelli, J. Chem. Phys., 110, 6858 (1999). S. H. Glarum, Mol. Phys., 24, 1327 (1972). J. W. Essex, and W. L. Jorgensen, J. Phys. Chem., 99, 17956 (1995). L. X. Dang, and B. M. Pettitt, J. Phys. Chem., 94, 4303 (1990). Z. Kurtovic, M. Marchi, and D. Chandler, Mol. Phys., 78, 1155 (1993). P. H. Fries, J. Richardi, and H. Krienke, Mol. Phys., 90, 841 (1997). J. Richardi, P. H. Fries, R. Fischer, S. Rast, and H. Krienke, Mol. Phys., 93, 925 (1998). J. Richardi, P. H. Fries, and H. Krienke, Mol. Phys., 96, 1411 (1999). (a) W. L. Jorgensen, J. Phys. Chem., 90, 1276 (1986); (b) J. Gao, D. Habibollazadeh, L. Shao, J. Phys. Chem., 99, 16460 (1995); (c) M. E. van Leeuwen, Mol. Phys., 87, 87 (1996); (d) L. Saiz, J. A. Padro, and E. Guardia, J. Phys. Chem. B, 101, 78 (1997). A. Brodka, and T. W. Zerda, J. Chem. Phys., 104, 6313 (1996). D. R. Wheeler, and R. L. Rowley, Mol. Phys., 94, 555 (1998). A.R. Vanbuuren, S.J. Marriuk, and H.J.C. Berendsen, J. Phys. Chem., 97, 9206 (1993). J.C. Shelley, G.N. Patey, D.R. Berard, and G.M. Torrie, J. Chem. Phys., 107, 2122 (1997). J.C. Shelley, and D.R. Berard, Rev. Comp. Chem., 12, 137 (1998). (a) M. Wintherhalter, and W. Helfrich, J. Phys. Chem., 96, 327 (1992); (b) W. Lorenz, J. Phys. Chem., 95, 10566 (1991); (c) S.J. Miklavic, J. Chem. Phys., 103, 4795 (1995). (a) A. Krämer, M. Vassen, and F. Foistman, J. Chem. Phys., 106, 2792 (1996); (b) R.D. Berard, G.N. Patey, J. Chem. Phys., 97, 4372 (1992); (c) R.J. Clay, N.R. Goel, and P.F. Buff, J. Chem. Phys., 56, 4245 (1972). (a) P. Gies, and R.R. Gerhardts, Phys. Rev. B, 33, 982 (1986); (b) R.D. Berard, M. Kinoshita, X. Ye, and G.N. Patey, J. Chem. Phys., 107, 4719 (1997). J.R. Mannier, and J.P. Perdew, Phys. Rev. B, 17, 2595 (1978). R. Saradha, M.V. Sangaranarayanan, J. Phys. Chem. B, 102, 5099 (1998). D.W. Brenner, O.A. Shenderova, and D.A. Areshkin, Rev. Comp. Chem., 12, 207 (1998). L.-J. Chen, M. Knackstedt, and M. Robert, J. Chem. Phys., 93, 6800 (1990). A. Ben-Naim, J. Phys. Chem., 82, 792 (1978); Water and Aqueous Solutions, Plenum Press, New York, 1987.
504
Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli
129 (a) A.J. Bard, and L.R. Faulkner, Electrochemical Methods: Fundamentals and Applications, Wiley, New York, 1980; (b) A.W. Adamson, Physical Chemistry of Surfaces, Wiley, New York, 1990; (c) L.R. Pratt, J. Phys. Chem., 96, 25 (1992). 130 M. Matsumoto, Y. Takaota, and Y. Kataoka, J. Chem. Phys., 98, 1464 (1993). 131 R. Bonaccorsi, E. Ojalvo, and J. Tomasi, Coll. Czech. Chem. Comm., 53, 2320 (1988). 132 D. Michael, and I. Benjamin, J. Phys. Chem., 99, 16810 (1994). 133 I. Benjamin, (a) in Molecular Dynamics: from Classical to Quantum Methods, P.B. Balbuena, and J.M. Seminario, Eds., Elsevier, Amsterdam, 1999; pp. 661-701; (b) Chem. Rev., 96, 1449 (1996).
9
Mixed Solvents Y. Y. Fialkov, V. L. Chumak Department of Chemistry National Technical University of Ukraine, Kiev, Ukraine
9.1 INTRODUCTION One of the basic problems in conducting a chemical process in solution is control of the process parameters. The most important is the yield of the reaction products and process rates. The equilibrium constants and the rate constants of processes in solutions are multi-factor dependencies, that is, they depend on temperature and many solvent properties. The realization of the process in individual solvent often complicates such controls, and in some cases makes it completely impossible. At the same time, it is possible to choose properties determining the process characteristics in mixed solvents directly. Such properties relate, first of all, to the density, ρ, viscosity, η, permittivity, ε, and specific solvation energy. In this chapter, the term “mixed solvent” refers to binary solvents - regardless of the relation of their components. The use of mixed solvents in scientific and industrial practice began at the beginning of the 20th century. But systematic theoretical-experimental research began in the sixties, and its intensity has been growing ever since. There are two basic reports on physical properties of mixed solvents. One is the published in four-volume reference edition by Timmermans1 in the fifties and published at that period by Krestov and co-author’s monograph.2 The detailed substantiation of information as well as the corresponding bibliography are included in the monograph.3
9.2 CHEMICAL INTERACTION BETWEEN COMPONENTS IN MIXED SOLVENTS The developments in physical chemistry and chemical physics of liquid state substantiated common viewpoint that liquids are associated. Assuming this approach as a limit, we may consider the individual solvent as divided into non-associated molecules having smaller energy of intermolecular interaction than the energy of molecules in thermal motion, and to molecules, having energy higher than kT. 9.2.1 PROCESSES OF HOMOMOLECULAR ASSOCIATION In mixed solvents formed of associated component A and non-associated component B (does not interact with component A), the following equilibrium takes place:
506
Y. Y. Fialkov, V. L. Chumak B B B mA ← → 2Am / 2 ← →L ← → Am
[9.1]
where: m
number of molecules A forming the homomolecular associate (in subscript - association degree)
The systems alcohol-carbon tetrachloride or carbon acids-cyclohexane are examples of mixed solvents of this type. Interesting examples of such mixed solvents are the systems formed of liquid tetraammonium salts, R4NA, and various liquids that are solvate-inert towards such compounds. The values of homomolecular association constants of alcohols and acids are high, typically in the range of 103-104. Therefore concentration of the component A is rather low, even at high content of solvate-inert component B, and may be neglected. The relative degree of the decomposition of heteromolecular associates into aggregates of a smaller association degree increases with increased permittivity of the non-associated solvate-inert component. In solutions of equal concentration of the two mixed solvents, acetic acid-cyclohexane (ε=1,88) and acetic acid-chlorobenzene (ε=5,6), the relative degree of acid heteromolecular association is higher in the first system than in the second system. In mixed solvents, formed of two associated components in individual states Am and Bn that do not interact with each other into specific solvation, the chemical equilibrium of mixed heteromolecular associates is established: Am + B n ↔ x (Am / x B n / x )
[9.2]
where: x
number of molecules of heteromolecular associate
Whether to consider these heteromolecular associates as real chemical adducts is more terminology than a chemical problem. In most cases investigators truly assume that in systems formed of two alcohols or of two carboxylic acids, specific interaction does not exist. In the case of mixed solvents formed of two carbon acids, that is true only when components have similar proton affinity, as is the case of a system such as acetic acid-propionic acid (see further paragraph 9.2.8). 9.2.2 CONFORMIC AND TAUTOMERIC EQUILIBRIUM. REACTIONS OF ISOMERIZATION Various conformers of the same compound differ in their dipole moments. Thus, in the mixed solvents A-B, where A is a liquid whose molecules give conformer equilibrium, and B is solvate-inert towards A, the A conformer ratio changes on addition of B component. In binary solvents A-B where component A can exist in two tautomeric forms A1 and A2 the ratio of concentrations of these forms changes on addition of component B. Common theory of this influence has been worked out by Kabachnic4,5 who proposed the common scheme of equilibrium in systems of such type: [9.3]
9.2 Chemical interaction between components
507
The scheme [9.3] includes the following stages occurring in sequence: I interaction of the first tautomeric form, A1H, with the solvent that leads to formation of a addition product, A1HS, (the solvate composition may be more complicated) II ionization of solvate with formation of ionic couple, A1- HS + III ionic couple may decompose into ions (stage III a), transfer of ionized complex into a product of solvent addition already in the second tautomeric form IV formation of the free tautomeric form, A2H A2H. The nature of the chemical type of influences on keto-enol equilibrium is obvious: the more basic the solvent, the higher the degree of keto-enol transformation into enol form. If solvent is indifferent with enough degrees of approximation, the well-defined dependence of the constants of keto-enol equilibrium on solvent permittivity is encountered. The difference in dipole moments of tautomeric forms is the basis of these relationships. In mixed solvents such as ACR-B, the first component may undergo rearrangement reactions: ACR ↔ ARC
[9.4]
These occur under the solvent influence or even with its direct participation: +B +B →L ← → ARC ACR ←
[9.4a]
9.2.3 HETEROMOLECULAR ASSOCIATION When components of the binary solvent A-B are solvent-active to one another, the sum of chemical equilibrium is established: mA + nB ↔ Am B n
[9.5]
In most scenarios, the interaction proceeds in steps, i.e., equilibrium solution represents the mixture of heteroassociates of different stoichiometry: AB, AB2, A2B, etc. Systems formed by O-, S-, N-, P-bases with various H-donors (e.g., amines-carboxylic acids, esters-carboxylic acids (phenols), dimethylsulfoxide-carboxylic acid) refer to this type of interaction. The systems having components which interact by means of donor-acceptor bond (without proton transfer) belong to the same type of solvents (e.g., pyridine-chloracetyl, dimethylsulfoxide-tetrachloroethylene, etc.). Components of mixed solvents of such type are more or less associated in their individual states. Therefore, processes of heteromolecular association in such solvents occur along with processes of homomolecular association, which tend to decrease heteromolecular associations. 9.2.4 HETEROMOLECULAR ASSOCIATE IONIZATION In some cases, rearrangement of bonds leads to formation of electro-neutral ionic associate in binary solvents where heteromolecular associates are formed: Am B n ↔ K pq+Lpq−
[9.6]
508
Y. Y. Fialkov, V. L. Chumak
In many cases, ionic associate undergoes the process of ionic dissociation (see paragraph 9.2.5). If permittivity of mixture is not high, interaction can limit itself to formation of non-ionogenic ionic associate, as exemplified by the binary solvent anisol-m-cresol, where the following equilibrium is established: C6H 5OCH 3 + CH 3C6H 4OH ↔ C6H 5OCH 3 •CH 3C 6H 4OH ↔ [C 6H 5OCH 3 • H ] •CH 3C6H 4O − +
9.2.5 ELECTROLYTIC DISSOCIATION (IONIC ASSOCIATION) In solvents having high permittivity, the ionic associate decomposes into ions (mostly associated) to a variable degree: K pq+Lpq− ↔ K pq+−1Lpq−−1 + K q+ + Lp − ↔L ↔ pK q+ + qLp −
[9.7]
Electrolyte solution is formed according to eq. [9.7]. Conductivity is a distinctive feature of any material. The practice of application of individual and mixed solvents refers to electrolyte solutions, having conductivities > 10-3Cm/m. Binary systems, such as carbonic acids-amines, cresol-amines, DMSO-carbonic acids, hexamethyl phosphorous triamide acids, and all the systems acids-water, are the examples of such solvents. The common scheme of equilibrium in binary solvents, where components interaction proceeds until ions formation, may be represented by the common scheme: [9.8]
where: Km, Kn Kadd Ki Ka
the constants of homomolecular association; respectively the constant of the process formation of heteromolecular adduct the ionization constant the association constant.
The system acetic acid-pyridine4 may serve as an example of binary solvent whose equilibrium constants of all stages of the scheme [9.8] have been estimated. 9.2.6 REACTIONS OF COMPOSITION As a result of interaction between components of mixed solvents, profound rearrangement of bonds takes place. The process is determined by such high equilibrium constant that it is possible to consider the process as practically irreversible: mA + nB → pC
[9.9]
where: m, n, p C
stoichiometric coefficients compound formed from A and B
Carbonic acid anhydrides-water (e.g., (CH3CO)2O + H2O → 2CH3COOH), systems isothiocyanates-amines (e.g., C3H5NCS + C2H5NH2 → C2H5NHCNSC3H5) are examples of such binary solvents. It is pertinent that if solvents are mixed at stoichiometric ratio (m/n), they form individual liquid.
9.2 Chemical interaction between components
509
9.2.7 EXCHANGE INTERACTION Exchange interaction mA + nB ↔ pC + qD
[9.10]
is not rare in the practice of application of mixed solvents. Carbonic acid-alcohol is an example of the etherification reaction. Carbonic acid anhydride-amine exemplifies acylation process (CH3CO)2O + C6H5NH2 ↔ C6H5NHCOCH3 + CH3COOH; carbonic acid-another anhydride participate in acylic exchange: RCOOH + (R1CO)2O ↔ RCOOCOR1 + R1COOH. The study of the exchange interaction mechanism gives forcible argument to believe that in most cases the process goes through formation of intermediate - “associate”: mA + nB ↔ [Am B n ] ↔ pC + qD
[9.11]
9.2.8 AMPHOTERISM OF MIXED SOLVENT COMPONENTS The monograph of one of the authors of this chapter7 contains a detailed bibliography of works related to this paragraph. 9.2.8.1 Amphoterism of hydrogen acids In the first decades of the 20th century, Hanch ascertained that compounds which are typical acids show distinctly manifested amphoterism in water solutions while they interact with one another in the absence of solvent (water in this case). Thus, in binary solvents H2SO4-CH3COOH and HClO4-CH3COOH-acetic acid accepts proton forming acylonium, CH3COOH2+, cation. At present time, many systems formed by two hydrogen acids are being studied. In respect to one of the strongest mineral acids, CF3SO3H, all mineral acids are bases. Thus, trifluoromethane sulfuric acid imposes its proton even on sulfuric acid, forcing it to act as a base:8 CF3 SO 3 H + H2 SO 4 ↔ CF3 SO 3− • H3 SO 4+ Sulphuric acid, with the exception of the mentioned case, and in the mixtures with perchloric acid (no proton transfer), shows acidic function towards all other hydrogen acids. The well-known behavior of acidic function of absolute nitric acid towards absolute sulphuric acid is a circumstance widely used for the nitration of organic compounds. Trifluoroacetic acid reveals proton-acceptor function towards all strong mineral acids, but, at the same time, it acts as an acid towards acetic and monochloroacetic acids. In mixed solvents formed by two aliphatic carbonic acids, proton transfer does not take place and interaction is usually limited to a mixed associate formation, according to the equilibrium [9.2]. 9.2.8.2 Amphoterism of L-acids In binary liquid mixtures of the so-called aprotic acids (Lewis acids or L-acids) formed by halogenides of metals of III-IV groups of the periodic system, there are often cases when one of the components is an anion-donor (base) and the second component is an anion-acceptor (acid): M I X m + M II X n ↔ M I X m+ −1 + M II X n−+1
[9.12]
510
Y. Y. Fialkov, V. L. Chumak
The following is an example of eq. [9.12]: In the mixture AsCl3 - SnCl4, stannous tetrachloride is a base and arsenic trichloride is an acid:
(
2AsCl 3 + SnCl 4 ↔ AsCl 2+
)
2
• SnCl 6−2
9.2.8.3 Amphoterism in systems H-acid-L-acid Amphoterism phenomenon is the mechanism of acid-base interaction in systems formed by H-acids or two L-acids in binary systems. For example, in stannous tetrachloridecarboxylic acid system, acid-base interaction occurs with SnCl4 being an acid: SnCl 4 + 2RCOOH ↔ [SnCl 4 (RCOO) 2 ]H2 The product of this interaction is such a strong hydrogen acid that it is neutralized by the excess of RCOOH:
[SnCl (RCOO) ]H 4
2
2
+ RCOOH ↔ [SnCl 4 (RCOO) 2 ]H − + RCOOH2+
The mechanism of acid-base interaction in binary solvents of the mentioned type depends often on component relation. In systems such as stannous pentachloride-acetic acid, diluted by SbCl5 solutions, interaction proceeds according to the scheme: 2SbCl 5 + 4HAc ↔ SbCl 3 Ac 2 + SbCl 6− + Cl − + 2H2 Ac + In solutions of moderate concentrations, the following scheme is correct: 2SbCl 5 + 2HAc ↔ SbCl 4 Ac + Cl − + H2 Ac + 9.2.8.4 Amphoterism in binary solutions amine-amine If amines differ substantially in their energies of proton affinity, one of them has a proton-donor function in binary mixtures of these amines, so that it acts as an acid: | | | | | | − + − + NH + N − ↔ N • H N ↔ N + H N − | | | | | |
[9.13]
In the overwhelming majority of cases, this acid-base interaction limits itself only to the first stage - formation of heteromolecular adduct. Interactions occurring in all stages of the scheme [9.13] are typical of binary solvents such as triethylene amine-pyridine or N-diethylaniline. Proton-donor function towards the amine component reveals itself distinctly in the case of diphenylamine. To conclude the section discussing amphoterism in binary solvents, we have constructed a series of chemical equilibria that help to visualize the meaning of amphoterism as a common property of chemical compounds:
9.3 Physical properties of mixed solvents
base
511
acid
Et3N
+
PhNH2
↔
Et3N⋅ PhNH2
PhNH2
+
Ph2NH
↔
PhNH2 ⋅ Ph2NH
Ph2NH
+
H2O
↔
Ph2NH⋅ H2O
H2O
+
HAc
↔L ↔
H3O + Ac
HAc
+
CF3COOH
↔L ↔
H2Ac + CF3COO
CF3COOH
+
HNO3
↔
CF3COOH⋅ HNO3
HNO3
+
H2SO4
↔
H2NO+3 + HSO-4 *
H2SO4
+
CF3SO3H
↔
H3SO+4 + CF3SO-3
+
-
+
-
*The first stage of absolute sulfuric and nitric acids
9.3 PHYSICAL PROPERTIES OF MIXED SOLVENTS 9.3.1 THE METHODS OF EXPRESSION OF MIXED SOLVENT COMPOSITIONS In the practice of study and application of mixed solvents the fractional (percentage) method of composition expression, according to which the fraction, Y, or percentage, Y*100, of i-component in mixed solvent are equal, is the most common: Yi = c i /
i =i
∑c i =1
Yi = c i /
[9.14]
i
i =i
∑c i =1
i
× 100%
[9.14a]
where: c
a method of composition expression.
The volume fraction, V, of a given component in the mixture equals to: Vi = v i / where: v
i =i
∑v i =1
[9.15]
i
initial volume of component
Molar fraction, X, equals X = mi /
i =i
∑m i =1
i
[9.16]
512
Y. Y. Fialkov, V. L. Chumak
where: m
the number of component moles.
Mass fraction, P, equals to: P = gi /
i =i
∑g i =1
[9.17]
i
where: g
mass of a component.
It is evident that for a solvent mixture, whose components do not interact:
∑V = ∑ X = ∑ P = 1
[9.18]
and expressed in percent
∑V
100
= ∑ X 100 = ∑ P100 = 100
[9.19]
Fractions (percentage) [9.19] of mixed solvents, since they are calculated for initial quantities (volumes) of components, are called analytical. In the overwhelming majority of cases, analytical fractions (percentages) are used in the practice of mixed solvents application. However, in mixed solvents A-B, whose components interact with formation of adducts, AB, AB2, it is possible to use veritable fractions to express their composition. For example, if the equilibrium A+B ↔ AB establishes, veritable mole fraction of A component equals to: NA = m A / (m A + m B + m AB )
[9.20]
where: mi
number of moles of equilibrium participants
In the practice of mixed solvents application, though more seldom, molar, cM, and molal, cm, concentrations are used. Correlations between these various methods of concentration expression are shown in Table 9.1.
Function
Table 9.1. Correlation between various methods of expression of binary solvent A-B concentration (everywhere - concentration of component A) Argument
x
V VθB / (θA − σ θV )
x V xθA / (σ θx + θB ) P xM A / (σ M x + M B ) cM
103ρx / (σ M x + M B )
cm 103 x / M B (1 − x)
cm cm M B / (103 + cm M B )
Pρ B / (ρ A − σ p P)
cm MB θA / (103 θB + cm MB θA )
Vρ A / (σ pV + ρ B ) 103V / θA
cM
P
PM B / ( M A − σ M P) cM θB / (103ρ − cM σ M) cM θA / 103 cM M A / 10 ρ 3
103ρcm / (103 + cm M A)
103ρP / M A
103 θBV / M B θA (1 − V ) 103 P / M A (1 − P)
cm M A / (103 + cm M A)
103 cM / (103ρ − M AcM)
σ is the difference between the corresponding properties of components, i.e., σy = yA - yB, ρ is density, and θ is molar volume.
9.3 Physical properties of mixed solvents
513
9.3.1.1 Permittivity In liquid systems with chemically non-interacting components, molar polarization, PM, according to definition, is additive value; expressing composition in molar fractions: i =i
PM = ∑ PM i x i
[9.21]
i =1
where: xi
molar fraction of corresponding mixture component
In the case of binary solvent system: PM = PM1 x + PM 2 (1 − x) = (PM1 − PM 2 )x + PM 2
[9.21a]
where: x
molar fraction of the first component.
Here and further definitions indices without digits relate to additive values of mixture properties and symbols with digits to properties of component solvents. As PM = f (ε)M / ρ = f (ε)θ i
f (ε)θ = ∑ f (ε i )θ i x i
then
[9.22]
1
or for binary solvent f (ε)θ = f (ε1 )θ1 x + f (ε 2 )θ 2 (1 − x)
[9.22a]
where: M ρ θ f(ε)
an additive molecular mass density molar volume one of the permittivity functions, e.g., 1/ε or (ε - 1)/(ε + 2)
In accordance with the rules of transfer of additive properties from one way of composition expression to another3 based on [9.22], it follows that i
f (ε) = ∑ f (ε i )Vi
[9.23]
1
and for a binary system f (ε) = f (ε1 )V + f (ε 2 )(1 − V)
[9.23a]
where: V
the volume fraction.
Substituting f(ε) function into [9.23], we obtain volume-fractional additivity of permittivity
514
Y. Y. Fialkov, V. L. Chumak i
ε = ∑ ε iVi
[9.24]
1
or for binary solvent ε = ε1V + ε 2 (1 − V)
[9.24a]
Vast experimental material on binary liquid systems1,2,3 shows that the equation [9.24] gives satisfactory accuracy for systems formed of components non-associated in individual states. For example, dielectric permittivity describes exactly systems n-hexane-pyridine, chlorobenzene-pyridine by equation [9.24a]. For systems with (one or some) associated components, it is necessary to consider fluctuations, which depend on concentration, density, orientation, etc. Consideration of ε fluctuations shows that calculation of dielectric permittivity for mixture of two chemically non-interacting liquids requires ∆ε-values determined from equation [9.24a] and substituted to equation: ∆ε = (ε1 − ε1 ) (∆ ρ) / [ε(2 + δ ε )] 2
[9.25]
where: ∆ρ δε ε1, ε2 εAdd εexp
fluctuation of density = (∂εAdd/∂V)/(∂εexp/∂V) dielectric permitivities of individual components additive value experimental value
The ∆ε value may be calculated from a simplified equation. For equal volume mixtures (V=0.5) of two solvents: ∆εV = 0. 5 = 0.043(ε1 − ε 2 ) / εV = 0. 5 2
[9.25a]
Thus, permittivity of equivolume mixture of two chemically non-interacting solvents can be calculated from the equation εV = 0. 5 = 0.5(ε1 − ε 2 ) + ε 2 − ∆εV = 0. 5
[9.26]
Having ε values of both initial components and the ε value for the mixture, it is not difficult to interpolate these values for mixtures for any given composition. For systems formed of two associated liquids, there are no reliable methods of ε isotherms calculation. An empirical method has been proposed, which is based on the assumption that each associated component mixed with the other associated component that does not interact chemically with the first one introduces the rigorously defined contribution into ε deviation from the volume-fractional additivity. The extent of these deviations for the first representatives of the series of aliphatic carbonic acids and alcohols of normal structure, as well as for some phenols, are given elsewhere.3 Permittivity of mixtures at different temperatures can be calculated with high accuracy from absolute temperature coefficients of permittivity and their values using equations identical in form to [9.24]- [9.26]. For example, the calculation of temperature coefficients
9.3 Physical properties of mixed solvents
515
of permittivity, α ε , of the equivolume mixture of two chemically non-interacting associated solvents is carried out using the equation identical in its form to [9.26]. ∆α ε,V = 0. 5 = 0.5( α ε1 − α ε 2 ) + α ε 2 − ∆α ε,V = 0. 5
[9.27]
where ∆α ε,V = 0. 5 = 0.043( α ε1 − α ε 2 ) / α ε,V = 0. 5
[9.28]
9.3.1.2 Viscosity There are a great number of equations available in literature intended to describe viscosity of mixtures of chemically non-interacting components. All these equations, regardless of whether they have been derived theoretically or established empirically, are divided into two basic groups. The first group includes equations relating mixture viscosity, η, to viscosity of initial components, ηi, and their content in mixture, c, (c in this case is an arbitrary method of concentration expression): η = f (η1, η 2 ,Lη i ; c 1, c 2 ,Lc i −1 )
[9.29]
The equations of the second group include various constants k1, k2, etc., found from experiment or calculated theoretically η = f (η1, η 2 ,Lη i ; c 1, c 2 ,Lc i −1 ; k1, k 2 ,L)
[9.30]
The accuracy of calculation of solvents mixture viscosity from [9.30] type of equations is not higher than from equations of [9.29] type. We, thus, limit discussion to equations [9.29]. The comparison of equations of type [9.29] for binary liquid systems3 has shown that, in most cases, viscosity of systems with chemically non-interacting components is described by the exponential function of molar-fractional composition i
η = ∏ η xi i
[9.31]
1
or for binary solvent: η = η1x η12− x
[9.31a]
where: xi
molar fractions of components of the binary system
Empirical expressions for calculation of viscosity of the system of a given type, which permit more precise calculations, are given in monograph.3 From [9.31], it follows that the relative temperature coefficient of viscosity β η = ∂η / η∂T = ∂lnη / ∂T for mixtures of chemically non-interacting liquids is described by equation: β η = xβ η1 + (1 − x)β η2
[9.32]
516
Y. Y. Fialkov, V. L. Chumak
9.3.1.3 Density, molar volume According to the definition, a density, ρ, of system is the volume-additive function of composition: i
ρ = ∑ ρ iVi
[9.33]
1
or for binary system: ρ = ρ 1V1 + ρ 2 (1 − V1 )
[9.33a]
From [9.33], it follows that molar volume of ideal system is a molar-additive function of composition: i
θ = ∑ θi x i
[9.34]
1
Molar mass of mixture, according to the definition, is a molar-additive function too: i
M = ∑ Mi x i
[9.35]
1
For two-component mixture of solvents, these relations are as follows: θ = θ1 x + θ 2 (1 − x)
[9.36]
M = M1 x + M 2 (1 − x)
[9.37]
Non-ideality of the system leads to relative deviations of density and molar volume, as well as of other volumetric properties, having magnitude of about 1%. Here, contraction and expansion of solutions are approximately equiprobable. In systems in conditions close to separation, considerable contraction is possible. However, two component systems formed from the most common solvents, used in practical applications, form solutions far from separation at usual temperature range. Taking into account the linear density dependence on temperature in equations [9.33] and [9.34], it is possible to calculate values of density and molar volume at any given temperature. 9.3.1.4 Electrical conductivity High conductivity of solvent mixture formed from chemically non-interacting components may be related to the properties of only one or both components. Very high conductivity of mineral acids, carboxylic acids, some complexes of acids with amines, stannous chloride, and some tetraalkylammonium salts increases conductivity of their mixtures with other solvents.3 Normal practice does not often deal with mixed solvents of such high conductivity. Therefore, the theory of concentration dependence of conductivity of binary solvents is briefly discussed here.3,10
9.3 Physical properties of mixed solvents
517
Two basic factors which influence conductivity of binary solvent mixture are viscosity and permittivity. The influence of these factors on specific conductivity is quantitatively considered in empirical equation: κ = 1 / η exp[ln κ 1η1 − L(1 / ε − 1 / ε1 )]
[9.38]
where: κ
conductivity (symbols without index belong to the mixture properties, index 1 belongs to electrolyte component) = aε1ε 2 / (ε1 − ε2 ) proportionality coefficient
L a
and in non-empirically equation: ln κ = const 1 + 1 / ε(const 2 + const 3 + const 4 ) + 1 / 2 ln c − ln η
[9.39]
where: c consts
concentration, mol l-1 calculated from the equations11 based on crystallographic radii of the ions, dipole moments of molecules of mixed solution, and ε of mixture components
From equations [9.38] and [9.39], it follows that isotherms of the logarithm of conductivity corrected for viscosity and concentration lnκηc-1/2 have to be a linear function of reciprocal ε. Validity of this assumption is confirmed in Figure 9.1. Control of properties of multi-component mixture of solvents is achieved by methods of optimization based on analytical dependencies of their properties in composition:
(
)
y I = f y 1I , y 2I ,Lc 1, c 2 ,Lc i −1 ,
(
y II = f y 1II , y 2II ,Lc 1, c 2 Lc i −1
)
[9.40]
where: yI, yII y 1I , y II2
any property of solvent initial property of solvent
Concentration dependencies of permittivity, viscosity, density (molar volume) and conductivity described here permit to select with certainty the composition of mixed solvent, characterized by any value of mentioned properties. 9.3.2 PHYSICAL CHARACTERISTICS OF THE MIXED SOLVENTS WITH CHEMICAL INTERACTION BETWEEN COMPONENTS In this section, we discuss physical characteristic change due to the changes of binary mixed solvent composition in the systems with chemical interactions between components (these systems are the most commonly used in science and technology). Comprehensive discussion of the physical and chemical analysis of such systems, including Figure 9.1. Dependencies lnκηc-1/2 on 1/ε at 393.15K stoichiometry and stability constants, deterfor systems containing tetrabutylammoniumbromide mination of the formation of heteromolecular and aprotic solvent (1-nitromethane, 2-acetonitrile, 3-pyridine, 4-chlorobenzene, 5-benzene).
518
Y. Y. Fialkov, V. L. Chumak
Figure 9.2. Classification of permittivity isotherms for binary solvents: 1 and 2 - isotherms for systems with interaction and without interaction, respectively.
associates can be found in a specialized monograph.3 In theory, if composition of the mixture is known, then calculation of any characteristic property of the multi-component mixture can be performed by the classical physical and chemical analysis methods.3 For binary liquid system, it is possible only if all chemical forms (including all possible associates), their stoichiometry, stability constants, and their individual physical and chemical properties are well determined. A large volume of correct quantitative thermodynamic data required for these calculations is not available. Due to these obstacles, data on permittivity, viscosity and other macro-properties of mixed solvents with interacting components are obtained by empirical means. Data on empirical physical properties of liquid systems can be found in published handbooks.1,2,10 Principles of characteristic changes due to the compositional change of liquid mixtures with interacting components are discussed here. Assessment of nature of such interactions can only be made after evaluation of the equilibrium constant (energy) of such interactions between solvents. 9.3.2.1 Permittivity Permittivity, ε, is the only property of the mixed liquid systems with chemical interaction between components that has not been studied as extensively as for systems without chemical interaction.10 When interaction between components is similar to a given in equation [9.8], formation of conductive solutions occurs. Determination of ε for these solutions is difficult, and sometimes impossible. Although, more or less successful attempts to develop such methods of permittivity determination are published from time to time, the problem until now has not been solved. Classification of the isotherms ε vs. composition of liquid systems3 is based on deviations of the experimental isotherms, which are then compared with isotherms of the system
9.3 Physical properties of mixed solvents
519
without chemical interaction between components. The latter isotherms are calculated from equations [9.24] - [9.26], (Figure 9.2). Type I isotherm corresponds to the experimental isotherm with isotherms [9.24] or [9.26] of the liquid systems without interaction. The additive ε increase is compensated by ε decrease, due to homomolecular association process. The system diethyl ether - m-cresol illustrates this type of isotherm. Type II is determined by the negative deviations from isotherm [9.26]. This type corresponds to systems with weak heteromolecular interactions between components, but with strong homomolecular association of one of the components. The system formic acidanisole is an example of this kind of isotherm. Furthermore, these isotherms are characteristic when non-associating components in pure state form heteromolecular associates with lower dipole moment, DM, then DM of both components. The average DM for such kind of interaction in mixed systems is lower than correspondent additive value for non-interacting system. The system 1,2-dichloroethane - n-butylbromide can be referenced as an example of this kind of mixed binary solvent. Type III combines ε isotherms, which are above the isotherm calculated from equation [9.26]. This kind of isotherm suggests interaction between components of mixed solvent. The variety of such systems allows us to distinguish between five isotherm subtypes. Subtype IIIa is represented by isotherms lying between the curve obtained from eq. [9.26] and the additive line. This isotherm subtype occurs in systems with low value of heteromolecular association constant. The system n-butyric acid - water can be mentioned as a typical example of the subtype. Subtype IIIb combines S-shaped ε isotherms. This shape is the result of coexistence of homo- and heteromolecular association processes. System pyridine-water is a typical example of this subtype. Subtype IIIc combines most common case of ε isotherms - curves monotonically convex from the composition axis (∂ε / ∂V ≠ 0, ∂ 2 ε / ∂V2 < 0). The system water-glycerol can serve as typical example of the subtype. Subtype IIId is represented by ε isotherms with a maximum. Typically, this kind of isotherm corresponds to systems with high heteromolecular association constant. Component interaction results in associates with greater values of DM than expected from individual components. Carboxylic acids-amines have this type of isotherms. Finally, subtype IIIe includes rare kind of ε isotherms with a singular maximum, which indicates equilibrium constant with high heteromolecular association. This kind of isotherm is represented by system SnCl4-ethyl acetate. 9.3.2.2 Viscosity Shapes of binary liquid system viscosity - composition isotherms vary significantly. The basic types of viscosity - composition isotherms for systems with interacting components are given in Figure 9.3. Type I - viscosity isotherms are monotonically convex in direction to the composition axis (∂η / ∂x ≠ 0, ∂ 2η / ∂x²>0). Chemical interaction influences the shape of viscosity isotherm that is typical when experimental isotherm is situated above the curve calculated under assumption of absence of any interaction (i.e., from equation [9.31a]). The means of increasing heteromolecular association and determination of stoichiometry of associates for this type of isotherms was discussed elsewhere.3 Piperidine-aniline system is an example of this kind of interacting system.
520
Y. Y. Fialkov, V. L. Chumak
Figure 9.3. Classification of viscosity isotherms for binary solvents.
Type II - S-shaped viscosity curves (∂η / ∂x ≠ 0, ∂ 2η / ∂x² = 0 at the inflection point). This type of isotherm is attributed to systems which have components differing substantially in viscosity, but have low yield of heteromolecular associates to form local maximum in isotherms. Systems such as sulfuric acid-pyrosulfuric acid and diphenylamine - pyridine are examples of this kind of viscosity isotherm. Type III combines viscosity isotherms, which are convex-shaped from composition axis as its common attribute, i.e., ∂ 2η / ∂x² < 0. In the whole concentration range, we have positive values of excessive viscosity ηE = η exp − [(η1 − η 2 )x+ η 2 , where ηexp is experimental value of viscosity. It shall be mentioned that, although the term ηE is used in literature, unlike the excessive logarithm of viscosity (lnη)E = lnηexp - (lnη1/η2 + lnη2), it has no physical meaning.Viscosity isotherms of this type for interacting liquid systems are most commonly encountered and have shape diversity, which can be divided into three subtypes. Subtype IIIa - isotherms monotonically convex shaped from composition axis, but without maximum (in the whole composition range ∂η / ∂x ≠ 0). Dependence viscositycomposition of water-monochloroacetic acid can serve as an example. Subtype IIIb - isotherms with “irrational” maximum (i.e., with maximum which does not correspond to any rational stoichiometric correlation of components in mixed solvent). This subtype is the most common case of binary mixed solvents with interacting compo-
9.3 Physical properties of mixed solvents
521
nents. System trifluoroacetic acid-acetic acid is common example. The true stoichiometry determination method for this case was described elsewhere.3 Subtype IIIc - isotherms with maximum at rational stoichiometry, which corresponds to composition of heteromolecular associate. This case can be exhibited by the system pyrosulfuric acid-monochloracetic acid. Occasionally, isotherms with singular maximum can also occur. This behavior is characteristic of the system mustard oil-amine. This classification covers all basic types of viscosity isotherms for binary mixed systems. Although the classification is based on geometrical properties of isotherm, heteromolecular associations determine specific isotherm shape and its extent. The relative level of interactions in binary mixed systems increases from systems with isotherms of type I to systems with type III isotherms. 9.3.2.3 Density, molar volume Unlike the dependencies of density deviation (from additive values) on system composition, correspondent dependencies of specific (molar) volume deviation from the additive values: ∆θ = θ exp − [(θ1 − θ 2 )x + θ 2 ]
[9.41]
where: ∆θ θexp θi x
departure from of molar volume experimental value of molar volume molar volumes of components molar fraction of the second component
allow us to make a meaningful assessment of the heteromolecular association stoichiometry.3 Classification of volume-dependent properties of binary mixed solvents are based on dependencies of ∆θ value (often called “excessive molar volume”, θE). It is evident both from “ideal system” definition and from equation [9.36], that for non-interacting systems ∆θ ≈ 0. Interaction is accompanied by formation of the heteromolecular associates. It can be demonstrated by analysis of volumetric equations for the liquid mixed systems, data on volume compression, i.e., positive density deviation from additivity rule, and hence negative deviations of experimental specific molar volume from partial molar volume additivity rule. There are three basic geometrical types of ∆θ isotherms for binary liquid systems with chemically interacting components12 (Figure 9.4.). Type I - rational ∆θ isotherms indicating that only one association product is formed in the system (or that stability constants for other complexes in system are substantially lower than for main adduct). Most often, adduct of equimolecular composition is formed, and the maximum on ∆θ isotherm occurs at x = 0.5. Fluorosulfonic acid-trifluoroacetic acid system is an example of the type. Type II - irrational ∆θ isotherms with maximum not corresponding to the stoichiometry of the forming associate. This irrationality of ∆θ isotherms indicates formation of more than one adduct in the system. In the case of formation of two stable adducts in the system, the ∆θ maximum position is between two compositions which are stoichiometric for each associate. If in the system, two adducts AB and AB2 are formed,
522
Y. Y. Fialkov, V. L. Chumak
Figure 9.4. Classification of isotherms of molar volume deviation from additivity for binary solvents.
then ∆θ position of maximum is between xA values of 0.5 and 0.33 molar fractions. This type of ∆θ isotherm is most common for interacting binary liquid systems, for example, diethyl aniline-halogen acetic acids. Type III includes ∆θ isotherms with a singular maximum. According to metrical analysis,3 dependencies of this type are typical of systems where adduct formation constant, K f → ∞. Mustard oil-amine is an example of this behavior. Relative extent of the interaction increases from systems with ∆θ isotherms of type I to systems of type III isotherms. This conclusion comes from analysis of metrical molar volume equations and comparison of binary systems composed of one fixed component and a number of components, having reactivity towards fixed component changing in well-defined direction. 9.3.2.4 Conductivity The shape of specific conductivity relation on composition of the binary mixed solvent depends on conductivity of the components and some other parameters discussed below. Because of molar conductivity calculation of binary mixed solvent system, the ion associate concentration should be taken into account (eq. [9.7]). This quantity is known in very rare instances,13,14 therefore, the conductivity of interacting mixed solvent is most often expressed in specific conductivity terms. Classification of conductivity shapes for binary liquid systems may be based on the initial conductivity of components. Principal geometrical types of the conductivity isotherms are presented in Figure 9.5. The type I isotherm includes the concentration dependence of conductivity of the systems (Figure 9.5). These systems are not often encountered in research and technology practice. The type I can be subdivided into subtype Ia - isotherms with a minimum (for example, selenic acidorthophosphoric acid) and subtype Ib - isotherms with a maximum (for instance, orthophosphoric acid-nitric acid). The type II isotherms describe the concentration dependence of conductivity for binary systems containing only one conductive component. In this case also, two geometric subtypes may be distinguished. Subtype IIa - has monotonically convex shape (i.e., without extreme) in the direction of the composition axis. The binary system, selenic acid-acetic acid, can serve as an example of this kind of dependency. Subtype IIb includes isotherms with a maximum (minimum) - most widespread kind of isotherms (for example: perchloric
9.3 Physical properties of mixed solvents
523
Figure 9.5.Classification of specific conductivity isotherms for binary solvents.
acid-trifluoroacetic acid). The next frequently observed case is separated as subtype IIb-1; these are isotherms with χ=0 in the middle of concentration range. The system sulfuric acidethyl acetate illustrates the subtype IIb-1. Geometrically, the conductivity isotherms of I and II types are similar to the χ isotherms for binary liquid systems without interaction (Section 9.3.1.4). But conductivity dependencies corrected for viscosity vs. the concentration of interacted system differ from the corresponding dependencies for non-interacting systems by the presence of maximum. Also, interacting and non-interacting systems differ based on analysis of effect of the absolute and relative temperature conductivity coefficients on concentration.3 The relative temperature electric conductivity coefficient, β χ , differs from the electric conductivity activation energy, Eact, by a constant multiplier.15 Systems formed by non-electrolyte components are the most common types of electrolyte systems.
524
Y. Y. Fialkov, V. L. Chumak
Conductivity of the systems results from the interaction between components. This interaction proceeds according to the steps outlined in the scheme [9.8]. Conductivity of the mixed electrolyte solution confirms the interaction between components. Subtype IIIa isotherms are the most widespread. Such isotherms have one maximum and bring the conductivity to the origin for pure components. Subtype IIIb isotherms have a minimum situated between two maxima. The maximum appears because of the significant increase of the solution viscosity due to the heteromolecular association process. When the conductivity is corrected for viscosity, the maximum disappears. Conductivity of the mixed solvent pyrosulfuric acid-acetic acid is an example of the system. Subtype IIIb-1 isotherms with a curve is a special case of subtype IIIb isotherms. The curve is caused by viscosity influence. Isotherm IIIb turns to subtype IIIb-1 at higher temperatures. A concentration dependence of conductivity for stybium (III) chloride-methanol is an example of the systems. Determination of stoichiometry of interaction between the components of the mixed solvent is discussed elsewhere.3,12 9.3.3 CHEMICAL PROPERTIES OF MIXED SOLVENTS Solvating ability of mixed solvent differs from solvating ability of individual components. In addition to the permittivity change and the correspondent electrostatic interaction energy change, this is also caused by a number of reasons, the most important of which are discussed in the chapter. 9.3.3.1 Autoprotolysis constants Let both components of mixed solvent, AH-BH, to be capable of autoprotolysis process. 2HA ↔ H2 A+ + A −
[9.42a]
2BH ↔ H2 B + + B −
[9.42b]
As Aleksandrov demonstrated,15 the product of the activities of lionium ions sum (a H A+ + a H B+ ) on the liate activities sum (a A - + a B - ) is a constant value in the whole concen2 2 tration range for a chosen pair of cosolvents. This value is named as ion product of binary mixed solvent.
(
HA − HB K ap = aH
2A
+
+ aH
2B
+
)(a
A−
+ aB− )
[9.43]
If components of binary mixed solvent are chemically interacting, then the equilibrium of protolysis processes [9.42] is significantly shifting to one or another side. In most such instances, concentrations of lionium ions for one of components and liate ions for another become so low that correspondent activities can be neglected, that is autoprotolysis constant can be expressed as: HA − HB = aH K ap
2A
+
+ aB−
[9.44a]
HA − HB K ap = aH
2B
+
+ aA−
[9.44b]
9.3 Physical properties of mixed solvents
525
As an example, let us consider mixed solvent water-formic acid. Both components of the system are subject to autoprotolysis: 2H2O ↔ H3O+ + OH- and 2HCOOH ↔ HCOOH+2 + HCOOIt is obvious, that in the mixed solvent, concentrations of the HCOOH+2 ions (strongest acid from possible lionium ions in the system) and the OH- ions (strongest base from two possible liate ions) are neglected, because these two ions are mutually neutralized. Equilibrium constant of the direct reaction is very high: HCOOH+2 + OH- ↔ HCOOH + H2O Therefore, in accordance with [9.44a] autoprotolysis constant for this mixed solvent can be expressed as: HA − HB = aH K ap
3A
+
+ a HCOO −
Because of lionium and liate ions activities, both mixed solvent components depend on solvent composition, as does autoprotolysis constant value for mixed solvent. Figure 9.6 demonstrates dependencies of pKap values on composition for some mixed solvents. Using the dependencies, one can change the mixed solvent neutrality condition in a wide range. That is, for mixed solvent composed of DMSO and formic acid, neutrality condition corresponds to concentration of H+ (naturally solvated) of about 3×10-2 g-ion*l-1 at 80 mol% formic acid, but only 10-4 g-ion*l-1 at 10 mol% acid component. The method of autoprotolysis constant value determination based on electromotive force measurement in galvanic element comFigure 9.6. Dependence of pKap (-logKap) on the mixed solvent composition: 1-DMSO-formic acid; 2- posed from Pt,H2(1 atm)|KOH(m), KBr(m), solvent|AgBr,Ag was proposed elsewhere.15a DMSO-acetic acid; 3- formic acid-acetic acid. The method was used for polythermal study of autoprotolysis constant of binary mixed solvent 2-methoxyetanol - water. On the basis of ionization constants, polytherms data, and the autoprotolysis process, thermodynamic data were calculated. A large data set on the electrophility parameter ET (Reichardt parameter65) for binary water-non-water mixed solvents was compiled.66
526
Y. Y. Fialkov, V. L. Chumak
Figure 9.7. ET and Z parameters for some mixed solvents.
9.3.3.2 Solvating ability Applying the widely-used chemical thermodynamics concept of free energy linearity, it is easy to demonstrate that any characteristic y which linearly depends on free energy or on free activation energy of process in the mixed solvent composed of specifically non-interacting components, shall be linear function of the components with partial molar concentration x. i
y = ∑ yi xi
[9.44]
1
The formula was developed analytically by Palm.16 In the mixed solvents with chemically interacting components, the function y=f(x) can substantially deflect from the linear form including appearance of extrema. Selection of the mixed solvent components allows us in most cases to provide controlled solvation of all substances, participating in the chemical process performed in solu-
9.4 Mixed solvent influence on the chemical equilibrium
527
tion. Often, it can be achieved by combination of solvate active and solvate inert components. For example, it is obvious that in all compositional range of mixture DMSOCCl4 (except 100% CCl4) specific solvation of acid dissolved in this mixture is realized by DMSO. Similarly in mixed solvent formic acid-chlorobenzene, solvation of the dissolved donor substance is performed exclusively by formic acid. Selection of the second (indifferent) component’s ε also provides means to control the universal solvation ability of mixed solvent. In the above mentioned examples, increasing the solvate inert component concentration results in the decrease of mixed solvent ε . On the contrary, addition of indifferent component (propylene carbonate) into the systems such as acetic acid-propylene carbonate or propylene carbonate-aniline causes ε to rise. Because in the last two systems acetic acid and aniline were chosen as solvate active components, it was obviously intended to use these mixtures for specific solvation of the dissolved donor and acceptor compounds respectively. 9.3.3.3 Donor-acceptor properties As was demonstrated,17 the parameters ET and Z of the binary mixed solvent 1,2-dibromoethane - 1,2-dibromopropane are strictly additive function of molar composition. For the mixed solvents, having components engaged into specific interaction, dependencies ET=f(x) and Z=F(x) are non-linear and even extremal, as can be seen from the examples in Figure 9.7. The method was proposed18 to linearize polarity index of mixed solvent by introducing a parameter, which connects the ET and Z values with fractional concentrations of components.
9.4 MIXED SOLVENT INFLUENCE ON THE CHEMICAL EQUILIBRIUM 9.4.1 GENERAL CONSIDERATIONS7 The chemical process established in a solvent can be represented in a general form: E ↔F
[9.45]
where: E F
all chemical forms of reaction reagents all chemical forms of reaction products
Considering the traditional thermodynamic cycle, we can use the general equation of Gibbs’ energy variation because of the process [9.45]: ∆G = ∆Gsolv, E − ∆Gsolv, F − ∆G (v )
[9.46]
where: ∆Gsolv,E, ∆Gsolv,F ∆G(v)
Gibbs’ solvation energy of [9.45] equilibrium members Gibbs’ energy of process [9.45] in vacuum
The process [9.45] takes place in mixed solvent A-B. Solvation energies of equilibrium members are the algebraic sum of those for each of the mixed solvent components (this sum also takes into account the energy of mixed solvates such as EAxBy and FAzBt): ∆Gsolv, i = ∆Gsolv, A + ∆Gsolv, B = σsolv, i Therefore the equation [9.46] can be represented in the form
[9.47]
528
Y. Y. Fialkov, V. L. Chumak
∆G = σsolv, E − σsolv, F − ∆G (v )
[9.48]
The variation of free energy due to any chemical process consists of both covalent and electrostatic components: ∆G = ∆G (cov) − ∆G ( el )
[9.49]
Substituting [9.49] in [9.48], we come to an equation that in general describes the mixed solvent effect on the equilibrium chemical process:
(
(cov) (cov) ∆G = −∆G (v ) + σsolv − σsolv ,E ,F
) + (σ
( el ) solv, E
el ) − σs(olv ,F
)
[9.50]
For universal (chemically indifferent) solvents where σ (el) >> σ (cov) , it may by assumed that ( el ) ( el ) − σsolv ∆G = −∆G (v ) + σsolv ,E ,F
[9.51]
Thus the mixed solvent effect on the equilibrium of the chemical process [9.45] is determined not only by the vacuum component but also by the solvation energy of each of the chemical forms of equilibrium members. From free energy to equilibrium constants, one can obtain the equation describing mixed solvent effect on the equilibrium constant of the [9.45] process:
[
(
(cov) (cov) ln K = ∆G (v ) + σsolv − σsolv ,F ,E
) + (σ
( el ) solv, F
el ) − σs(olv ,E
)] / RT
[9.52]
In the special case of universal media, i.e., mixed solvent formed by both solvation-inert components, this equation can be presented as:
[
]
( el ) ( el ) ln K ( univ ) = ∆G (v ) + σsolv − σsolv / RT ,f ,E
[9.53]
Because the energy of all types of electrostatic interaction is inversely proportional to permittivity, these equations can be rewritten in the form:
[
(
(cov) (cov) ln K = ∆G (v ) + σsolv − σsolv ,F ,E
) + (β
solv, F
]
− βsolv, E ) / ε / RT
[9.52a]
and
[
]
ln K ( univ ) = ∆G (v ) + (βsolv, F − βsolv, E ) / ε / RT
[9.53a]
where: β
the multipliers of the magnitudes of reciprocal permittivity in equations of energy of the main types of electrostatic interactions such as dipole-dipole, ion-dipole, and ion-ion interactions
As follows from [9.52a], in binary mixed solvents formed from solvation-indifferent components (i.e., universal media), equilibrium constants of the [9.45] process are expo-
9.4 Mixed solvent influence on the chemical equilibrium
529
nent dependent on the 1/ε values (i.e., it is a linear correlation between lnK(univ) and 1/ε magnitudes). The vacuum component of energy of the [9.45] process can be obtained by the assumption of hypothetical media with ε → ∞ and 1/ε → 0: ) ∆G (v ) = RT ln K1(/univ ε→ 0
[9.54]
For binary solvents formed by solvation (active component A and indifferent component B) analysis of equation [9.52a] demonstrates that there is also a linear correlation between lnK and 1/ε. Such mixed solvents are proposed to be called as conventionally universal. To analyze the solvent effect on the process [9.45], it is often convenient to represent the temperature and permittivity dependencies of lnK in approximated form:
(
)
ln K = a 00 + a 01 / T + a 02 / T 2 +L+ a10 + a11 / T + a12 / T 2 +L / ε +
(
i =m j =n
)
+ a 20 + a 21 / T + a 22 / T 2 +L / ε 2 +L = ∑ i =0 j =o
[9.55]
a ij ε jT i
For the universal and conventionally-universal media, this dependence described by the equation can be represented in the form ln K ( univ ) = a 00 + a 01 / T + a10 / ε + a11 / εT
[9.56]
9.4.2 MIXED SOLVENT EFFECT ON THE POSITION OF EQUILIBRIUM OF HOMOMOLECULAR ASSOCIATION PROCESS All questions this part deals with are considered based on the example of a special and wide-studied type of homomolecular association process, namely, monomer-dimer equilibrium: 2E ↔ E 2
[9.57]
For this process, equation [9.52] can be presented in the form:
[
(
) ]
(cov) (cov) ln Kdim = ∆Gdim + ∆Gsolv / ε / RT , E 2 − 2 ∆Gsolv , E + β solv , E 2 − 2β solv , E
[9.58]
where: β µ
value at the 1/ε is calculated in accordance with the following equation ∆Geld - d = −120.6 µ 2 / rd3- d kJ / mol dipole moment
Because of the electrostatic component of process [9.57], the free energy is conditioned by dipole-dipole interactions such as dimer-solvent and monomer-solvent. In mixed solvent, CCl4-C6H5Cl, universal relation to acetic acid (because the mixed solvent components do not enter into specific solvation with the acid), the dimerization constant dependence on the temperature and permittivity in accordance with [9.52.a] and [9.56] is described by equation:19
530
Y. Y. Fialkov, V. L. Chumak
lnKdim = -1.72 + 1817.1/T + 0.92/ε + 1553.4/εT The dimer form concentration, cM (mol/l), is related to initial analytical concentration of dissolved compound and dimerization constant in equation:
(
)
c M = 4K dim c Mo + 1 − 8K dim c Mo + 1
1/ 2
/ 8K dim
[9.59]
One can calculate, with the help of this equation, that in hexane at c oM = 0.1, half of the dissolved acetic acid is in dimer form (Kdim ≈ 1.5×103). In nitrobenzene solution (Kdim ≈ 102), only a third of analytical concentration of acid is in dimer form. For phenol these values are 12% and 2%, respectively. Therefore the composition variation of universal mixed solvents, for instance, using hexane-nitrobenzene binary solvent, is an effective method to control the molecular composition of dissolved compounds able to undergo homomolecular association, in particular, through H-bonding. Typically, in specific solvents, the process of monomer formation [9.57] is characterized by considerably lower dimerization constants, as compared with those in universal media. In fact, acetic acid dimerization constant in water-dioxane binary solvent, the components of which are solvation-active in respect to the acid, vary in the 0.05-1.2 range. Replacing the solvent is often the only method to vary the molecular association state of dissolved compound. To achieve dimer concentration in 0.1 M solution of phenol in n-hexane equal to dimer concentration in nitrobenzene solution (50% at 25oC), it would be necessary to heat the solution to 480oC, but it is impossible under ordinary experimental conditions. 9.4.3 MIXED SOLVENT INFLUENCE ON THE CONFORMER EQUILIBRIUM Equilibrium took place in solutions Conformer I ↔ Conformer II
[9.60]
and the special case, such as Cis-isomer ↔ Trans-isomer
[9.61]
when investigated in detail. But the concentration-determining methods of equilibrium [9.60, 9.61] members are often not very precise. Dielcometry, or in our measurement the mean dipole moment, can be used only in the low-polar solvents. The NMR methods allowed us to obtain the equilibrium constants of processes by the measurement of chemical dislocation and constants of spin-spin interactions of nucleuses, which are quite precise but only at low temperatures (that is why the numerous data, on conformer and isomerization equilibrium, in comparison with other types of equilibria, are in solutions at low temperatures). It is evident that the accuracy of the experimental definition of equilibrium constants of the investigated process is insufficient. Thus, the summarization of experimental data on solvent effect on [9.60, 9.61] equilibrium encounters several difficulties. As a rule, the differences in conformer energies are not very high and change from one tenth to 10-12 kJ/mol. This is one order value with dipole-dipole interaction and specific solvation energies, even in low-active solvents. Moreover, the dipole moments of con-
9.4 Mixed solvent influence on the chemical equilibrium
531
former highly differ one from another, so ε is one of the main factors which affects equilibrium, such as in equations [9.60, 9.61]. If equations of processes are identical to that of scheme [9.45] in the form and maintenance, it is possible to apply the equations from Section 9.4.1 to calculate the equilibrium constants of the investigated process. Abrahem20 worked out details of the solvent effect theory on changing equilibrium. This theory also accounts for the quadruple interactions with media dipoles. It demonstrates that the accuracy of this theory equations is not better than that obtained from an ordinary electrostatic model. Because of low energy of [9.60, 9.61] processes, it is not possible to assign any solvents strictly to a universal or specific group. That is why only a limited number of binary solvents can be used for the analysis of universal media influence on conformer equilibrium constants. It is interesting to note that over the years, the real benzene basicity in liquid phase has not been sufficiently investigated. There were available conformer equilibrium constants in benzene, toluene, etc., highly differing from those in other low-polar media. Such phenomenon is proposed to be called the “benzene effect”.21 Sometimes rather than equilibrium constants, the differences in rotamer energies - for example, gosh- and trans-isomers - were calculated from the experiment. It is evident that these values are linearly proportional to the equilibrium constant logarithm. Different conformers or different intermediate states are characterized by highly distinguished values of dipole moments.22 Indeed, the media permittivity, ε, change highly influences the energy of dipole-dipole interaction. Therefore, according to [9.53a], it is expected that conformer transformation energies and energies of intermediate processes in universal solvents are inversely proportional to permittivity. But equilibrium constants of reactive processes are exponent dependent on1/ε value, i.e., there is a linear correlation between lnKconf and reciprocal permittivity: E conf ( turning ) = A + B / ε
[9.62]
ln K conf ( turning ) = a + b / ε
[9.63]
The analysis of experimental data on conformer and intermediate equilibrium in universal media demonstrates that they can be described by these equations with sufficient accuracy not worse than the accuracy of experiment. The differences in rotamer energies of 1-fluoro-2-chloroetane in mixed solvents such as alkane-chloroalkane can be described by equation:20 ∆E = Egosh - Etrans = -2.86 + 8.69/ε, kJ/mol According to equation [9.51], the permittivity increase leads to decreasing absolute value of electrostatic components of conformer transformations free energies in universal solvents. For instance, conformer transformation free energy of α-bromocyclohexanone in cyclohexane (ε=2) is 5.2 kJ/mol, but in acetonitrile (ε=36) it is -0.3 kJ/mol. Specific solvation is the effective factor which controls conformer stability. The correlation between equilibrium constants of the investigated processes [9.60] and [9.61] and polarity of solvation-inert solvents is very indefinite. For instance, diaxial conformer of 4-methoxycyclohexanone in acetone (as solvent) (ε=20.7) is sufficiently more stable than in
532
Y. Y. Fialkov, V. L. Chumak
methanol (ε=32.6). But the maximum stability of that isomer is reached in low-polar trifluoroacetic acid as solvent (ε=8.3). Formation of the internal-molecular H-bonding is the cause of the increasing conformer stability, often of gosh-type. Therefore solvents able to form sufficiently strong external H-bonds destroy the internal-molecular H-bonding. It leads to change of conformer occupation and also to decrease of the internal rotation barrier. Due to sufficiently high donority, the benzene has a comparatively high degree of conformer specific solvation (“benzene effect”). This leads to often stronger stabilization of conformer with higher dipole moment in benzene than in substantially more polar acetonitrile. In a number of cases, the well-fulfilled linear correlation between conformer transformation constants and the parameter ET of mixed solvent exists (see Section 9.3.4.3). It is often the solvent effect that is the only method of radical change of relative contents of different conformer forms. Thus, with the help of the isochore equation of chemical reaction, the data on equilibrium constants and enthalpies of dichloroacetaldehyde conformer transformation allow us to calculate that, to reach the equilibrium constant of axial rotamer formation in cyclohexane as solvent (it is equal to 0.79) to magnitude K=0.075 (as it is reached in DMSO as solvent), it is necessary to cool the cyclohexane solution to 64K (-209oC). At the same time, it is not possible because cyclohexane freezing point is +6.5oC. By analogy, to reach the “dimethylsulfoxide” constant to value of “cyclohexane”, DMSO solution must be heated to 435K (162oC). 9.4.4 SOLVENT EFFECT ON THE PROCESS OF HETEROMOLECULAR ASSOCIATION Solvent effect on the process of heteromolecular association mE + nF ↔ E m Fn
[9.64]
has been studied in detail.22,23 In spite of this, monographs are mainly devoted to individual solvents. In universal media formed by two solvate-inert solvents according to equation [9.53a], equilibrium constants of the process of heteromolecular adduct formation depend exponentially on reciprocal permittivity; thus ln K add = a + b / ε
[9.65]
The process of adduct formation of acetic acid (HAc) with tributyl phosphate (TBP) nHAc•TBP in the mixed solvent n-hexane-nitrobenzene7 can serve as an example of the [9.65] dependence validity. Equilibrium constants of the mentioned above adducts in individual and some binary mixed solvents are presented in Table 9.2. The concentration of adduct EF cM is related to constant Kadd and initial concentration of the components c oM by equation
(
)
c M = 2K add c Mo + 1 − 4K add c Mo + 1
1/ 2
/ 2K add
[9.66]
9.4 Mixed solvent influence on the chemical equilibrium
533
Table 9.2. Equilibrium constants of adduct (nHAc•TBP) formation in the mixed solvent n-hexane - nitrobenzene (298.15K) and coefficients of equation [9.65] Kadd
Solvent
HAc•TBP
2HAc•TBP
3HAc•TBP 5
4HAc•TBP 1.46×10-7
Hexane (ε=2.23)
327
8002
6.19×10
H+NB (ε=9.0)
28.1
87.5
275
1935
H+NB (ε=20.4)
19.6
45.0
88.1
518
Nitrobenzene (ε=34.8)
17.4
36.2
60.8
337
Coefficients of equation [9.65]: a
2.69
3.28
3.58
5.21
b
5.82
10.73
18.34
21.22
When stoichiometric coefficients are equal to m and n (equation [9.64]), Kadd expression is an equation of higher degree relative to cM
(
K E m Fn = c M / c Mo , E − nc M
) (c n
o M, F
− mc M
)
m
[9.67]
Let us consider an example of the interaction between acetic acid and tributyl phosphate (the change of permittivity of universal solvent permits us to change essentially the output of the reaction product). When initial concentration of the components equals to c oM =0.1 mol/l, the output (in %) of complexes of different composition is listed in table below: 1:1
2:1
3:1
4:1
In hexane
84
46
32
22
In nitrobenzene
43
13
4.5
2
Relative concentration changes the more essentially, the larger the value of electrostatic component of the process free energy, ∆Gelε=0 . Just as in the previous cases, the solvent use in this case is an effective means of process adjustment. To reduce the output of adduct HAc•TBF in nitrobenzene to the same value as in hexane solution, nitrobenzene solution must be cooled down to -78oC (taking into account that enthalpy of adduct formation equals 15 kJ/mol).24 Naturally, the process is not possible because the nitrobenzene melting point is +5.8oC. Let us consider the effect of specific solvation on equilibrium constant of the heteromolecular association process as an example of associate formation with a simplest stoichiometry: E + F ↔ EF
[9.68]
534
Y. Y. Fialkov, V. L. Chumak
Only when A is a solvate-active component in the mixed solvent A-B, in the general case, both A and B initial components undergo specific solvation: E + A ↔ EA
[9.69]
F + A ↔ FA
[9.70]
where: EA, FA
solvated molecules by the solvent (it is not necessary to take into account solvation number for further reasoning)
Since specific solvation of the adduct EF is negligible in comparison with specific solvation of initial components, A-B interaction in the solvent can be presented by the scheme: EA + FA ↔ EF + 2A
[9.71]
i.e., the heteromolecular association process in the specific medium is a resolvation process, since both initial components change their solvative surrounding. That is why equilibrium constants of the heteromolecular association process, calculated without consideration of this circumstance, belong indeed to the [9.71] process but not to the process [9.68]. Let us develop a quantitative relation between equilibrium constants for the process [9.68] K EF = [EF ] / [E ][F ]
[9.72]
and for the resolvation process [9.71] K us = [EF ] / [EA][FA]
[9.73]
Concentration of the solvate active solvent or the solvate active component A of mixed solvent in dilute solution is higher than the initial concentration of equilibrium components [E]0 and [F]0. Then activity of equilibrium components is equal to their concentrations. Then equations of material balance for components A and B can be set down as
[E ]0 = [E ] + [EF ] + [EA]
[9.74]
[F ]0 = [F ] + [EF ] + [FA]
[9.75]
Hence equilibrium constants of the process [9.69, 9.70] are presented by expressions K EA = [EA] / [E ] = [EA] / ([E ]0 − [EF ] − [EA])
[9.76]
K FA = [FA] / [F ] = [FA] / ([F ]0 − [EF ] − [FA])
[9.77]
Definition [9.72] is presented in the form K EF = [EF ] / ([E ]0 − [EF ] − [EA]) ([F ]0 − [EF ] − [FA]) and re-solvation constant
[9.78]
9.4 Mixed solvent influence on the chemical equilibrium
K us = [EF ] / ([E ]0 − [EF ]) ([F ]0 − [EF ])
535
[9.79]
Inserting [9.76] in [9.77] and [9.79] in [9.78], we come to the equation relating the equilibrium constants of all considered processes: K EF = K us (1 + E A )(1 + FA )
[9.80]
A similar form of equation [9.80] has been proposed elsewhere.25 If it is impossible to neglect the solvent concentration (or the concentration of solvate component of the mixed solvent) in comparison with [E]0 and[F]0, equation [9.80] is written in the form:
(
A K EF = K us 1 + K EA
)(1+ K ) S FA
[9.81]
Let E be acidic (acceptor) reagent in reaction [9.68], F basic (donor) reagent. Then, if A is acidic solvent, one can neglect specific solvation of the reagent E. Thus equations for equilibrium constant of the process [9.68] can be written in the form
(
A K EF = K us (1 + K FS ) K EF = K us 1 + K FS
)
[9.82]
If A is a basic component, equation [9.82] can be re-written to the form
(
A K EF = K us (1 + K EA ) K EF = K us 1 + K EA
)
[9.83]
Equations [9.82] and [9.83] were developed26-28 for the case of specific solvation. Thus for KEF calculation one must obtain the equilibrium constant of processes: [9.71] - Kus, [9.69] - KEA and [9.70] - KFS from conductance measurements. The constant KEF is identified in literature as “calculated by taking into account the specific solvation”.29 The value KEF characterizes only the universal solvation effect on the process of heteromolecular associate formation. The approach cited above can be illustrated by equilibrium: (CH3)2SO + o-CH3C6H4OH ↔ (CH3)2SO•CH3C6H4OH
[9.84]
studied in binary mixed solvents: A) CCl4 - hepthylchloride30 formed by two solvate inert components [9.84] B) CCl4 - nitromethane31 formed by solvate inert component (CCl4) and acceptor (nitromethane) component C) CCl4 - ethyl acetate32 formed by solvate inert component (CCl4) and donor (ester) component; The isotherms lnK vs. 1/ε (298.15K) are presented in Figure 9.8. These dependencies (right lines 1,2,4,5) are required for calculation of equilibrium constants of the heteromolecular association process free from specific solvation effect. It can be seen from Figure 9.8 that the values lnK, regardless of solvent nature, lie on the same line 3, which describes the change of equilibrium constants of the process [9.84] in the universal solution CCl4-heptylchloride.
536
Y. Y. Fialkov, V. L. Chumak
The data obtained for equilibrium [9.84] in binary mixed solvents of different physical and chemical nature are in need of some explanatory notes. First of all, if solvents are universal or conditionally universal ones, in accordance with general rules of equilibrium constants, the dependence on permittivity and the dependence of K= f(1/ε) in all cases is rectilinear. As appears from the above, specific solvation of one component decreases equilibrium constants in comparison with equilibrium constants in isodielectric universal solvent. The increase of KEF with permittivity increase in DMSO - o-cresol system required supplementary study.33 The method of study of specific solvation effect on the process of heteromolecular association has been described in work,34 devoted to study of the following interaction:
Figure 9.8. Dependence of equilibrium constant for the formation process of addition product DMSO to o-cresol on permittivity in solution based on heptylchloride (o), ethyl acetate (x) and nitromethane (•):1 - resolvation of o-cresol solvated in ethyl acetate in DMSO in mixed solvent CCl4-ethyl acetate; 2 resolvation of DMSO solvated in nitromethane in o-cresol in mixed solvent CCl4-nitromethane; 3 - process [9.84] in mixed solvent CCl4-heptylchloride; 4 solvation of o-cresol in ethyl acetate; 5 - solvation of DMSO in nitromethane.
C6H3(NO2)3 + C6H5N(CH3)2 ↔ C6H3(NO2)3•C6H5N(CH3)2
[9.85]
in the binary liquid solvent formed from a solvate-inert component (heptane) and one of the following solvate-active component: trifluoromethylbenzene, acetophenone or p-chlorotoluene. Therefore trinitrobenzene, TNB, is solvated in initial mixed solvents, so its interaction with donor represents the resolvation process. C6H3(NO2)3•A + C6H5N(CH3)2 ↔ C6H3(NO2)3•C6H5N(CH3)2 + (A) [9.85a] Equation [9.62] applied to process of the heteromolecular association is written in the form:
[
(
(cov) (cov) (cov) − σGsolv − σGsolv ln K = ∆G (v ) + σGsolv , EF ,E ,F
) + σβ
solv
]
/ ε / RT
[9.86]
But for equilibrium [9.85a] in mixed solvents one can assume σG(cov) solv,EF ≈ 0. That is why equation [9.52] can be converted for the process in conditionally universal media:
[
]
(cov) ln K us = ∆G (v ) − σGsolv + σβsolv / ε / RT ,E
[9.86a]
9.4 Mixed solvent influence on the chemical equilibrium
537
It follows from equation [9.86a] that dependence of lnKus vs. 1/ε will be rectilinear. Experimental data of equilibrium constants for process [9.85a] are in agreement with the conclusion. Approximation of data by equation [9.65] is presented in Table 9.3. It also follows from equation [9.86a] that RT, unlike in the cases of chemical equilibrium in universal solvents, describes, not the vacuum component of the process free energy, (el) but the remainder ∆G(v) − σG(cov) solv,TNB and σG solv,TNB . The values of the remainder as well as electrostatic component of the free energy for the described chemical equilibrium are summarized in Table 9.4. Although the solvate-active components are weak bases, their differences influence (el) the values ∆G(v) − σG(cov) solv,TNB and σG ε=1 . The first value characterizes the difference between the covalent components of solvation energy for trinitrobenzene in any solvate-active solvent. It follows from the table, that contributions of the covalent and electrostatic components are comparable in the whole concentration range of the mixed solvents. This indicates the same influence of both donor property and permittivity of the solvent on the [9.85a] process equilibrium. Table 9.3. Coefficients of equation [9.65] equilibrium constants of the [9.85a] process at 298.15K (r-correlation coefficient) Solvent
a
b
r
Heptane-trifluoromethylbenzene
-0.80
5.50
0.997
Heptane-acetophenone
-1.49
6.74
0.998
Heptane-chlorotoluene
-2.14
7.93
0.998
Table 9.4. Free energy components (kJ/mol) of resolvation process [9.85a] in mixed solvents at 298.15K ε
-∆G in system heptane - S, where S: trifluoromethylbenzene
acetophenone
chlorotoluene
2
6.8
8.3
9.8
4
3.4
4.1
4.9
6
2.3
2.8
3.3
1.7
2.1
-
-2.0
-3.7
-5.3
13.6
16.7
19.6
8 ∆G
(v)
σG
(el) ε=1
− σG
(cov) solv,TNB
According to the above energy characteristics of the heteromolecular association process (resolvation) in specific media, the solvent exchange affects the products’ output (the relationship of output cM and Kus is estimated from the equation [9.66]). This shows that the product output (with initial concentration of reagents 0.1M) can be changed from 34% (pure heptane) to 4 % (pure n-chlorotoluene) by changing the binary mixed solvent composition. The processes [9.85a] and [9.85] can be eliminated completely when the solvate active component (more basic then chlorotoluene) is used.
538
Y. Y. Fialkov, V. L. Chumak
Possibility of management of the products` output may by illustrated by the data from reaction equilibrium [9.84]. As follows from Figure 9.8, maximum equilibrium constant (in investigated solvents) is for pure chloroheptane; the adducts output in this solvent equals to 86% (at initial concentration of reagent 0.1 M). Minimum equilibrium constant of the process is in pure ethyl acetate, where the products` output equals to 18 % (at initial concentration of reagent 0.1 M). Thus the products` output of the reaction [9.84] can be changed directionally in the range 90 up to 20% by means of choice of corresponding binary or individual solvent. If selected solvent is more basic than ethyl acetate or more acidic than nitromethane, the process is not possible. Heteromolecular association process of o-nitrophenol and triethylamine can be an example of the management of products` output. Association constant of this process has been determined for some solvents.35 Use of hexane instead of 1,2-dichloroethane, DHLE, increases products’ output from 6% to 93%. When both components of the mixed solvent are solvate-active towards the reagents of equilibrium [9.64], the following interactions take place: E + A ↔ EA; F + A ↔ FA; E + B ↔ EB; F + B ↔ FB
[9.87]
The process of heteromolecular association [9.64] is due to displacement of the solvent components and formation of completely or partially desolvated adduct: xEA + yFA + (m − x )EB ↔ E m Fn + (A) + (B)
[9.88]
It is easy to develop the equation for binary solvent formed from two solvate-active components similar to [9.90] by using the above scheme for the binary solvent with one solvate active component (equations [9.76 - 9.80]) and introducing equilibrium constant of the process [9.88] such as Kus: K EF = K us (1 + K EA )(1 + K FA )(1 + K EB )(1 + K FB )
[9.89]
where: Ki
equilibrium constant of the processes [9.87]
9.4.4.1 Selective solvation. Resolvation When component B is added to solution E in solvent A (E is neutral molecule or ion), resolvation process takes place: EAP + qB = EB q + pA
[9.90]
Equilibrium constant of this process in ideal solution (in molar parts of the components) equals to: K us = m EBq m Ap / m EAp m Bq
[9.91]
where: mi p, q
a number of moles stoichiometric coefficients of reaction
On the other hand, resolvation constant Kus equals to the ratio of equilibrium constants of solvation processes E + pA ↔ EAp(I); E + qB ↔ EBq(II):
9.4 Mixed solvent influence on the chemical equilibrium
539
K us = K II / K I
[9.92]
The method of Kus determination, based on the differences between free energy values of electrolyte transfer from some standard solvent to A and B, respectively, leads to a high error. If the concentration ratio of different solvative forms is expressed as α, the concentrations A and B are expressed as 1-xB and xB, respectively, (xB - molar part B), equation [9.91] may be presented in the form: K us = αx Bp / (1 − x B )
q
[9.91a]
or ln K us = ln a + p ln x B − q ln(1 − x B )
[9.93]
The last equation permits to calculate value for the solvent of fixed composition and xB determination at certain composition of the solvate complexes p and q and resolvation constant Kus. It follows from the equations [9.91] and [9.91a]
[
)]
(
(
K us = m E( q ) q / m Eo − m E( q ) p x 0( q ) 1 − x B( q )
)
q
[9.91b]
where: m oE m (q) E
a number of moles E in solution a number of moles E solvated by the solvent A
Hence K us = x ′B x Bp /
[(1− x ′ )(1− x ) ] q
B
[9.91c]
B
where: x B′
molar fraction of B in solvate shell
Equation [9.91c] developed for the ideal solution E-A-B permits us to establish the relationship between the composition of mixed solvent xB and the solvate shell composition xB′ . For the special case of equimolar solvates, the expression of resolvation constant is written in the form: K us = αx B / (1 − x B )
[9.91d]
It follows that even in ideal solution of the simplest stoichiometry, α is not a linear function of the mixed solvent composition xB. Dependence of isotherm α on the solvent composition at Kus = 1 is presented in Figure 9.9a. Analytical correlation between the composition of solvate shell x′B and the mixed solvent composition can be developed:
[
K us = [x ′B / (1 − x ′B )] x Bp / (1 − x B )
q
]
[9.94]
540
Y. Y. Fialkov, V. L. Chumak
Figure 9.9. Characteristics of solvate shells in the mixed solvent A-B: a - dependence α = EA/EB on composition of the mixed solvent; b - dependence of the solvate shell composition, x ′B , on composition xB of the mixed solvent E-A-B.
Figure 9.10. Selective solvation of NaI in mixed solvent cyanomethane-methanol (a) and DMFA-methanol (b, curve1), DMFA - cyanomethane (b, curve 2): xB - composition of mixed solvent; x B′ - composition of solvate shell in molar parts of the second component.
The dependence for Kus = 1 is given in Figure 9.9b. As it follows from the analysis of equation [9.94] and Figure 9.9, the compositions of solvate shell and mixed solvent are different. A similar approach has been developed for calculation of solvate shell composition Na+ and I- in the mixed solvent formed by the components with close values of permittivity. Such solvent selection permits us to eliminate the permittivity effect on solvation equilibrium. Resolvation constants have been determined from the calorimetric study. The composition of anions solvate complex has been determined from experimental data of electrolyte Bu4NI assuming lack of the cation specific solvation. Experimental data are presented in Figure 9.10.
9.4 Mixed solvent influence on the chemical equilibrium
541
Padova36 has developed this approach to non-ideal solutions. He has proposed an equation based on electrostatic interaction which relates molar fractions of the components (xB - in the mixed solvent and xB′ - in the solvate shell) to the activity coefficient of components of the binary solvent: α = ln[(1 − x ′B ) / (1 − x B )] = ln γ B2
[9.95]
Strengthening or weakening interaction (ion-dipole interaction or dipole-dipole interaction) of universal solvation leads to re-distribution of molecules in the mixed solvate and to the change of the composition of solvate shell in contrast to the composition of mixed solvent. The method for determination of average filling of molecules` coordination sphere of dissolved substance by molecules of the mixed solvent (with one solvate-inert component) has been proposed.37 The local permittivity is related to average filling of molecules’ coordination sphere expressed by the equation: ε p = ε A x ′A + ε B x ′B
[9.96]
where: x ′A = z A / (z a + z B ); x ′B = z B / (z A + z B )
[9.97]
where: zA, zB
average numbers of A and B molecules in the first solvate shell
The last equations can be used for development of the next expression permitting to calculate the relative content of B molecules in the solvate shell x ′B = ( ε p − ε A ) / (ε B − ε A )
[9.98] where: εp εA, εB
Figure 9.11. Selective solvation of 3-aminophthalimid (A) by n-butanol (B) from the mixed solvent hexanen-butanol.
permittivity of binary solvent permittivities of components
Value xB′ can be found from the equation linking the location of maximum of absorption band of IR spectrum with refraction index and ε of the solution. The data on selective solvation of 3-aminophthalimid by butanol from butanol-hexane mixture are presented in Figure 9.11. The data have been calculated from the equations presented above. Alcohol content in solvate shell has higher concentration than in solvent composition even at low concentration of alcohol in the solvent. For example, when molar fraction of n-butanol in the mixture was 7%, the relative molar fraction of n-butanol in the solvate shell of
542
Y. Y. Fialkov, V. L. Chumak
Figure 9.12. Dependence of concentration (in molar parts) of solvate forms Li(EDA)mSn on molar fraction, xB, of the second component in the binary solvents ethylenediamine-DMSO (a) and ethylenediamine-DMFA (b): 1m=4, n=0;2- m=n=2;4-m=0, n=4.
aminophthalimide was 30%. Resolvation process completes at n-butanol concentration in solution ≈90%. Mishustin37 proposed a strict and accurate method for selective solvation study. The method is based on data of free energy transfer of electrolyte from individual solvent A to mixed solvent A - B. The method takes into account non-ideality of the system, and allows calculation of the concentration of different solvate forms and their dependence on the mixed solvent composition. An example of application of this method is in the work.39 Authors have calculated relative concentration of different solvate forms of Li+ in the mixed solvent ethylenediamine DMSO and ethylenediamine-DMFA (Figure 9.12). Free energy of lithium transfer from DMSO (DMFA) in the mixed solvent has been calculated from the time of spin-lattice relaxation of kernel 7Li. The curves presented in Figure 9.12 depict quantitatively the selectivity of Li+ relative to ethylenediamine, which is more basic component in contrast to the second components of the mixed solvent, namely DMSO and DMFA. The following systems can serve as examples of the effect of composition of the mixed solvent on the solvate shell composition: [Cr(NH)S(H2O)m(DMSO)n]3+ - H2O - DMSO40 [Be(H2O)m(EG)n]SO4 - H2O - ethylene glycol41 [Be(H2O)m(HMPTA)n]SO4 - H2O - HMPTA42 [Ni(H2O)mSn]ClO4 - H2O - S (where S is methanol, ethanol, propanol, DMSO)43 Data presented in Figure 9.13 contain information on the composition of solvate shell as a function of molar fraction of water in the mixed solvent H2O-other solvents.43 Monograph44 contains collection of data on resolvation constants of the ions in the mixed solvents. The above presented dependencies of the composition of solvate shell on the mixed solvent composition as well as resolvation constants permit calculation of the solvate composition by varying solvent composition. The dependence of resolvation constants on the permittivity of the solvent is discussed in the example of the proton resolvation process.
9.4 Mixed solvent influence on the chemical equilibrium
Figure 9.13. The composition of solvate shell Ni2+ (in molar parts of water) in the mixtures of solvents formed from water and propanol (1), ethanol (2), methanol (3) and DMSO (4). Data from Ref. 43.
[
HA+p + B ←→ HAp −1B
]
+
543
In the mixed solvents water - non-aqueous solvent, in spite of its donor and polar properties, water is a preferred solvating agent. This generalization has some exceptions (solvation in systems Ag+ - H2O acetonitrile, Cr3+ - H2O - DMSO, F-- H2O-ethylene glycol).45 Solvation energy of proton by donor solvents is very high. The regularities of the proton selective solvation and re-solvation processes were studied in more detail in comparison with other ions. Let us consider the changes in the system, when donor component is added to protic acid HA in solvent A. Anion solvation can be neglected, if both solvents have donor character. The interaction influences the proton re-solvation process.
+B ← →L ←→ HB p+ + A
[9.99]
or HA+p + qB ←→L ←→ HB +q + pA
[9.99a]
The model related to eq. [9.99a] was evaluated,46 resulting in the supposition that the equilibrium of two forms of solvated proton: HA +p and HB+q is important. Solvation stoichiometry was not considered. Both proton solvated forms are denoted as HA+ and HB+. If B is the better donor component (it is a necessary requirement for equilibrium [9.99] shift to the right hand side of equation), the equation from work46 can be simplified to the form:47
(
Ka = KaA + KaB − KaA
){[K
us
]}
][
x B / (1− x B ) / 1+ Kus x B / (1− x B )
[9.100]
where: Ka K Aa , K Ba Kus xB
ionic association constant the ionic association constants of acid in individual solvents A and B a constant of resolvation process molar fraction of B
For the calculation of resolvation constant, one must determine the experimental constant of HA association in the mixed solvent and determine independently K Aa and KBa . When the resolvation process is completed at low concentration of the second component, the change of permittivity of mixed solvent A-B may be ignored. Thus, one may assume that K Aa and KBa are constant and calculate Kus from the equation [9.100] in the form:
(
)
(
) {[(1− x ) / x ][1/ K (K
1/ Ka − KaA = 1/ KaB − KaA +
B
B
us
B a
− KaA
)]}
[9.101]
544
Y. Y. Fialkov, V. L. Chumak
Thus Kus is calculated as a slope ratio of coordinates: 1/( Ka - K Aa ) = (1-xB)/xB and (Ka K Aa ) remainder is obtained as Y-intercept. Dependence of eq. [9.101] is presented in Figure 9.14. Kus for proton was calculated in a series of mixed solvents. It was shown47 that, when pyridine, dimethylalanine or diphenylamine (resolvating agents with decreasing donor numbers) are added to the solution of trifluoroacetic acid in DMSO, proton decreases consecutively and its values are equal 4 3 Figure 9.14. Dependence of 1/(Ka - K Aa ) for CF3COOH to 2.7x10 ; 4.2x10 and 35.4, respectively. on the mixed solvent composition DMSO (A) Consideration of permittivity of the dimethylalanine at different temperatures. mixed solvent has allowed calculation of proton in whole concentration range of the mixed solvent DMSO - diphenylamine. The data have been approximated using equation [9.55]. ln Kus = -12.2 + 3400/T + 453.7/ε - 75043/εT Equilibrium constants for exchange process of alcohol shell of solvates to water shell were calculated:48 ROH+2 + H2 O ↔ H3 O + + ROH
[9.102]
The solutions of HCl and HOSO3CH3 in aliphatic alcohol (i.e., CnH2n+1OH) - normal alcohol C1-C5 and isomeric alcohol C3-C5 have been studied. If the components taking part in resolvation process are capable of H-bonding, the anion solvation by these components cannot be neglected. The differences in Kus values for both acids in different solvents may be explained as follows. The dependence of Kus on permittivity and temperate is described by the equation: ln K = a 00 + a 01 / T + a 02 / T 2 + (a10 + a11 / T ) / ε
[9.103]
The coefficients of equation [9.103] are presented in Table 9.5. Table 9.5. Coefficients of the equation [9.103] for constants of resolvation process System
a00
a01×10-2
a02×10-4
a10
a11×10-3
±δ
HCl - n-alcohol
12.4
46.0
56.3
186.8
71.1
0.3
HCl - isomeric alcohol
13.1
5.9
-36.9
318.4
94.7
0.2
HOSO3CH3 - n-alcohol
6.3
4.4
-5.4
177.7
60.9
0.3
The solvent effect in the following resolvation process was studied:49
9.4 Mixed solvent influence on the chemical equilibrium
545
ROH+2 + Py ↔ HPy + + ROH
[9.104]
The solutions of trifluoroacetic acid in ethanol and methanol in the temperature range 273.15-323.15K were investigated. The dependence of Kus on permittivity and temperature is described by the equation: ln Kus = 9.44 + 1768/T + 77.8/ε - 11076/εT Unlike the processes considered above, in the case of process [9.104] permittivity decrease leads to decreasing Kus. The explanation of the results based on covalent and electrostatic components of resolvation process enthalpy is given.50 The reaction [9.104] also has been studied for isodielectric mixtures of alcoholchlorobenzene with ε=20,2 (permittivity of pure n-propanol) and ε=17.1 (permittivity of pure n-butanol) to investigate the relative effect of universal and specific solvation on the resolvation process. The mixtures were prepared by adding chlorobenzene to methanol, ethanol, and C1-C3 alcohol. Alcohol is a solvate-active component in these isodielectric solvents. Kus data are given in Table 9.6. Table 9.6. Equilibrium constants of the process [9.104] in isodielectric solvents Solvents with ε=20.2
Kus×10-5
Solvents with ε=17.1
Kus×10-5
Methanol +23.5% chlorobenzene
3.9
Methanol + 32.8% chlorobenzene
1.4
Ethanol + 13% chlorobenzene
4.9
Ethanol + 24% chlorobenzene
2.0
n-Propanol
7.3
n-Butanol
4.9
Insignificant increase of Kus in n-butanol (or n-propanol) solution in comparison to methanol is due to relaxation of the proton-alcohol bond, when the distance of ion-dipole interaction increases. The change of donor property of the solvate-active component is not significant. The equations relating Kus to ε permit to divide free energy of resolvation process into the components. Corresponding data are presented in Table 9.7. Table 9.7. The components of free energy (kJ mol-1) of proton resolvation process at 298.15K σ∆Gel in process Solvent
ε
[9.102]
[9.104]
HCl
HOSO3CH3
CF3COOH
Methanol
32.6
3.9
2.0
-3.1
Ethanol
24.3
5.3
2.7
-4.1
n-Propanol
20.1
6.4
3.3
-5.0
n-Butanol
17.1
7.5
3.9
-5.9
n-Pentanol
14.4
8.9
4.6
-
546
Y. Y. Fialkov, V. L. Chumak
σ∆Gel in process Solvent
-δ∆G
ε
cov
−δ∆G
el ε=1
[9.102]
[9.104]
HCl
HOSO3CH3
CF3COOH
8.2
17.7
-38.1
128
66.0
-100.6
In contrast to the processes considered earlier, the vacuum electrostatic component of resolvation process has high value whereas −δ∆Gel values are comparable with the covalent component, δ∆Gcov. δ∆G values according to [9.51] are equal to: δ∆G = ( ∆G HB + + ∆G A − ∆G HA + + ∆G B )sol For small additions of B (to component A), (∆GA)sol=0 and (∆GHA)sol=0 , then δ∆G = (∆GHB+ + ∆GB)sol Solvation energy of complex HB+ by solvent A is small because coordination vacancies of the proton are saturated to a considerable extent. Therefore the interaction energy between A and B influences significantly the value of σ∆G. That is why, the mixed solvents (alcohol-water and alcohol-pyridine, for instance) are different because of the proton resolvation process. This can be explained in terms of higher energy of heteromolecular association for the alcohol-water in comparison with alcohol pyridine. The concept of solvent effect on the proton resolvation process was confirmed by quantum chemical calculations.51 Above phenomena determine the dependence of resolvation constant on physical and chemical properties. Let the resolvation process proceeds at substantial abundance of the component A in mixed solvent and initial concentrations HA (HA+) and B to be equal. The output of the process can be calculated from the equation similar to equation [9.66]. The large value of Kus in all considered processes of proton resolvation indicates the effect of permittivity change on the yield of complex HB+ formation. The output of resolvated proton in process [9.104] proceeding in methanol equals 100%, whereas in the same process in low polarity solvent (e.g., methanol-hexane), with abundance of the second component, the equilibrium is shifted to the left, resulting in solvate output of less than 0.1%. Kus values in single alcohol solvents are large, thus the output of reaction does not depend on solvent exchange. The process H•DMSO+ + B ↔ HB+ + DMSO may be considered as an example of the effect of chemical properties of B on the output of the reaction [9.99]. The output of HB+ equals to 98%, if pyridine is included in the process at initial concentration of 0.1M. Use of diphenylamine, having lower donor properties, decreases the output to 60%. The output differs even more at smaller concentrations of component, such as 10-3 M, which gives yields of 83 and 33%, respectively. 9.4.5 MIXED SOLVENT EFFECT ON THE ION ASSOCIATION PROCESS The ion association process (or opposite to it - ion dissociation process) has been studied in detail in comparison with other types of chemical equilibrium in solutions. The modern
9.4 Mixed solvent influence on the chemical equilibrium
547
state of the theory of individual solvent effects on equilibrium constant of ion association process (Ka) is described in monographs by Izmailov53 and Barthel.54 The formation of free (preferably solvated) ions is due to a successive equilibrium states proceeding in solution: K
K
K
add mE + nF ← → E mFn ←i →Cat pq + • An qp − ←→L ←a → pCat q + + qAn p−
[9.105]
where: Kadd Ki Ka
a constant of adduct formation ionization constant association constant
The true constant of ion association is a ratio of ionic associate concentration Cat qp + • An pq - to ion concentration product. If ionic associate concentration is unknown (as is true in many cases), ion association constant is calculated from the analytical concentration of dissolved substance:
{
[
] [
K a = c 0 − p Cat q+ − q An p −
]} / [Cat ] [An ] q+
p
p−
q
where: c0
initial concentration of electrolyte
The general theory of ionic equilibrium53,54 leads to the conclusion that ion association constant in universal or conditionally universal media, in accordance to the equation [9.53a], depends exponentially on reciprocal value of permittivity: ln K a = a 00 + a 01 / ε
[9.106]
where: a00, a01
approximation coefficients
Figure 9.15. Dependence Ka of (C2H5)4NBr on permittivity in mixed solvents formed by propylene carbonate with o-dichlorobenzene (1,2-o), pyridine (1,2-x) and acetic acid (3) at 298.15K.
If ion association process has high energy, the solvents are solvate-inert because of large ions, such as (CnH2n+1)4N+, (CnH2n+1)4P+, etc. Solution of (C2H5)4NBr in mixed solvents propylene carbonate - o-dichlorobenzene and propylene carbonate - pyridine55 serve as an example. The components of these binary solvents (with the exception of inert dichlorobenzene) possess donor properties, though they do not solvate the large anion R4N+ because it is solvate-inert component in relation to anion. That is why lnKa - 1/ε isotherms follow the same line (see Figure 9.15). But the lnKa - 1/ε isotherm for solution of propylene carbonate - acetic acid differs from the other two because acetic acid is a solvate-active component in relation to anion Br-.
548
Y. Y. Fialkov, V. L. Chumak
Coefficients of equation 9.56 for the various systems are presented in Table 9.8. Table 9.8. Coefficients of equation [9.56] for ionic association of tetraethyl ammonium bromide at 298-323K Solvent
-a00
a10
a01
a11
Propylene carbonate - o-dichlorobenzene
5.2
1440
78.8
9523
Propylene carbonate - pyridine
6.3
1850
114
0
Propylene carbonate - acetic acid
7.7
2400
110.5
-11500
The lnKa - 1/ε dependence for universal media is linear in full range of permittivity. Validity of the following equation was evaluated7 for the mixed solvent propylene carbonate - 1,4-dioxane in the range of permittivity ε=65±3 for the solutions Et4NBr lnKa = 1.32 + 80.05/ε Comparison of the free energy of
Table 9.9. Components of free energy (kJ mol-1) of ion association of methyl octyl ammonium components for ion association promethyl sulfate in universal solvents at 298.15K cess in different solvents permits us to
estimate the energy of specific solvation interaction. Solvent ε σ Electrical component of the Pyridine-DMFA 25.0 7.6 change of free energy, depending on ion-ion interaction in accordance with Pyridine-acetonitrile 19.2 9.9 the nature of ion association process, Pyridine-propylene carbonate 16.0 11.9 must exceed a sum of vacuum and coNitrobenzene-acetic acid 11.9 16.0 valent components in equation [9.52]. σ elsolv for all above considered systems Propylene carbonate-acetic acid 10.6 17.9 is higher than 85 - 90% because of the general change of the process free energy. The components of free energy of association of CH3(C8H17)3NOSO3CH3 in different solvents7 are presented in Table 9.9. Evidently, the change of permittivity of the universal solvent is the most effective method of affecting the ion concentration. Using the equation relating ion concentration c*M for every type of 1-1 electrolyte with Ka and analytical concentration of electrolyte, cM: el solv
(
)
c M = 4K a c M0 + 1
1/ 2
− 1 / 2K a
[9.107]
one can ascertain, that ε decrease from 185 to 6 changes the ion concentration of methyl octyl ammonium methyl sulfate in solution from 71 to 0.3%. In mixed solvent containing propylene carbonate, the ionic concentration of Et4NBr is also changed from 80% (in propylene carbonate) to 3% (in pyridine), and even to 0.8% (in o-dichlorobenzene or acetic acid). Thus, affecting electrolyte strength by solvent choice permits to change a strong electrolyte into non-electrolyte.
9.4 Mixed solvent influence on the chemical equilibrium
549
A similar phenomenon is observed in conditionally-universal media A-B (A is solvate-active component, B is solvate-inert component). Association constant can be calculated from equations [9.56] or [9.104] for LiBr and KCNS solutions.55,57 Table 9.10. Coefficients of equations [9.56] and [9.106] for association constants of salts [9.56]
Solvent
a00
[9.106] (296K)
a10
a01
a11
a00
a01
Lithium bromide (298.15-323.15K) Propylene carbonate-o-dichlorobenzene
3.38
-808
-361
1.53×105
0.67
152.4
Propylene carbonate-pyridine
-5.56
2040
135
-6923
1.29
111.8
Propylene carbonate-acetic acid
-3.55
1617
70.1
-6160
1.87
49.4
Potassium thiocyanate (298.15-348.15K) Acetonitrile-chlorobenzene Propylene carbonate-chlorobenzene DMSO-chlorobenzene
3.80 1.88 6.18
Figure 9.16. Dependence of lnKa for lithium bromide on 1/ε in the mixed solvent based on propylene carbonate at 298.15K.
-521 -229 -1580
σ el = (βCat⋅ An − βCat + − β An − ) / ε
3.26×104
2.05
84.0
-1.53
4
1.11
85.7
4
0.88
74.9
-16.7
2.60×10 2.73×10
The dependence of lnKa and 1/ε for LiBr in mixed solvent [9.53] shows an influence of solvate-active component on Ka. The difference of LiBr solvation energy in the presence of pyridine and acetic acid may be evaluated in accordance to [9.53], assuming low solvation energy of Li+ by propylene carbonate in comparison with pyridine and negligible value of solvation energy of anion in solvents containing dichlorobenzene (DCB) and pyridine. The data presented in Table 9.10 permit us to calculate the components of free energy for ion association processes. To interpret the data, one must take into account:
σ cov = ∆σvCat⋅ An + ( σsol,Cat⋅ An − σsol,Cat + − σsol, An − ) correspondingly:
-25.4
cov
550
Y. Y. Fialkov, V. L. Chumak
A small value σcov in all considered systems can be explained by low energy of interaction between the filled orbital of cation and anion. On the contrary, contribution of ion-ion interaction to the process energy leads to high value σel in comparison with the covalent component even in media of high permittivity (not to mention media of low permittivities). From [9.106], isodielectric solution formed by two mixed solvents I and II follows the relationship: ln K a, I = (a 01, I / a 01, II ) ln K a, II − (a 01, I / a 01, II )a 00, II + a 00, I
[9.108]
where: 00, 01 I, II
subscripts of coefficients of approximation for dependence on temperature and permittivity subscripts for solvents I and II
The leveling (leveling solvent reduces differences between constants of ion association) or differentiating (differentiating solvent enhances these differences) effect of the first mixed solvent is defined by the ratio of coefficients of equation [9.106] or by the ratio of vacuum electrostatic component of free energy of association process, δ∆Gεel=1. One can calculate ratio a01, PC-DCB/a01, PC-Py =1.36 using the data from Tables 9.10 and 9.11 and assess the leveling effect of the solvent propylene carbonate-pyridine on electrolyte strength. The ratio aPC-DCB/aPC-HAc = 3.1 indicates that propylene carbonate-acetic acid has more pronounced leveling effect on electrolyte strength than propylene carbonate-pyridine. Table 9.11. The components of free energy change (kJ mol-1) in ion association process of different salts in conditionally-universal media at 298.15K -σel
Solvent
At max. ε
At min. ε
-σcov
σelε=1
Lithium bromide Propylene carbonate-o-dichlorobenzene
5.8
37.7
1.6
377
Propylene carbonate-pyridine
4.2
22.9
3.2
277
Propylene carbonate-acetic acid
1.9
20.0
4.6
122
Potassium thiocyanate Acetonitrile-chlorobenzene
5.7
37.1
5.1
208
Propylene carbonate-chlorobenzene
3.2
37.9
2.7
212
DMSO-chlorobenzene
4.0
33.1
2.2
185
ε change of the solvent and universal media affect the ion concentration in solution. The ion concentration in 0.1M solution of LiBr in the mixed solvent propylene carbonateo-dichlorobenzene varies relative to its analytical concentration from 50% in propylene carbonate to 0.1% in dichlorobenzene. The ion concentration in 0.1M solution of KSCN in the solvent propylene carbonate-chlorobenzene varies from 60% in propylene carbonate to 0.06% in chlorobenzene.
9.4 Mixed solvent influence on the chemical equilibrium
551
The same effect can be obtained by means of solvent heating. For instance, LiBr solution in o-dichlorobenzene must be heated to ≈ 215oC to reach the same value of association constant as in propylene carbonate at room temperature. For this reason, solvent may be used as an effective means for tailoring electrolyte strength in conditionally universal media as well. The influence of solvate active property of components of the mixed solvent is revealed in acid solution. The data for some acids in different mixed solvents are presented in Table 9.12.7 Table 9.12. The properties of ion association process of H-acids in conditionally universal media at 298.15K
Acid
Solvent
Coefficients of equation [9.106] a00
a01
Components of energy of ion association process, kJ mol-1 -σcov
σelε=1
σel At max. ε
At min. ε
CH3SO3H
Methanol - n-hexanol
2.49
80.6
6.2
200
6.1
15.0
HSO3F
Methanol-n-butanol
-1.50
165.0
3.7
408
12.5
23.9
CF3COOH
4.86
190.9
12.0
473
15.5
26.7
H2SO4*
-1.82
165
-4.5
408
12.6
23.1
DMSO - CCl4
-1.55
215
-3.8
532
11.4
133
DMSO-o-dichlorobenzene
-1.34
183
-3.3
453
9.7
75.6
DMSO-benzene
0.55
145
1.4
359
7.7
120
DMSO-1,4-dioxane
0.55
129
1.4
319
6.9
80
DMSO-pyridine
1.50
90
3.7
223
4.8
15
Propylene carbonate -DMSO
0.38
146
0.9
361
5.6
7.8
Methanol-n-butanol
-2.06
162
-5.1
401
12.3
22.7
H2SeO4*
*data for the first constant of ion association
The degree of ion association of acids as well as ionophores depends significantly on solvents` permittivity. The influence of the specific solvation by the solvate active component is pertinent from the comparison lnKa - 1/ε isotherms for acid solutions in conditionally universal media S(1) - S(2), S(1) - S(3)... Figure 9.1758 gives lnKa - 1/ε relationships for solutions of sulfuric acid in binary solvent mixture containing DMSO and CCl4, benzene, dioxane, pyridine. The increase in the second component basic capacity according to [9.50] leads to decrease in lnKa. General analysis of the binary solvent mixtures formed by two solvate active components (these solvents are often used in analytical and electrochemistry) was conducted to evaluate their effect on H-acids.59 The analysis was based on an equation which relates the constant of ion association, Ka, of the solvent mixture and constants of ion association of the acid K Aa and KBa of each component of the mixed solvent, using equilibrium constants of scheme [9.105] - heteromolecular association constant, Kadd; ionization constant of the
552
Y. Y. Fialkov, V. L. Chumak
Figure 9.18. Dependence of Ka for HSO3F on solvent composition in mixed solvents: A - n-butanol at 298.15K.7 B : 1 - n-hexane; 2 - nitrobenzene; 3 - methanol; 4 - N,N- dimethyl aniline. Figure 9.17. Dependence of Ka of sulfuric acid on 1/ε in mixed solvents based on DMSO at 298.15K.
adduct, Ki; and resolvation constant, Kus, (e.g., equilibrium constant of process HAn$A + B ↔ HAn$B + A), and with molar fractions of components of the mixed solvent xA and xB:
(
K a = K aB + K aB − K aA
)[1− (K / (1+ K B a
add
x B (1 + K i ))
)]
[9.109]
The equation [9.109] changes into equation [9.100] when both components of the mixed solvent have donor character, e.i., Kus = 0. The analysis of equation [9.100]59 shows that the solvate active property of components of the mixed solvent A-B have to be changed to influence the dependence of Ka on composition of the mixed solvent. Some examples of such influence are presented in Figure 9.18. 9.4.6 SOLVENT EFFECT ON EXCHANGE INTERACTION PROCESSES Chemical processes of exchange interaction, given by equation: EF+ HL ↔ EL + HF
[9.110]
are common in research and technology, but the solvent effect on their equilibrium was not frequently studied. Let us systematize the existing experimental data on these processes.60 Systems with non-associated reagents The process of exchange interaction between bis-(carbomethoxymethyl)mercury and mercury cyanide (CH3OCOCH2)2Hg + Hg(CN)2 ↔ 2CH3OCOCH2HgCN
[9.111]
has been studied in the mixed solvent DMSO-pyridine. The components of solvent possess almost equal ability to specific solvation of mercury-organic compounds. This ability is
9.4 Mixed solvent influence on the chemical equilibrium
553
confirmed by equality of the constants of spin - spin interaction 2I(1H - 199Hg) for different organic compounds in these solvents. This was also confirmed61 for participants of the equilibrium [9.111]. Systems with one associated participant of equilibrium Methanolysis of triphenylchloromethane may serve as an example of such reaction: Ph3CCl + CH3OH ↔ Ph3COCH3 + HCl
[9.112]
This process has been studied in benzene, 1,2-dichloroethane, chloroform, trifluoromethylbenzene and in mixed solvents hexane - nitrobenzene, toluene nitromethane, toluene - acetonitrile. Low constant of the ion associate formation process of triphenylchloromethane, high constant of the ion association process, and low constant of the heteromolecular association process of HCl (HCl solutions in listed solvents obey the Henry’s Law) show that only methanol is an associative participant of the equilibrium. Systems with two associated participants of equilibrium These systems have been studied be means of reaction of acidic exchange between acids and anhydrides with different acidic groups: CH2ClCOOH + (CH3CO)2O ↔ CH2ClCOOCOCH3 + CH3COOH
[9.113a]
CH3COOH + (CH2ClCO)2O ↔ CH2ClCOOCOCH3 + CH2ClCOOH
[9.113b]
in binary mixed solvents formed from tetrachloromethane with chlorobenzene, trifluoromethylbenzene and nitrobenzene. The system trifluoroacetic acid and methanol, undergoing esterification reaction: CF3COOH + CH3OH ↔ CF3COOCH3 + H2O
[9.114]
also belongs to the group. The equilibrium process has been studied in binary mixed solvents such as hexane-chloroform, hexane-chlorobenzene, chloroform-chlorobenzene. Also, re-esterification CH3COOC3H7 + CX3COOH ↔ CX3COOC3H7 + CH3COOH
[9.115]
(where X is F or Cl) has been studied for the full concentration range of propyl acetate trifluoro- and trichloroacetic acids.62 Etherification process of ethanol by caproic or lactic acids has been studied by these authors.63 The reaction of acetal formation was studied64 RCHO + R’OH ↔ RCH(OR’)2 + H2O
[9.116]
in systems of isoamyl alcohol-benzaldehyde (a) and ethanol - butyric aldehyde (b). The relationship between the maximum output of the reaction of exchange interaction and associative state of equilibrium [9.110] participants follows the Gibbs-Dugem-Margulis equation:
554
Y. Y. Fialkov, V. L. Chumak
x s d ln as + Σx i d ln a i = 0
[9.117]
where: xS, xi as, ai
molar fractions activity of the solvent and participants of the chemical process, correspondingly
Izmaylov65 showed that maximum output of reaction depends on initial composition of reagents. Associative and dissociative processes of the chemical system components are accompanied by the change of their chemical potentials if the solvent potential depends on total activity of solution (it depends on concentration in dilute solutions): i
µ s = µ so − RT ∑ ln c i
[9.118]
i =1
Hence the condition of chemical equilibrium in non-ideal (regular) systems is presented in the following form:
(∂G) P ,T = (∂GS ) P ,T
+ ∑ (∂G i ) P ,T
i
∂ lnK ch = ns ∂V
∂∑ c i i =1
∂V
[9.119] i
∂ − ∂V
∂ γ − ∏ ∂ V i =1 ι
vi i
∑µ i =1
o i
v i + ns ∆µ so RT
[9.120]
Inserting values of chemical potentials for equilibrium participants [9.110] and solvent in equation [9.119], one can obtain the equation relating the equilibrium constant, Kch, change relative to the solvent composition (in volume parts, V), mol number, nS, activity coefficients, γi, stoichiometric coefficients, vi, of process [9.110] and chemical potential, µ oi , for every participant of equilibrium at concentration 1M. Equation [9.120] changes to [9.121], if only EF and EL undergo homomolecular association: ∂ ln K ch 1 ∂ 1 o ≈ 2ns c EF − EL EF ∂V ∂V K K dim dim
dK ch + dε
[9.121]
where: Kdim
equilibrium constant of homomolecular association
Equations [9.120] and [9.121] describe the solvent effect on Kch for a non-ideal liquid system. The dependence of lnKch and reciprocal ε is linear: ln K ch = a 00 + a 01 / ε
[9.122]
Coefficients of the equation and some thermodynamic properties of exchange interaction processes are presented in Table 9.13. The degree of heteromolecular association of esters in reaction [9.121] is higher than for acids when Kdim values are equal. Thus the change of equilibrium constant according to [9.121] depends on electrostatic component. Kch for these reactions also depends on exponentially reciprocal ε.
9.4 Mixed solvent influence on the chemical equilibrium
Figure 9.19. Dependence of equilibrium constants for triphenylchloromethane methanolysis on 1/ε of mixed solvent at 298.15K: 1 - hexane-nitrobenzene; 2 - toluene-nitromethane; 3 - toluene-acetonitrile.
555
The dependence of equilibrium constant for reaction [9.116] on ε is similar in a concentration range from 0.5 to 1.0 molar fraction of aldehyde, if the degrees of homomolecular association for aldehydes and acetals are smaller than for alcohol and water. Kch increase with increasing ε of media is a general property of all processes of exchange interaction presented in the Table 9.13. Dependence of lnKch - 1/ε for process [9.111] is not linear in all mixed solvents (Figure 9.19). Extreme Kch dependence on 1/ε is explained by chemical effect of solvation (it is explained in detail elsewhere60). Correlations with different empirical parameters of individual solvent parameters are held true for this process. The best of them is correlation (r=0.977) with parameter ET.
Table 9.13. Characteristics of some processes of exchange interaction
Reaction
Coefficients of equation [9.122]
Free energy components, kJ/mol
a00
-a01
-σ
σelε=1
σG at max. ε
σG at min. ε
[9.111]
2.40
97.6
5.9
241
5.2
19.0
[9.115], a, (F)
1.72
19.6
4.3
48.5
3.1
6.3
[9.115], b, (Cl)
2.55
17.1
6.3
42.3
5.1
7.0
[9.116], a
1.24
47
3.1
116.4
7.5
9.9
[9.116], b
2.47
16.7
6.1
41.3
3.4
4.1
cov
The influence of molecular state of participants of exchange interaction on the process equilibrium is reflected in Kch dependence (processes [9.113]) on the mixed solvent composition (Figure 9.20). The change of Kdim for acetic acid is less pronounced than for monochloroacetic acid, when CCl4 is replaced by chlorobenzene. Thus Kch should increase for reaction [9.113a] and it should decrease for reaction [9.113b], with ε increasing in accordance to equation [9.121]. Correlation of Kdim for acetic and monochloroacetic acid is obtained when chlorobenzene is replaced by nitrobenzene. This leads to decrease of Kdim when ε increases in the case of reaction [9.113a], and the opposite is the case for reaction [9.113b]. That is why a maximum is observed in the first process (see Figure 9.20 on the left hand side), and a minimum is observed for the second process.
556
Y. Y. Fialkov, V. L. Chumak
Figure 9.20. Dependence of Kch on the mixed solvent composition for processes [9.113a] and [9.113b] at 298.15K: 1-tetrachloromethane-nitrobenzene; 2- tetrachloromethane-chlorobenzene.
At equilibrium [9.114] (with three associated participants), association degree of associated participants decreases with media polarity increasing. It leads to Kch increasing and to Kadd decreasing in the process of heteromolecular association of acid and alcohol. Thus, Kch depends to a low extent on the composition of studied mixed solvents. Vacuum component of free energy of processes of exchange interaction (Table 9.13) is low; σGel values are low, too. This indicates, that ε change does not lead to the change of the reaction output. Solvent replacement leads to essential change of reaction output in process [9.111], characterized by the high value of σel (see Table 9.13), from 6% in pyridine to 60% in DMSO (in 0.1M solutions). Kch increase in pyridine solution to its value in DMSO requires cooling to -140oC. 9.4.7 MIXED SOLVENT EFFECT ON PROCESSES OF COMPLEX FORMATION Non-aqueous solvent effect on equilibrium and thermodynamic of complexation processes are summarized in monographs.67-69 The solvent effect on complex formation processes has two aspects. The first is a change of coordination sphere composition, changing participation of solvent as a chemical regent:
[ML m ]
p+
+A ← →[ML m −1 A] + L ←→L ←→[MAn ] + mL p+
p+
[9.123]
where: Lm A, B
a ligand in initial complex components of binary solvent
If both components of the mixed solvent are solvate-active, equilibrium [9.123] may be presented as:
[ML m ]
p+
+A+B ← →[ML m − 2 AB ] + 2L ↔L ↔ [MAn − q B q ] + mL [9.123a] p+
p+
9.5 The mixed solvent effect on the chemical equilibrium thermodynamics
557
Coordination solvents A and B taking part in the processes are the same ligands as L. The second aspect of the solvent effect is related to the thermodynamic properties of the complex, such as formation constant. Discussing the problem of solvent effect on complex stability, authors often neglect the change of a complex nature, due to the change in solvent as follows from equations [9.123] and [9.123a]. These processes may also be regarded as resolvation processes. When the components of mixed solvate are solvate-inert in relation to the complex, equation [9.53a] takes the following form: ln K k = a 00 + a 01 / ε Figure 9.21. Dependence of stability constants for cadmium thiocyanate complexes on 1/ε in mixed solvent water-methanol at 298.15K.
[9.124]
Coefficient a00 has a physical meaning of logarithm of formation constant at ε → ∞ related to the standard ionic strength I = 0. In Figure 9.21, this dependence is presented.
Table 9.14. Complexation process characteristics for system Cd2+-thiourea in mixed solvent water-methanol at 298.15K Coefficient of equation Constant K2 K3 K4 K5
a00
a01
Correlation coefficient r
-0.42 1.07 7.79 2.72
367.7 561.9 631.3 837.0
0.996 0.987 0.991 0.988
σGcov -1.0 2.6 19.3 6.7
Component σG, kJ/mol -σGcov σelε=1 910 1390 1563 2072
in H2O
in MeOH
11.6 17.7 19.9 26.4
27.9 42.7 47.9 63.6
The data for cadmium thiourea complexes in water-methanol mixed solvent70 are presented in Table 9.14. Dependence of stability constant on the solvent composition is very complex in the mixed solvents formed by two solvate-active components as defined by the relative activity, L, of A and B. Dependencies of stability constant for some complexes in water-B solution (where B is methyl acetate, methanol, etc.69) are presented in Figures 9.21 and 9.22 for complex NiEn2+ (En=ethylenediamine) in water-non-aqueous solution (the constants are presented relative to stability constant in water).
9.5 THE MIXED SOLVENT EFFECT ON THE CHEMICAL EQUILIBRIUM THERMODYNAMICS In most cases, thermodynamic characteristics of equilibrium processes are determined by temperature dependence of equilibrium constant K = f(T), according to the classical equa-
558
Y. Y. Fialkov, V. L. Chumak
Figure 9.23. Isotherms lnK - f(ε) for different temperatures (T1
tion: ∆H=RT2(dlnK/dT) and ∆S = RlnK + T(dlnK/dT) or ∆S = ∆H/T + RlnK. Because equilibrium constant is a function of both temperature and permittivity [9.52a], i.e., Figure 9.22. Stability constant of amine complexes of some Ni(NH3)+ in water-organic solvents: 1-NiEn2+; 2-[N; NH3]2+; 3-[CdCH3NH2]2+; 4-[AgCH3NH2]2+; 5-[AgEn2]2+; 6-[NiEn2]2+.
K = f (T, ε)
[9.125]
A change of constant during temperature change is a consequence of enthalpy of process (not equal to zero) and temperature dependence of permittivity. Thus, the change of equilibrium constant with temperature changing depends on both self-chemical equilibrium characteristics (here ∆H) and solvent characteristics (namely, solvent permittivity influence on equilibrium constant). This is pertinent from Figure 9.23, which shows dependence of equilibrium constant logarithm on reciprocal permittivity (isotherms). Equilibrium constants, obtained experimentally (dotted line a - b) correspond to various values of permittivity. The thermodynamic characteristics of a chemical process, calculated from equilibrium constant polytherms, are integral characteristics (∆Hi and ∆Si), i.e., they consist of terms related to process itself (∆HT and ∆ST) and terms depended on permittivity change with temperature changing (∆Hε and ∆Sε): ∆Hi = ∆HT + ∆Hε
[9.126a]
∆S i = ∆ST + ∆S ε
[9.126b]
The thermodynamic description of a process in a solvent has to lead to defined values of ∆HT and ∆ST, which are called van’t Hoff or original parameters.7 It follows from Figure 9.23 that one has to differentiate the equilibrium constant of polytherms corresponding to isodielectric values of the solvent on temperature (e.g., to equilibrium constant values of c-d dotted line).
9.5 The mixed solvent effect on the chemical equilibrium thermodynamics
559
Division of equation [9.126] into the constituents must be done with full assurance to maintain influence of different solvents. For this reason, it is necessary to have a few isotherms of equilibrium constant dependencies on permittivity (K=f(ε)T). We have developed an equation of a process depicted by scheme [9.45] after approximating each isotherm as lnK vs. 1/ε function and by following approximation of these equations for ε-1, ε-2, ε-3...ε-j conditions. We then can calculate the integral value of process entropy by differentiating relationship [9.55] versus T, e.g., ∆G= -RTlnK: ij==mn ∆S i = R ∑ (1 − i )a ji T i ε i i = 0 j = 0
(
i =m j =n
) − j (d ln ε / dT ) / ε ∑ a j −1
i =0 j =0
ji
/ T i −1 / ε i
[9.127]
Here, the first term of the sum in brackets corresponds to ∆ST/R value and the second term to ∆S/R. We can determine “van’t Hoff’s” (original) constituents of the process entropy if all the terms containing dlnε/dT are equal to zero: i =m j =n
(
∆ST = R ∑ (1 − i )a ji / T i ε j i =0 j =0
)
[9.128]
Van’t Hoff’s constituent of enthalpy is determined in analogous manner as ∆H=∆G+T∆S (∆HT): i =m j =n
(
∆HT = −R ∑ a ji / T i −1 ε j i =0 j =0
)
[9.129]
Such approach may by illustrated by dependence of formic acid association constant on temperature and permittivity in mixed solvents: water-ethylene glycol,72 which is approximated from equation ln K = a 00 + a 01 / T + a 02 / T 2 + a10 / ε + a11 / (εT ) where: a00 a01 a02 a10 a11
= 40.90 = -2.17×104 = 3.11×106 = -425.4 = 2.52×105
Hence ∆Hi = −R[a 01 + 2a 02 / T + a11 / ε + T (d ln ε / dT )(a10T + a11 ) / ε] ∆S i = R[a 00 − a 02 / T + a10 / ε − (d ln ε / dT )(a10T + a11 ) / ε] and consequently,
560
Y. Y. Fialkov, V. L. Chumak
∆HT = −R(a 01 + 2a 02 / T + a11 / ε)
∆ST = R(a 00 − a 02 / T + a10 / ε)
Comparison of ∆Hi with ∆HT and ∆Si with ∆ST shows that integral values of thermodynamic characteristics of the ionic association process for HCOOH are not only less informative than van’t Hoff’s characteristics, but they contradict the physical model of the process in this case. Indeed, corresponding dependencies of compensative effect are described by equations: ∆Hi = -0.704∆Si + 22.87 kJ/mol ∆HT = 0.592∆ST - 22.22 kJ/mol Thus, if the compensative effect for integral thermodynamic functions is interpreted, one should conclude that characteristic temperature (this is tanα = ∆H/∆S) is a negative value, but that is, of course, devoid of physical sense. In common cases, when the dependence [9.125] is approximated using equation [9.56], “van’t Hoff’s” (original) thermodynamic constituents of equilibrium process are equal to: ∆HT = −R (a 01 + a11 / ε)
[9.130]
∆ST = R (a 00 + a10 / ε)
[9.131]
As it is evident from [9.130] and [9.131], the original thermodynamic characteristics of the process are summarized by the terms where the first of them does not depend, and the second one depends, on permittivity. These terms accordingly are called covalent and electrostatic constituents of the process in solution, i.e., ∆H cov = −Ra 01 ∆S cov = Ra 00
and and
∆H el = −Ra11 / ε ∆S el = Ra10 / ε
[9.132] [9.133]
It should be emphasized that ∆GT values are substantially more informative than integral thermodynamic values for thermodynamic analysis of chemical equilibrium in solutions of mixed solvents. Indeed, in frequent cases of equilibrium constant polytherm characterized by an extreme (i.e., sign of ∆Hi changes), one may arrive at the wrong conclusion that the process nature changes greatly at temperature of extreme value. As has been shown,71 appearance of extreme value of equilibrium constant polytherms is caused in the majority of cases by temperature change of solvent permittivity. The condition of extreme appearance for the process described by equation [9.55] is: a 01 exp(αTextr + γ) + αa11Textr (a10Textr + a11 ) = 0
[9.134]
where: α,γ γ Textr
the coefficients of lnε = αT + γ equation activity coefficient temperature corresponding to extremum of isotherm of equilibrium constant
9.5 The mixed solvent effect on the chemical equilibrium thermodynamics
561
Let us illustrate the principle of integral thermodynamic characteristics of chemical equilibrium in mixed solvent based on the example of ionic association process of methylsulfuric acid HSO3CH3 and its tetraalkylammonium salt CH3(C8H17)3N + OSO3CH3 in mixed solvent - methanol - n-butanol.73 The dependencies of ionic association constants for these electrolytes on temperature and permittivity are described by equations based [9.56] the relationship: Ka,acid = 14.53 - 3436/T - 130.28/ε + 5.84×10/εT Ka,salt = 11.8 - 2514/T - 198.70/ε + 7.79×10/εT Thermodynamic characteristics for these systems, calculated based on [9.130][9.133] equations, are given in Tables 9.15 and 9.16. Table 9.15. Thermodynamic characteristics of ionic association process of HSO3CH3 in mixed solvent: methanol - n-butanol (298.15 K) (∆Hcov = 28.56 kJ/mol, ∆Scov = 120.8 J/(mol*K)) BuOH mol%
ε
60.39
∆H, kJ/mol
∆S, J/(mol* K)
∆Hi
∆HT
∆Hel
∆Si
∆ST
∆Sel
20
20.62
4.29
-24.28
121.4
66.6
-54.1
51.60
22
20.90
6.50
-22.07
119.9
71.6
-49.2
37.24
24
21.19
9.89
-20.23
118.8
75.7
-45.1
25.32
26
21.49
8.34
-18.67
118.0
79.1
-41.7
15.82
28
21.79
11.22
-17.34
117.6
82.1
-38.7
8.76
30
22.09
12.38
-16.18
117.3
84.7
-36.1
Table 9.16. Thermodynamic characteristics of ionic association process of CH3(C8H17)3N + OSO3CH3 in mixed solvent: methanol - n-butanol (298.15 K) (∆Hcov=20.90 kJ/mol, ∆Scov= 98.1 J/(mol K)) BuOH mol%
ε
60.39
∆H, kJ/mol
∆S, J/(mol K)
∆Hi
∆HT
∆Hel
∆Si
∆ST
∆Sel
20
3.85
-11.23
-32.13
66.1
15.5
-82.6
51.60
22
5.00
- 8.31
-29.21
67.7
23.0
-75.1
37.24
24
6.00
- 5.88
-26.78
69.1
29.3
-68.8
25.32
26
6.90
- 3.82
-24.72
70.5
34.6
-63.5
15.82
28
7.70
-2.05
-22.95
71.8
39.1
-59.0
8.76
30
8.45
- 0.52
-21.42
73.1
43.0
-55.1
The analysis of data leads to conclusion that ∆Hi depends weakly on permittivity for both acids and salts. This contradicts the nature of the process. In contrast, the rate of ∆HT
562
Y. Y. Fialkov, V. L. Chumak
change is large enough. The analogous observations can be made regarding the character of ∆Si and ∆ST values changes along with permittivity change. If considerable endothermic effect of the ionic association for acid can be explained by desolvation contribution of ionic pair formation (i.e., solvated molecule of acid), the endothermic effect of ionic pair formation by such voluminous cations contradicts the physical model of the process. Estimation of desolvation energy of ionic pair formation of salt ions according to the equation74 shows that this energy is two orders of magnitude lower than the energy of heat movement of solvent molecules. For this reason, the process of ionic association of salt ions is exothermal, as seen from ∆HT values. The exothermal character increases with permittivity decreasing, i.e., with increase of ion-ion interaction energy. Dependencies ∆Hi on ∆Si for salt and acid are not linear, i.e., the compensative effect is not fulfilled. Consequently, the substantial change of the ionic association process is due to change of solvent composition. Considering the same chemical characteristics of mixed solvent components, this conclusion is wrong because the compensative effect if fulfilled for van’t Hoff’s components is: ∆HT = 0.448 ∆ST - 28.5 kJ/mol ∆HT = 0.389 ∆ST - 17.3 kJ/mol The smaller the magnitude of the characteristic temperature for salt (389K), in comparison to acid, the smaller the barrier of the ionic pairs formation process. That is true because the association process in the case of acid is accompanied by energy consumption for desolvation.
REFERENCES 1
Timmermans J., The Physico-Chemical Constants of Binary Systems in Concentrated Solutions V1-4, N-Y-London, Intern. Publ. ING,1959. 2 Krestov G., Afanas’ev V., Efimova L., Physico-chemical Properties of Binary Solvents, Leningrad, Khimiya,1988. 3 Fialkov Yu., Physical-Chemical Analysis of Liquid Systems and Solutions, Kiev, Naukova dumka,1992. 4 Kabachnik M., Doklady Akademii Nauk USSR, 83, 6, 1952 859-862 5 Kabachnik M., Izvestiya Akademii Nauk USSR, 1, 1955, 98-103 6 Fialkov Yu., Diner L., J. Gener. Chem. (russ.), 48, 2, 1978, 253-256. 7 Fialkov Yu., Solvent as an Agent of Chemical Process Control, Leningrad, Khimiya, 1990. 8 Fialkov Yu., Ligus V., Doklady Akademii Nauk USSR, 197, 6,1971, 1353-1354. 9 Akhadov Ya., Dielectric Properties of Binary Solutions, Moskow, Nauka, 1977. 10 Fialkov Yu., Grishenko V., Electrodeposition of Metals from Non-Aqueous Solutions, Kiev, Naukova dumka, 1985. 11 Fialkov Yu., Chumak V., Kulinich N., Electrochem.(russ.), 18, 8, 1024-1027. 12 Anosov V., Ozerova M., Fialkov Yu., Bases of Physical-Chemical Analysis, Moskow, Nauka, 1976, 1978. 13 Rudnickaya A., Maiorov V., Librovich N., Fialkov Yu., Izvestiya Akademii Nauk USSR, 5, 1981, 960-966. 14 Rudnickaya A., Maiorov V., Librovich N., Fialkov Yu., Izvestiya Akademii Nauk USSR, 11, 1981, 2478-2484. 15 Alexsandrov V., Acidity of Non-Aqueous Solutions, Kharkov, Visha skola,1981. 15a Prabir K.Guna, Kiron K.Kundu, Can. J. Chem.,1985, 63, 804-808. 16 Palm V., The Basis of Quantitative Theory of Organic Reactions, Leningrad, Khimiya,1977 17 Balakrishnan S., Eastael A., Austr. J. Chem., 1981, 34, 5, 933-941. 18 Langhals M., Chem. Ber., 1981, 114, 8, 2907-2913. 19 Fialkov Yu., Barbash V., Bondarenko E., Sovyet Progress in Chemistry, 53, 5, 490-491. 20 Abrachem R. J., Bretchnaider E., Internal Rotation of Molecules, 1977, 405-409. 21 Simoshin V., Zefirov N., J. Mendeleev-Soc., 1984, 29, 5, 521-530.
9 Mixed solvents
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
563
Minkin V., Osipov O., Zhdanov Yu., Dipole Moments in Organic Chemistry, Leningrad, Khimiya,1968. Christian S., Lane E., in Solutions and Solubilities, V1, 1975, John Wiley & Sons, N-Y, 327-377. Komarov E., Shumkov V., Chemistry Chem. Thermodyn. Solution, Leningrad, Univers,1973, 3, 151-186. Bishop R., Sutton L., J. Chem. Soc., 1964, 6100-6105. Bhownik B., Srimani P., Spectrochim. acta, 1973, A29, 6, 935—942. Ewall R., Sonnesa A., J. Amer. Chem. Soc., 1970, 92, 9, 3845-3848. Drago R., Bolles T., Niedzielski R., J. Amer. Chem. Soc., 1966, 88, 2717-2721. Fialkov Yu., Barbash V., Theoret. Exper. Chem.(russ.), 1986, 22, 248-252. Barbash V., Fialkov Yu., Shepet`ko N., J. Gener. Chem.(russ.), 1983, 53, 10, 2178-2180. Fialkov Yu., Barbash V., Sovyet Progress in Chemistry, 1985, 51, 4, 354-367. Barbash V., Fialkov Yu., Sovyet Progress in Chemistry, 1983, 49, 11, 1157-1160. Barbash V., Golubev N., Fialkov Yu., Doklady Akademii Nauk USSR, 278, 2, 1984, 390-391. Borovikov A., Fialkov Yu., J. Gener. Chem.(russ.), 1978, 48, 2, 250-253. Pavelka Z., Sobczyk L., Roczn. chem., 1975, 49, 7/8, 1383-1394. Padova J., J. Phys. Chem., 1968, 72, 796-801. Bachshiev, Spectroscopy of Intermolecular Interactions, Leningrad, Nauka, 1972. Mishustin A., J. Phys. Chem.(russ.), 1987, 61, 2, 404-408. Mishustin A., Kruglyak A., J. Gener. Chem.(russ.), 1987, 57, 8, 1806-1810. Reynolds W., Inorg.Chem., 1985, 24, 25, 4273-4278; J. Chem. Soc. Comm., 1985, 8, 526-529. Busse J., Ber. Buns. phys. Chem., 1985, 89, 9, 982-985. Fuldner H., Ber. Buns. phys. Chem., 1982, 86, 1, 68-71. Kobayashi K., Bull. Chem. Soc. Japan, 1987, 60, 467-470. Marcus Y., Ion Solvation, N-Y, Wiley, Chichester,1985. Gordon J., The Organic Chemistry of Electrolyte Solutions, 1977, N-Y-London-Sydney-Toronto, John Wiley&Sons. Buleishvili M., Fialkov Yu., Chumak V., Sovyet Progress in Chemistry, 1983, 49, 6, 599-602. Fialkov Yu., Chumak V., Rudneva S., Doklady Akademii Nauk Ukraine,"B", 1984, 4, 49-53. Dorofeeva N., Kovalskaya V., Fialkov Yu., Shlyachanova V., Sovyet Progress in Chemistry, 1984, 50, 5, 476-479. Lukyunova S., Fialkov Yu., Chumak V., Sovyet Progress in Chemistry, 1987, 53, 3, 252-255. Levinskas A., Shmauskaite S., Pavidite Z., Electrochemistry(russ.), 1978, 14, 12, 1885-1889. Fialkov Yu., Pinchuk V., Rudneva S., Chumak V., Doklady Akademii Nauk USSR, 1984, 277, 1, 136-138. Fialkov Yu., Dorofeeva N., Chumak V., Thermodynamic Properties of Solutions, Ivanovo, 1984, 95-100. Izmailov N., Electrochemistry of Solutions, Khimiya, Moskow, 1966. Busse J., Ber. Buns. phys. Chem., 1985, 89, 9, 982-985. Fialkov Yu., Chumak V., Borchashvili A., Sovyet Progress in Chemistry, 1984, 50, 7, 10, 705-708, 1023-1027. Fialkov Yu., Tarasenko Yu., Kryukov B., Doklady Akademii Nauk Ukraine,"B", 1985, 2, 59-61. Kryukov B., Tarasenko Yu., Fialkov Yu., Chumak V., Sovyet Progress in Chemistry, 1985, 51, 1, 34-38. Fialkov Yu., Chumak V., Chem. Chem. Technol.(russ.), 1976, 19, 2, 265-268. Buleishvili M., Fialkov Yu., Chumak V., Sovyet Progress in Chemistry, 1983, 50, 2, 214-216. Fialkov Yu., Zhadaev B., Thermodynamic Properties of Solutions , Ivanovo, 1979, pp.90-104. Zhadaev B., Fialkov Yu., Doklady Akademii Nauk USSR,1978, 242, 6,1350-1351. Fialkov Yu., Fenerli G., J. Gener. Chem.(russ.), 1964, 34, 10, 3146-3153. Fialkov Yu., Fenerli G., J. Gener. Chem.(russ.), 1966, 36, 6, 967-973. Fialkov Yu., Fenerli G., J. Gener. Chem.(russ.), 1966, 36, 6, 973-981. Reichardt Ch., Solvents and Solvent Effects in Organic Chemistry, VCH, Weinheim, 1988. Krigovski T., Wrona P., Zielikowska U., Reichardt Ch., Tetrahedron, 1985, 41, 4519-4527. Gutmann V., Coordination Chemistry in Non-Aqueous Solvents, 1968, Springer-Verlag,Wien-N-Y. Burger K. Solvatation, Ionic and Complex Formation Reactions in Non-Aqueous Solvents, 1983, Akademia Kiado, Budapest. Complexing in Non-Aqueous Solution, Ed. G. Krestov, Moscow, Nauka, 1989. Migal P., Ciplakova V., J. Neorg. Chem.(russ.), 1964, 9, 3, 601-605. Fialkov Yu., J. Chem. Thermodynamic Thermochemistry (russ.), 1993, 2, 2, 113-119.
564
72 73 74
Y. Y. Fialkov, V. L. Chumak Alexsandrov V., Mamina E., Bondarev, Problems of Solvation and Complexing in Solutions, Ivanovo, 1984, 1, 174-176. Fialkov Yu., Koval`skaya V., Chem. Phys. Reports, 1996, 15, 11, 1685-1692. Fialkov Yu., Gorbachev V., J. Neorg. Chem.(russ.), 1977, 42, 8, 1401-1404.
10
Acid-base Interactions 10.1 GENERAL CONCEPT OF ACID-BASE INTERACTIONS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Acid-base interactions have found numerous applications in research dealing with adsorption of molecules of liquids on the surfaces of solids. The main focus of this research is to estimate the thermodynamic work of adhesion, determine mechanism of interactions, analyze the morphology of interfaces and various surface coatings, develop surface modifiers, study the aggregation of macromolecular materials, explain the kinetics of swelling and drying, understand the absorption of low molecular weight compounds in polymeric matrices, and determine the properties of solid surfaces. In addition to these, there are many other applications. Several techniques are used to determine and interpret acid-base interactions. These include: contact angle, inverse gas chromatography, IGC, Fourier transform infrared, FTIR, and X-ray photoelectron spectroscopy, XPS. These methods, as they are applied to solvents are discussed below. Contact angle measurements have long been used because of common availability of instruments. In recent years, they have been developed from simple optical devices to the present day precise, sophisticated, computer-controlled instruments with sufficient precision. Van Oss and Good1,2 developed the basic theory for this method. Their expression for surface free energy is used in the following form: γ = γ LW + γ AB = γ LW + 2 γ + γ −
[10.1.1]
where: γ LW γ AB γ+ γ-
Lifshitz-van der Waals interaction acid-base interaction Lewis acid parameter of surface free energy Lewis base parameter of surface free energy.
The following relationship is pertinent from the equation [10.1.1]: γ AB = 2 γ + γ −
[10.1.2]
A three-liquid procedure was developed1-3 which permits the determination of the acid-base interaction from three measurements of contact angle:
566
George Wypych
(γ (1 + cos θ ) = 2( γ (1 + cos θ ) = 2( γ
γ L1 (1 + cos θ1 ) = 2 γ L2 γ L3
LW S
) γ γ ) γ γ )
γ LW γ S+ γ L−1 + γ S− γ +L1 L1 +
2
LW S
γ LW γ S+ γ L− 2 + L2 +
3
LW S
γ LW γ S+ γ L− 3 + L3 +
− S
+ L2
− S
+ L3
[10.1.3]
Solving this set of equations permits the calculation of solid parameters from equation [10.1.1]. The work of adhesion, Wa, between a solid and a liquid can be calculated using the Helmholtz free energy change per unit area:
(
−∆GSL = WaSL = γ S + γ L − γ SL = 2
γ SLW γ LW + γ S+ γ L− + γ S− γ +L L
)
[10.1.4]
Determination is simple. However, several measurements (usually 10) should be taken to obtain a reliable averages (error is due mostly to surface inhomogeneity). There is a choice between measuring the advancing or the receding contact angle. Advancing contact angles are more representative of equilibrium contact angles.3 Inverse gas chromatography data are interpreted based on Papirer’s equation:4,5 RT lnVN = 2N( γ SD )1/ 2 a( γ DL )1/ 2 + c
[10.1.5]
where: R T VN N γD S, L a c
gas constant temperature net retention volume Avogadro’s number dispersion component of surface energy indices for solid and liquid, respectively molecular area of adsorbed molecule integration constant relative to a given column.
Several probes are used to obtain the relationship between RTlnVN and aγ DL . From this relationship the reference retention volume, VNREF , is calculated and used to calculate the acid-base interaction’s contribution to the free energy of desorption: ∆G AB = RT ln
VN VNREF
[10.1.6]
If data from a suitable temperature range for ∆GAB can be obtained, the acid-base enthalpy, ∆HAB can be calculated using the following equation: ∆HAB = K a DN + K b AN where: Ka Kb DN AN
acid interaction constant base interaction constant donor number acceptor number
[10.1.7]
10.1 General concept of acid-base interactions
567
323.5
199
323
Cl 2p3/2, eV
Binding energy, eV
199.5
198.5
198
197.5
1
1.5
2
2.5
3
3.5
-1
-∆H, kcal mol
Figure 10.1.1. Correlation of binding energy and ∆HAB for several polymers. [Adapted, by permission, from J F Watts, M M Chehimi, International J. Adhesion Adhesives, 15, No.2, 91-4 (1995).]
322.5
322
321.5
0
20
40
60
80 100 120 140
-∆H , kJ mol-1 AB
Figure 10.1.2. Correlation of binding energy and ∆HAB for several solvents. [Adapted, by permission, from J F Watts, M M Chehimi, International J. Adhesion Adhesives, 15, No.2, 91-4 (1995).]
Finally the plot of ∆HAB/AN vs. DN/AN gives Ka and Kb. Further details of this method are described elsewhere.4-9 It can be seen that the procedure is complicated. The various conditions of experiments conducted in various laboratories were sufficiently different to prevent correlation of data between laboratories.10 To rectify this situation, a large body of data was obtained for 45 solvents and 19 polymers tested under uniform conditions.10 Fowkes11 showed that the carbonyl stretching frequency shifts to lower values as the dispersion component of surface tension increases. The following empirical relationship was proposed: ∆HAB = 0.236∆v AB
[10.1.8]
where: ∆HAB ∆vAB
the enthalpy change on acid-base adduct formation projection of carbonyl stretching frequency on dispersive line.
It should be noted that, as Figure 7.1.16 shows, the change in frequency of carbonyl stretching mode is related to the process of crystallization.12 XPS is emerging as very precise method for evaluating acid-base interactions based on the works by Chehimi et al.6-9 There is very good correlation between XPS chemical shift and the change in exothermic enthalpy of acid-base interaction. Drago’s equation is used for data interpretation:13 −∆HAB = E A E B + CA CB
[10.1.9]
where: A, B EA, EB CA, CB
subscripts for acid and base, respectively susceptibility of acid or base species to undergo an electrostatic interaction susceptibility of acid or base species to undergo a covalent interaction.
568
George Wypych
50
1
Cl2p/ClLMM
PMMA surface content, %
CCl4
0.9 0.8 TLC 0.7
DCM 0.6
DCE
0.5 THF
CHCl3
DXN 0.4 20 15 10
5
0
5
DN
10 15 20 AN
40 30 20 10 0 0.4
0.5
0.6
0.7
0.8
0.9
1
Cl2p/ClLMM
Figure 10.1.3. Intensity ratio vs DN and AN. [Adapted, Figure 10.1.4. PMMA surface content vs. intensity ratio. by permission, from M L Abel, M M Chehimi, [Data from M L Abel, M M Chehimi, Synthetic Metals, Synthetic Metals, 66, No.3, 225-33 (1994).] 66, No.3, 225-33 (1994).]
Thickness, angstrom
10 8 6 4
Figure 10.1.6. Schematic diagram of surface roughness of polypyrrole and deposition of PMMA from two different solvents. [Adapted, by permission, from M L Abel, J L Camalet, M M Chehimi, J F Watts, P A Zhdan, Synthetic Metals, 81, No.1, 23-31 (1996).]
2 0 0.4
0.5
0.6
0.7
0.8
0.9
1
The determination of properties of unknown system requires that master curve be Figure 10.1.5. Thickness of PMMA overlayers vs. constructed. This master curve is deterintensity ratio. [Data from M L Abel, M M Chehimi, mined by testing a series of different polySynthetic Metals, 66, No.3, 225-33 (1994).] mers exposed to a selected solvent. An example of such a relationship is given in Figure 10.1.1. The enthalpy change caused by the acid-base adduct formation is obtained from a study of the same solid with different solvent probes (see Figure 10.1.2). Having this data, coefficients E and C can be calculated from the chemical shifts in any system. Figure 10.1.3 shows that there is a correspondence between DN and AN values of different solvents and Cl(2p)/Cl(LMM) intensity ratios.7 Figure 10.1.4 shows that the type of solvent (measured by its Cl(2p)/Cl(LMM) intensity ratio) determines adsorption of basic PMMA on acidic polypyrrole. Figure 10.1.5 shows that also the thickness of PMMA overlayer corresponds to Cl(2p)/Cl(LMM) intensity ratio of solvent. Cl2p/ClLMM
10.1 General concept of acid-base interactions
569
8
-0.04
6
RT ln VN
4
-0.08 χ
2
-0.12
0 -2 -4
5
6
7
8
Number of carbon atoms Figure 10.1.7. Relationship between RT ln VN and number of carbon atoms in n-alkanes. [Adapted, by permission from M M Chehimi, E Pigois-Landureau, M Delamar, J F Watts, S N Jenkins, E M Gibson, Bull. Soc. Chim. Fr., 9(2) 137-44 (1992).]
9
-0.16 1.76
1.8
1.84 1.88 1000/T, K-1
1.92
Figure 10.1.8. Flory interaction parameter vs. inverse temperature for polyamide. [Adapted, by permission, from L Bonifaci, G Cavalca, D Frezzotti, E Malaguti, G P Ravanetti, Polymer, 33(20), 4343-6 (1992).]
XPS analysis combined with analysis by atomic force microscopy, AFM, determines differences in the surface distribution of deposited PMMA on the surface of polypyrrole depending on the type of solvent used (Figure 10.1.6).6 AFM, in this ex1735 periment, permitted the estimation of the surface roughness of polypyrrole. When PMMA was deposited from tetrahydrofuran, a poor solvent, it assumed a surface roughness equivalent to that of 1725 15 25 35 45 55 polypyrrole. Whereas PMMA, when deγd posited from a good solvent - CHCl3,6 L developed a smooth surface. The above exFigure 10.1.9. Carbonyl stretching frequency vs. dispersion contribution to surface tension of solvent. [Adapted, amples show the importance of solvents in coating morphology. by permission from M M Chehimi, E Pigois-Landureau, M Delamar, J F Watts, Figure 10.1.7 shows the correlation S N Jenkins, E M Gibson, Bull. Soc. Chim. Fr., 129(2) between the number of carbon atoms in 137-44 (1992).] n-alkanes and the net retention volume of solvent using IGC measurements.8 Such a correlation must be established to calculate the acid-base interaction’s contribution to the free energy of desorption, ∆GAB, as pointed out in discussion of equations [10.1.5] and [10.1.6]. Figure 10.1.8 shows that the Flory interaction parameter (measured by IGC) increases as the temperature increases. vC=O, cm
-1
1745
570
S. Bistac, M. Brogly
Figure 10.1.9 shows a good correlation between carbonyl stretching frequency as determined by FTIR and the dispersion contribution of surface tension of the solvent. The above shows a good correspondence between the data measured by different methods. This was also shown by a theoretical analysis of the effect of acid-base interactions on the aggregation of PMMA.14 But there is as yet no universal theory for the characterization of acid-base interaction. The following Sections show specific applications of acid-base interactions and present data on solvents applied to various phenomena in which acid base interaction is important.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
C J van Oss, L Ju, M K Chaudhury, R J Good, J. Colloid Interface Sci., 128, 313 (1989). R J Good, Contact Angle, Wetting, and Adhesion, VSP, 1993. K-X Ma, C H Ho, T-S Chung, Antec ‘99 Proceedings, 1590 and 2212, SPE, New York, 1999. C Saint Flour, E. Papirer, Colloid Interface Sci., 91, 63 (1983). K C Xing, W. Wang, H P Schreiber, Antec ‘97 Proceedings 53, SPE, Toronto, 1997. M L Abel, J L Camalet, M M Chehimi, J F Watts, P A Zhdan, Synthetic Metals, 81, No.1, 23-31 (1996). M L Abel, M M Chehimi, Synthetic Metals, 66, No.3, 225-33 (1994). M M Chehimi, E Pigois-Landureau, M Delamar, J F Watts, S N Jenkins, E M Gibson, Bull. Soc. Chim. Fr., 129(2) 137-44 (1992). J F Watts, M M Chehimi, International J. Adhesion Adhesives, 15, No.2, 91-4 (1995). P Munk, P Hattam, Q Du, A Abdel-Azim, J. Appl. Polym. Sci.: Appl. Polym. Symp., 45, 289-316 (1990). F M Fowkes, D O TIschler, J A Wolfe, L A Lannigan, C M Ademu-John, M J Halliwell, J. Polym. Sci., Polym. Chem. Ed., 22, 547 (1984). B H Stuart, D R Williams, Polymer, 36, No.22, 4209-13 (1995). R S Drago, G C Vogel, T E Needham, J. Am. Chem. Soc., 93, 6014 (1971). S J Schultz, Macromol. Chem. Phys., 198, No. 2, 531-5 (1997). L Bonifaci, G Cavalca, D Frezzotti, E Malaguti, G P Ravanetti, Polymer, 33(20), 4343-6 (1992).
10.2 EFFECT OF POLYMER/SOLVENT ACID-BASE INTERACTIONS: RELEVANCE TO THE AGGREGATION OF PMMA S. Bistac, M. Brogly Institut de Chimie des Surfaces et Interfaces ICSI - CNRS, MULHOUSE France
10.2.1 RECENT CONCEPTS IN ACID-BASE INTERACTIONS Polymer solvent interactions determine several properties, such as, solubility, solvent retention, plasticizer action, wettability, adsorption and adhesion. The solubility parameter is an important criterion for the choice of solvents. However, acid-basic characters of both solvent and polymer are also determinant parameters, which can affect the solution and final film properties. This part, devoted to the influence of acid-base interactions on the aggregation of poly(methyl methacrylate) will first present some recent concepts in acid-base interactions, followed by two practical examples based on experimental results obtained for PMMA/solvent systems.
10.2 Effect of polymer/solvent acid-base interactions
571
10.2.1.1 The nature of acid-base molecular interactions 10.2.1.1.1 The original Lewis definitions The Lewis definitions of acid-base interactions are now over a half a century old. Nevertheless they are always useful and have broadened their meaning and applications, covering concepts such as bond-formation, central atom-ligand interactions, electrophilicnucleophilic reagents, cationic-anionic reagents, charge transfer complex formation, donor-acceptor reactions, etc. In 1923 Lewis reviewed and extensively elaborated the theory of the electron-pair bond,1 which he had first proposed in 1916.2 In this small volume which had since become a classic, Lewis independently proposed both the proton and generalized solvent-system definitions of acids and bases. He wrote:
“An acid is a substance which gives off the cation or combines with the anion of the solvent; a base is a substance which gives off the anion or combines with the cation of the solvent”. The important point that Lewis revealed is that though the acid-base properties of species are obviously modified by the presence or absence of a given solvent, their ultimate cause should reside in the molecular structure of the acid or base itself, and in light of the electronic theory of matter, not in a common constituent such as H+ or OH-, but in an analogous electronic structure. He states that a basic substance is one which has a lone pair of electrons which may be used to complete the stable group of another atom (the acid) and that an acid substance is one which can employ a lone pair from another molecule (the base) in completing the stable group of one of its own atoms. Moreover by tying his definitions to the concept of chemical bond, Lewis linked their usefulness to contemporary views on the nature of the chemical bond itself. Hence, the shared electron-pair bond model explains the existence of both non-polar bond and polar link from the same premises. As the electrochemical natures of two atoms sharing an electron pair began to differ more and more, the pair should become more and more unequally shared, eventually becoming the sole property of the more electronegative atom and resulting in the formation of ions. Ionic and non-polar bonds appear as logical extremes of a continuum of intermediate bond type. Differences in the continuum are only attributable to variations in the electron-pair donation, i.e., in the way in which the charges are localized within the molecule. As a consequence the distinctions between salts, acids and bases, coordination compounds and organic compounds are not of fundamental nature. 10.2.1.1.2 Molecular Orbital (MO) approach to acid-base reactions Translated into the idiom of molecular orbital theory,3 the acid-base definitions should be read as follows: - A base is a species, which employs a doubly occupied orbital in initiating a reaction. - An acid is a species, which employs an empty orbital in initiating a reaction. The term species may mean a discrete molecule, a simple or complex ion or even a solid exhibiting non-molecularity in one or more dimensions (graphite as an example). Free atoms seldom act as Lewis acids and bases. They usually have one or more unpaired electrons and their reactions are more accurately classified as free radical. The donor orbital is usually the highest occupied molecular orbital HOMO, and the acceptor orbital is usually the lowest unoccupied molecular orbital or LUMO. The molecular orbital definitions have a number of important consequences:
572
S. Bistac, M. Brogly
First, it is not necessary that the donor and acceptor orbitals be localizable on a single atom or between two atoms, as implied by Lewis dot structures. That is, the orbitals may be multi-centered even in a relatively localized representation. Thus donor-acceptor interactions involving delocalized electron systems (π-ring)4-5 are naturally subsumed by the definitions. Second, the HOMO or donor orbital on a base is likely to be either bonding or non-bonding in character, the latter always being the case for mono-atomic species. The LUMO or acceptor orbital on an acid is likely to be either anti-bonding or non-bonding in character, the latter always being the case for mono-atomic species. Third, all degrees of electron donation are possible, ranging from essentially zero in the case of weak intermolecular forces and idealized ion associations to the complete transfer of one or more electrons from the donor to the acceptor. This continuity can be represented [10.2.1] by wave functions,6 were the degree of donation increases as the ratio (a/b).2 ψ AB = aψ A + bψ B
[10.2.1]
where: ψAB ψA ψB a, b
wave function of the acid-base one-to-one adduct ground state wave function of the acid ground state wave function of the base weighting coefficients
We have reported in Table 10.2.1 the possible adducts classified in terms of the bonding properties of the donor and acceptor orbitals of the acid and base. Complete description of the mechanisms involved during the following adducts formation can be found in Jensen7 complete review. Table 10.2.1. Possible acid-base adducts in terms of orbitals properties Acceptor orbital of the acid Non-bonding
Donor orbital of the base
Non-bonding Bonding
Anti-bonding
n
σ*
π*
n
n⋅n
n⋅σ*
n⋅π*
σ
σ⋅n
σ⋅σ*
σ⋅π*
π
π⋅n
π⋅σ*
π⋅π*
Even if the nature of an acid or a base is highly relative, a brief classification based on bonds properties and orbitals symmetry can be proposed: n donor: π donor: σ donor: n acceptor: π acceptor:
Lewis bases, complex and simple anions, carbanions, amines, oxides, sulfurs, phosphines, sulfoxides, ketones, ethers, alcohols .... unsaturated and aromatic hydrocarbons having electron donor substituents..... saturated hydrocarbons, CO, single links like C-C, C-H, polar links like NaCl, BaO, silanes.... Lewis acids, simple cations.... N2, SO2, CO2, BF3, dienic and unsaturated and aromatic hydrocarbons having electron acceptor substituents....
10.2 Effect of polymer/solvent acid-base interactions σacceptor:
573
Brönsted acids, boranes and alkanes having strong acceptor substituents, such as, CHCl3, halogens....
To summarize how acid-base reactions do work on the basis of molecular orbitals perturbation theory, we have reported on Figure 10.2.1, the relative energies (as perturbed by the field of the other reactant) of the frontier orbitals HOMO and LUMO of a hypothetical species A and of the frontier orbitals of several hypothetical reaction partners B, C, D, E and F. This figure is intended to represent possible variations of donor-acceptor properties in the broadest possible context i.e. not only those Figure 10.2.1. How do donor-acceptor molecular orbital species encountered in aqueous solution but also those stabilized by non-aqueous enviinteractions work? ronments. According to Figure 10.2.1, with respect to B, complete electron transfer from B to A will be favorable and A will act as an oxidizing agent. With respect to C, the A(LUMO) C(HOMO) perturbation will be favorable and A will act as an acid. With respect to D, the A(HOMO) - D(LUMO) perturbation will be favorable and A will act as a base. Lastly, with respect to E, complete electron transfer from A to E will be favorable and A will act as a reducing agent. For F species, the frontiers orbitals are quite degenerated with those of A. Here neither species is clearly the donor nor acceptor and species may display both behavior simultaneously (case of multisite interactions encountered in concerted organic cycloaddition reactions). 10.2.1.1.3 The case of hydrogen bonding In addition to the general discussion of Lewis acid-base interactions, hydrogen bonding represents a special case. According to Pauling, hydrogen bonding is partly covalent and partly ionic (polar). Nevertheless, it is obvious that electrostatic and charge transfer interactions are predominant for hydrogen bonds. In most cases the principal charge transfer contribution is derived from the proton acceptor - proton donor charge transfer complex through the σ-type interactions. Moreover hydrogen bonds complexes are linear and the angle between the H bond and the molecular axis of the proton acceptor are mainly 0° or 60°. To conclude this first theoretical part, the chemical phenomena subsumed by the category of acid-base reaction, are the following: • Systems defined by the Arrhenius description, solvent system, Lux-Flood and proton acid-base definitions • Traditional coordination chemistry and “non-classical” complexes • Solvation, solvolysis and ionic dissociation phenomena in both aqueous and non-aqueous solutions • Electrophilic and nucleophilic reactions in organic and organometallic chemistry • Charge transfer complexes, molecular addition compounds, weak intermolecular forces, hydrogen bonds • Molten salt phenomena, salt formation
574
S. Bistac, M. Brogly
10.2.1.2 Quantitative determination of acid-base interaction strength 10.2.1.2.1 Perturbation theory Hudson and Klopman8 proposed an equation to describe the effect of orbital perturbation of two molecules on chemical reactivity. Their hypothesis is that the initial perturbation determines the course of a reaction or an interaction. They applied quantum-mechanical method9 to treat the encounter of two interacting systems as reactivity. Their equation for interaction energy can be simplified by including only two terms [10.2.2]: the Coulombic interaction and the frontier orbital interaction between HOMO and LUMO. 2(CN CE β) QN QE + εR E HOMO − E LUMO 2
∆E int = −
[10.2.2]
where: QN QE CN CE β ε R EHOMO ELUMO
total charges of nucleophile N total charges of electrophile E coefficient of the atomic orbital N coefficient of the atomic orbital E resonance integral permittivity interatomic distance energy level of the HOMO orbital energy level of the LUMO orbital
The relative magnitudes of the numerator and denominator of the second term determine the extent of perturbation and the type of reactivity. For electrostatic interactions, the first term dominates: this is the case of highly polar acceptors of low electron affinity and donors highly electronegative, while for electron donor - electron acceptor interaction, the second term dominates. Thus, a molecular interaction encompasses chiefly both acid-base and donor-acceptor interactions. On the other hand if the second frontier orbital term dominates, the extent of perturbation is large and the charge transfer is large, leading to a large gain in stability. The frontier orbital controlled processes are favored by the presence of weakly polar species lacking high charge density but possessing reactive atoms with large orbital radii. This second term is also benefitted by bases of low electronegativity and acids of high electron affinity. 10.2.1.2.2 Hard-Soft Acid-Base (HSAB) principle The energy gap between HOMO and LUMO has been equated to the absolute hardness of the HSAB principle. This principle10 describes some basic rules about kinetics and equilibrium of the acid-base interactions in solutions. The HSAB principle will be described as it has evolved in recent years on the basis of the density-functional theory.11 For organic interactions, the following statements were proposed: • A hard acceptor has a high energy LUMO and usually a positive charge • A soft acceptor has a low energy LUMO but does not necessarily have a positive charge • A hard donor has a low energy HOMO and usually a negative charge • A soft donor has a high energy HOMO but does not necessarily have a negative charge
10.2 Effect of polymer/solvent acid-base interactions
575
The principle states that hard acid prefers to interact with a hard base, and vice versa, a soft acid with a soft base. A hard-hard interaction is fast because of large Coulombic attraction, a soft-soft interaction is fast because of large orbital overlap between HOMO and LUMO. However the problem is, what is the physical meaning of hardness? 10.2.1.2.3 Density functional theory Parr and al.12 gave in 1988 a theoretical support to the absolute hardness. In the density functional theory two basic parameters were introduced. Any chemical system can be characterized by its chemical potential, µ, and its absolute hardness, η. The chemical potential measures the escaping tendency of an electronic cloud, while absolute hardness determines the resistance of the species to lose electrons. The exact definitions of these quantities are: ∂E µ = ∂N ν
and
1 ∂ 2E η = 2 ∂N 2
ν
[10.2.3]
where : µ η E N ν
chemical potential absolute hardness electronic energy number of electrons potential due to the nuclei plus any external potential.
However, according to frontier orbital method,13 the relationship between η and the HOMO and LUMO energies is reduced to: η≈−
1 (E HOMO − E LUMO ) 2
[10.2.4]
where : η EHOMO ELUMO
absolute hardness energy level of the HOMO orbital energy level of the LUMO orbital.
Of course the absolute softness is the reciprocal of the absolute hardness. The apparent success of the density-functional theory is to provide two parameters from which we can calculate the number of electrons transferred, resulting mainly from the charge transfer between two molecules, i.e., from electrons flow until chemical potential reaches an equilibrium. As a first approximation, the number of electron transferred is given14 by: NTrans =
µ B −µ A 2 (η A − η B )
[10.2.5]
where: Ntrans µA µB ηA ηB
number of electrons transferred chemical potential of the acid chemical potential of the base absolute hardness of the acid absolute hardness of the base.
This number varies from 0 to 1 and it is in most cases a fractional number. As an example15 for the interaction between Cl2 and substituted aromatic compounds, Ntrans varies
576
S. Bistac, M. Brogly
from -0.013 to 0.20. Calculation for various solid polymers could be found in a paper of Lee.16 10.2.1.2.4 Effect of ionocity and covalency: Drago’s concept Similar to the perturbation theory, Drago and Wayland17 proposed a four-parameter equation for predicting reactions enthalpies between acid and base species. Both species are each characterized by two independent parameters: an E value which measures their ability to participate in electrostatic bonding, and a C value which measures their ability to participate in covalent bond. Both E and C values are derived empirically to give the best curve fit of calculated to experimental heats of formation for the largest possible number of adducts, leading to: −∆H AB = E A E B + CA CB
[10.2.6]
where: ∆HAB EA EB CA CB
enthalpy of acid-base adduct formation ability of the acid to participate in electrostatic bonding ability of the base to participate in electrostatic bonding ability of the acid to participate in covalent bonding ability of the base to participate in covalent bonding.
A self-consistent set of E and C values is now available18 for 33 acids and 48 bases, allowing ∆H prediction for over 1584 adducts. It is assumed that the conditions under which measurements are made (gas phase or poorly coordinating solvents) give rather constant entropy contribution and that most of adducts are of one-to-one stoichiometry. Table 10.2.2 gathers Drago’s parameters, given in (kcal/mol)0.5 of some common solvents. Table 10.2.2. Drago’s parameters of some common solvents [after reference 18] Acid
CA
EA
Phenol
0.442
4.330
Chloroform
0.159
Water
Base
CB
EB
Acetone
2.330
0.987
3.020
Benzene
0.681
0.525
2.450
0.330
Ethyl acetate
1.740
0.975
tert-Butyl alcohol
0.300
2.040
Pyridine
6.400
1.170
Iodine
1.000
1.000
Methylamine
5.880
1.300
Pyrrole
0.295
2.540
Tetrahydrofuran
4.270
0.978
Trifluoroethanol
0.451
3.88
Dimethylformamide
2.480
1.230
The major importance of the above four parameters is their relationship with the HSAB principle. Actually, through the plot of EA versus CA for several liquids on a solid, one can obtain indirectly the values of the chemical softness 1/η from the slope of C/E, as represented by the following equation : EA =−
C ∆H AB − CA B EB EB
[10.2.7]
10.2 Effect of polymer/solvent acid-base interactions
577
With the chemical hardness, η, and the electronegativity (or negative chemical potential), one can easily apply the HSAB principle to acid-base interaction and calculate the number of electron transferred. 10.2.1.2.5 Effect of amphotericity of acid-base interaction: Gutmann’s numbers Solvatation, solvolysis and ionic dissociation phenomena, in both aqueous and nonaqueous solutions are subsumed by the Lewis definitions. In addition to the previous discussion of the dual polarity character of Lewis acids and bases, it should be noted that many of them are amphoteric, by definition. Donor number, DN, was developed19 in order to correlate the behavior of a solute in a variety of donor solvents with a given basicity or donicity. A relative measurement of the basicity of a solvent D is given by the enthalpy of its reaction with an arbitrarily chosen reference acid (SbCl5 in the Gutmann’s scale). Latter Mayer20 introduced an acceptor number, AN, as the relative 31P NMR shift induced by triethylphosphine, and relative to acidic strength (AN=0 for hexane and 100 for SbCl5). In 1989, Riddle and Fowkes21 modify these AN numbers, to express them, AN*, in correct enthalpic unit (kcal/mol). Table 10.2.3 gathers electron acceptor number AN and AN* and electron donor number DN for amphoteric solvents. Table 10.2.3. Acceptor number, AN, and donor number, DN, for common solvents [After references 19, 21] Amphoteric solvent
AN
AN*, kcal/mol
DN, kcal/mol
Acetone
12.5
2.5
17.0
Diethyl ether
3.9
1.4
19.2
Formamide
39.8
9.3
26.6
Tetrahydrofuran
8.0
0.5
20.0
Pyridine
14.2
0.14
33.1
1.5
17.1
Ethyl acetate Benzonitrile
15.5
0.06
11.9
Nitromethane
20.5
4.3
2.7
Water
54.8
15.1
18.0
The most important assumption of Gutmann’s approach is that the order of base strengths established remains constant for all other acids (solutes), the value of the enthalpy of formation of a given adduct is linearly related to the donor number of the base (solvent) through the equation [10.2.8]: −∆HAB = a A DNB + b B where: ∆HAB DNB aA, bA
enthalpy of acid-base adduct formation donor number of the base constants characteristic of the acid
[10.2.8]
578
S. Bistac, M. Brogly
Graphically this means that a plot of the DN for a series of donor solvents versus -∆H of their adducts formation with a given acid gives a straight line, allowing the determination of aA and bA. By experimentally measuring the enthalpy of formation of only two adducts for a given acid, one can predict, through the resulting aA an bA values, the enthalpy of adduct formation of this acid with any other donor solvent for which DN is known. Gutmann also proposes that the enthalpy of acid-base interaction could be approximated by a two-parameters equation of the form: −∆HAB =
ANA DNB 100
[10.2.9]
where: ∆HAB DNB ANA
enthalpy of acid-base adduct formation donor number of the base acceptor number of the acid
The factor of 100 converts the AN value from a percentage of the SbCl5 value to a decimal fraction. But one had to remind that on the 171 DN values reported in the literature22 only 50 were determined precisely, i.e., calorimetrically. 10.2.1.2.6 Spectroscopic measurements: Fowkes’ approach Fowkes23 has proposed that for specific functional groups involved in acid-base interaction, the enthalpy of acid-base adduct formation is related to the infrared frequency shift, ∆ν, of its absorption band according to the following equation: ∆HAB = k AB ∆ν AB
[10.2.10]
where: ∆HAB ∆νAB kAB
enthalpy of acid-base adduct formation infrared frequency shift characteristic correlation constant between IR wavenumber shift and enthalpy
kAB is a characteristic constant of the functional group determined on the basis of compared infrared and microcalorimetrical results of adduct formation. As an example, the latter is equal to -0.99 kJ/mol/cm for the carbonyl group C=O. The stretching frequency of the C=O vibration band is decreased by an amount ∆νAB proportional to the enthalpy of acid-base bonding ∆HAB according to kAB. Such a methodology has recently been nicely confirmed not only for polymer-solvent adduction, but also for polymer/polymer24 and polymer/metal25 adduction. 10.2.2 EFFECT OF POLYMER/SOLVENT INTERACTIONS ON AGGREGATION OF STEREOREGULAR PMMA 10.2.2.1 Aggregation of stereoregular PMMA PMMA chains are able to form aggregates in the presence of solvent26 (in diluted solution, concentrated solution, gel or solid). Aggregation between isotactic and syndiotactic chains, after mixing in some solvents, leads to stereocomplexes, and self-aggregation corresponds to the aggregation of isotactic or syndiotactic chains together. These two kinds of aggregates result from the development of physical interactions between polymer chains. The formation of aggregates depends mainly on the PMMA degree of stereoregularity and on the nature of the solvent, but also on the temperature, the mixing time, and the isotactic/syndiotactic stoichiometric ratio for stereocomplexes.26,27 Complexing solvents fa-
10.2 Effect of polymer/solvent acid-base interactions
579
vor the formation of aggregates and non-complexing solvents hinder the formation of aggregates. Table 10.2.4 presents the solvents classification as a function of complexing power for PMMA stereocomplexes formation. Table 10.2.4. Solvents classification as a function of complexing power for PMMA stereocomplexes formation (after reference 26) Strongly complexing solvents
CCl4, acetonitrile, DMF, DMSO, THF, toluene, acetone
Weakly complexing solvents
benzene, o-dichlorobenzene, dioxane
Non-complexing solvents
chloroform, dichloromethane
Several works have shown that the aggregation of isotactic and syndiotactic chains leads to the formation of stereocomplexes for which the iso/syndio stoichiometry is found equal to 1/2,28,29 probably with a structure composed of a double-stranded helix of a 30/4 helicoidal isotactic chain surrounded by a 60/4 helicoidal syndiotactic chain.30 Syndiotactic PMMA self-aggregates exhibit similar structures, with conformations close to extended chains.31 Experimental data indicate that, in self-aggregated syndiotactic PMMA in solution, some of the ester groups are close in contact, probably in a double helix structure32 with solvent molecules included in the cavities of inner- and inter-helices.33 Isotactic PMMA self-aggregates also exhibit conformational helix structures. These results prove that the presence of a complexing solvent leads to the formation of well-ordered structure made of paired PMMA chains. A high degree of stereoregularity is needed to allow the formation of aggregates. Syndiotactic and isotactic sequences lengths have to be larger than a critical length in order to be involved in paired association. As the aggregation power of the solvent increases, the critical sequence length decreases. Many authors explain aggregation as resulting from interactions between PMMA chains, probably through their ester functional groups.34 The incidence of the nature of the solvent in the aggregate formation is not yet totally clarified in the literature. Apparently, there is no obvious relationship between the solvent polarity and the ability to induce aggregation of PMMA chains. 10.2.2.2 Relation between the complexing power of solvents and their acid-base properties A tentative explanation of the effect of solvent has been proposed recently by using the Lewis acid-base concept.35 The acid or basic character of a given solvent was studied simultaneous with its strength of complexation in order to establish a relationship between both parameters. PMMA stereocomplexes and self-aggregates exhibit some similar aspects, particularly the complexing power of the solvent or the helical structure of paired chains. PMMA is classified as a basic polymer (electron donor, according to the Lewis concept), due to the presence of ester functional groups, where the carbonyl oxygen atom is the basic site.36 PMMA is therefore able to exchange strong acid-base interactions with an acidic solvent, such as, chloroform, which is also a non-complexing solvent. According to Fowkes works, PMMA and chloroform can form acid-base complexes, resulting from the
580
S. Bistac, M. Brogly
interaction between the ester basic group of PMMA and the hydrogen acid atom of chloroform. According to Drago’s classification, it appears therefore in Table 10.2.2 that chloroform exhibits an acid character.37 PMMA presents a basic character, with Eb=0.68 and Cb=0.96. The fact that an acidic solvent, such as, chloroform is able to exchange strong acid-base interactions with PMMA ester groups is able to explain its non-complexing behavior. Strong acid-base interactions between PMMA and chloroform molecules hinder the aggregation of PMMA chains, reflecting greater PMMA/solvent interactions compared to PMMA/PMMA interactions. In the presence of basic solvents, the lower PMMA/solvent acid-base interactions lead to inter-chains associations (and consequently aggregation), energetically favorable. Moreover, solvents, such as, acetone or THF possess a high degree of self-association (31%, 27% respectively) and Fowkes has shown that the contribution of solvent/solvent acid-base interactions to the heat of vaporization is proportional to the degree of self-association.38 Acetone and THF are therefore strongly acid-base self-associated solvents. As a consequence, they would develop preferentially self-associations (between solvent molecules) rather than PMMA/solvent interactions. Hence the PMMA chains aggregation would be also favored. Gutmann’s numbers of complexing and Table 10.2.5. AN and DN numbers for non-complexing solvents are listed in Table complexing and non-complexing 10.2.5.The results indicate also that chloroform is solvents (after reference 19) an acidic solvent, with a high Acceptor Number (acidic character), and a Donor Number (basic AN DN, kcal/mol character) equal to zero. The advantage of DMF 16.0 26.6 Gutmann’s approach is to consider the potential amphoteric character. As an example, benzene, Acetone 12.5 17.0 which is classified as a basic solvent by Drago, THF 8.0 20 exhibits an amphoteric character with a low AN, but also a DN close to zero. This result could exDMSO 20.4 29.8 plain that, in some cases, benzene (which is genBenzene 8.2 0.1 erally a complexing solvent) is described as a Chloroform 23.1 0 non-complexing solvent for syndiotactic PMMA.31 In this case, amphoteric character of benzene favors acid-base interactions with PMMA. To resume this part, it is possible to correlate the complexing power of solvents towards stereoregular PMMA with their acid-base character. It appears that complexing solvents exhibit basic character, like PMMA. Moreover, some of them are strongly self-associated. Solvent/solvent and chain/chain interactions are consequently favored, leading to the formation of PMMA aggregates. On the contrary, chloroform is an acidic solvent, and the development of acid-base interactions with PMMA ester groups helps to hinder the formation of chains aggregates. However, this original view on the relationship between complexing power and solvent acid-base character does not take into account the way of action of the solvent in terms of steric effect and macromolecular conformations of aggregates.
10.2 Effect of polymer/solvent acid-base interactions
581
10.2.3 INFLUENCE OF THE NATURE OF THE SOLVENT ON THE α AND β-RELAXATIONS OF CONVENTIONAL PMMA 10.2.3.1 Introduction Thin polymer films are generally obtained by solution casting, in many applications, such as paints, varnishes, or adhesives. The properties of polymer films obtained from a solution differ from the original bulk properties, and this effect can have some significant consequences on the expected behavior of the final film. Recent studies have analyzed the influence of the nature of the solvent on the relaxation temperatures of conventional PMMA solid films.39,40 PMMA was dissolved in various solvents and the solid films (after solvent evaporation) were analyzed by dielectric spectroscopy. The purpose of the study was to investigate the influence of the nature of the solvent on the α and β-relaxations of PMMA solid films. The α-relaxation is related to the glass transition of PMMA and corresponds to the rotation of lateral groups around the main chain axis.41 The β-relaxation, which occurs at a lower temperature, is induced by the rotation of the acrylate groups around the C-C bonding which links them to the main chain. Different solutions of conventional (atactic) PMMA in good solvents (chloroform, acetone, toluene and tetrahydrofuran) are cast on a metallic substrate (aluminum). After solvent evaporation at room temperature, solid films of PMMA are analyzed by dielectric spectroscopy (DETA) at 100 Hz and 10 KHz from -40 to 135°C with a scanning temperature of 2°C/min. The variation of the loss factor tan δ is studied as function of temperature. The temperatures of the tan δ peaks are related to the relaxations temperatures of the polymer.42 The reference DETA spectrum is obtained by analyzing a bulk PMMA film obtained from heating press (without solvent). At 10 KHz, only one peak is detectable (whatever the sample) which is attributed to the α-relaxation. However, the peak is not symmetrical, with a broadening towards the lower temperatures, due to the contribution of the β-relaxation. At 100 Hz, two peaks are present, the major at lower temperature, corresponding to the β-relaxation and the minor at higher temperature, attributed to the α-relaxation. 10.2.3.2 Dielectric spectroscopy results Table 10.2.6 reports the temperatures of the α-relaxation (measured at 10 KHz) and the β-relaxation (measured at 100 Hz) for the reference sample (bulk PMMA) and the different solution-cast films. Table 10.2.6. Temperature of α and β-relaxations of bulk PMMA and solution-cast films (after reference 40) PMMA
PMMA/
PMMA/
PMMA/
PMMA/
bulk
chloroform
toluene
acetone
THF
T° of α peak, °C at 10KHz
108
75
96
85
82
T° of β peak, °C at 100 Hz
40
34
49
45
47
The results show that the temperature of the α-transition is significantly reduced for cast films compared to the bulk polymer, with a lower value for chloroform samples. The decrease of the α-transition temperature can be explained by the presence of residual solvent, inducing a plasticizing effect. On the contrary, an increase of the β-transition tempera-
582
S. Bistac, M. Brogly
ture is generally observed for the cast films, except for the ‘chloroform’ sample. An increase of the β-transition temperature indicates a lower mobility of the acrylate groups compared to the bulk polymer. However, for chloroform sample, this mobility is increased. Two different effects appear and two groups of solvents can then be distinguished: the first group, with chloroform, which induces a higher mobility o f both the main chain and the acrylate group (compared to the bulk polymer), and the second group, including acetone, toluene and THF which induce also an increase of the mobility of the main chain, but a decrease of the mobility of the lateral acrylate groups. The influence of the nature of the solvent on the relaxation temperature can be explained by analyzing the acid-base properties of the polymer and the solvents. PMMA, which is a basic polymer can exchange strong acid-base interactions with an acidic solvent, such as, chloroform. In the solid cast films, acid-base interactions between acrylate groups and residual chloroform occur: some acrylate groups interact with chloroform molecules and PMMA/PMMA self-associations partially disappear, leading to a lower value of the β-transition temperature. The mobility of the lateral acrylate groups interacting with chloroform molecules is then higher than the mobility of acrylate groups interacting with other acrylate groups as in bulk PMMA. Toluene, THF and acetone are described as basic solvents. The residual solvent molecules can therefore only weakly interact with the acrylate groups. Interactions between chains are then favored and self-associations between PMMA chains appear. The rotation of the acrylate groups becomes more difficult compared to the bulk PMMA, probably due to self-aggregation of some PMMA chains (even if the studied polymer is not stereoregular). To resume, residual solvent molecules present in solid conventional PMMA films are able to significantly modify the polymer relaxation properties. The effect of residual solvent depends strongly on the nature of the solvent, specially its acid-base character. 10.2.4 CONCLUDING REMARKS
These works have shown that acid-basic character of solvent is able to have a major influence on polymer film properties. Acid-base interactions between stereoregular PMMA and some solvents can lead to the formation of aggregates, which modify the solution properties. Elsewhere, residual solvent molecules trapped in solid conventional PMMA films have an effect on the polymer chains mobility, depending on the acid-base character of the solvent. Both solvent effects, due to acid-base interactions, are able to modify the wetting and the adhesion properties of the films, but also the mechanical and the durability behavior of the final film. It is then necessary to take into account, in the choice of a solvent, its acid-base properties, in addition to its solubility parameter, especially for PMMA. REFERENCES 1 2 3 4 5 6 7 8 9 10
G.N. Lewis, Valence and the structure ofatoms and molecules, The Chemical Catalog Co., N-Y, 1923. G.N. Lewis, J. Am. Chem. Soc., 38, 762 (1916). R.S. Mulliken, W.B. Pearson, Molecular complexes : a lecture and reprint volume Wiley-Interscience, N-Y, 1969. M. Brogly, M. Nardin, I. Schultz, J. Adhesion, 58, 263 (1996). M.F. Hawthorne, G.B. Dunks, Science, 178, 462 (1972). G. Klopman, Chemical reactivity and Reaction paths, Wiley-Intersciences, New-York, 1974. W.B. Jensen The Lewis acid-base concepts: an overview J.Wiley & Sons, New-York, 1979. R.F. Hudson, G. Klopman, Tetrahedron Lett., 12, 1103 (1967). S.R. Cain, Acid-base interactions : relevance to adhesion science and technology, K.L. Mittal and H. Anderson, Jr (Eds), VSP, Zeist, The Netherlands, 1991. R.G. Pearson, J. Am Chem. Soc., 85, 3533 (1963).
10.3 Solvent effects based on pure solvent scales
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
583
R.G. Parr, W. Yang in Density functional theory of atoms and molecules, Oxford University Press, New-York, 1989. M. Berkowitz and R.G. Parr, J. Chem. Phys., 88, 2554 (1988). I. Fleming, Frontier orbitals and organic chemical reactions, John Wiley, London, 1976. S. Shankar, R.G. Parr, Proc. Natl. Acad. Sci., USA, 82, 264 (1985). R.G. Pearson, J. Org. Chem., 54, 1423 (1989). L.H. Lee, J. Adhesion Sci. Technol., 5, 71 (1991). R.S. Drago, B. Wayland, J. Am. Chem. Soc., 87, 3571 (1965). R.S. Drago, Struct. Bonding, 15, 73 (1973). V. Gutmann, The donor-acceptor approach to molecular interaction, Plenum Press, New-York, 1977. U. Mayer, V. Gutmann, W. Greger, Montatsh. Chem., 106, 1235 (1975). F.L. Riddle, Jr, F.M. Fowkes, J. Am. Chem. Soc., 112, 3259 (1990). Y. Marcus, J. Solution Chem., 13, 599, (1984). F.M. Fowkes et al., J. Polym. Sci. Polym. Chem. Ed., 22, 547 (1984). M. Brogly, M. Nardin, J. Schultz, Polymer, 39, 2185 (1998). M.Brogly, S. Bistac, J. Schultz, Macromolecules, 31, 3967 (1998). J. Spevacek, B. Schneider, Adv. Coll. Interf. Sci., 27, 81 (1987). M. Berghmans, S. Thijs, M. Cornette, H. Berghmans, F.C. De Schryver, P. Moldenaers, J. Mewis, Macromolecules, 27, 7669 (1994). K. Ohara, Coll. Polym. Sci., 259, 981 (1981). K. Ohara, G. Rehage, Coll. Polym. Sci., 259, 318 (1981). F. Bosscher, G. Ten Brinke, G. Challa, Macromolecules, 15, 1442 (1982). J. Spevacek, B. Schneider, J. Dybal, J. Stokr, J. Baldrian, Z. Pelzbauer, J. Polym. Sci. : Polym. Phys. Ed., 22, 617 (1984). J. Dybal, J. Spevacek, B. Schneider, J. Polym. Sci. : Polym. Phys. Ed., 24, 657 (1986). H. Kusuyama, N. Miyamoto, Y. Chatani, H. Tadokoro, Polym. Comm., 24, 119 (1983). J. Spevacek, B. Schneider, Colloid Polym. Sci., 258, 621 (1980). S. Bistac, J. Schultz, Macromol. Chem. Phys., 198, 531 (1997). F.M. Fowkes, J. Adhesion Sci. Tech., 1, 7 (1987). R.S. Drago, G.C. Vogel, T.E. Needham, J. Amer. Chem. Soc., 93, 6014 (1971). F.M. Fowkes in Acid-Base Interactions, K.L. Mittal & H.R. Anderson, Ed, VSP, Utrecht 1991, pp. 93-115. S. Bistac, J. Schultz, Prog. in Org. Coat., 31, 347, (1997). S. Bistac, J. Schultz, Int. J. Adhesion and Adhesives, 17, 197 (1997). R.F. Boyer, Rubb. Chem. Techn., 36, 1303 (1982). N.G. Mc Crum, B.E. Read, G. Williams, Anelastic and Dielectric Effect in Polymeric Solids, Dover Publications, New York, 1991.
10.3 SOLVENT EFFECTS BASED ON PURE SOLVENT SCALES Javier Catalán Departamento de Química Fisíca Aplicada Universidad Autónoma de Madrid, Madrid, Spain
INTRODUCTION The solvent where a physico-chemical process takes place is a non-inert medium that plays prominent roles in chemistry. It is well-known1 that, for example, a small change in the nature of the solvent can alter the rate of a reaction, shift the position of a chemical equilibrium, modify the energy and intensity of transitions induced by electromagnetic radiation or cause protein denaturation. As a result, the possibility of describing the properties of solvents in terms of accurate models has aroused the interest of chemists for a long time. The interest of chemists in this topic originated in two findings reported more than a century ago that exposed the influence of solvents on the rate of esterification of acetic acid
584
Javier Catalán
by ethanol, established in 1862 by Berthelot and Saint-Gilles,2 and on the rate of quaternization of tertiary amines by alkyl halides, discovered in 1890 by Menschutkin.3 In his study, Menschutkin found that even so-called “inert solvents” had strong effects on the reaction rate and that the rate increased by a factor about 700 from hexane to acetophenone. Subsequent kinetic studies have revealed even higher sensitivity of the reaction rate to the solvent. Thus, the solvolysis rate of tert-butyl chloride increases 340 000 times from pure ethanol to a 50:50 v/v mixture of this alcohol and water,4,5 and by a factor of 2.88×1014 from pentane to water.6 Also, the decarboxylation rate of 6-nitrobenzisoxazol 3-carboxylate increases by a factor of 9.5×107 from water to HMPT.7 10.3.1 THE SOLVENT EFFECT AND ITS DISSECTION INTO GENERAL AND SPECIFIC CONTRIBUTIONS Rationalizing the behavior of a solvent in a global manner, i.e., in terms of a single empirical parameter derived from a single environmental probe, appears to be inappropriate because the magnitude of such a parameter would be so sensitive to the nature of the probe that the parameter would lack predictive ability. One must therefore avoid any descriptions based on a single term encompassing every potential interaction of the solvent, often concealed under a global concept called “solvent polarity”. One immediate way of dissecting the solvent effect is by splitting it into general (non-specific) interactions and specific interactions. In relation to general interactions, the solvent is assumed to be a dielectric continuum. The earliest models for this type of interaction were developed by Kirkwood8 and Onsager,9 and were later modified with corrections for the effect of electrostatic saturation.10,11 The intrinsic difficulty of these models in accurately determining the dimensions of the cybotactic region (viz. the solvent region where solvent molecules are directly perturbed by the presence of solute molecule) that surrounds each solute molecule in the bulk solvent, have usually raised a need for empirical approximations to the determination of a parameter encompassing solvent polarity and polarizability. An alternative approach to the general effect was recently reported that was derived from liquid-state theories. One case in point is the recent paper by Matyushov et al.,12 who performed a theoretical thermodynamic analysis of the solvent-induced shifts in the UV-Vis spectra for chromophores; specifically, they studied p-nitroanisole and the pyridinium-N-phenoxide betaine dye using molecular theories based on long-range solute-solvent interactions due to inductive, dispersive and dipole-dipole forces. A number of general empirical solvent scales have been reported;1,13-15 according to Drago,14 their marked mutual divergences are good proof that they do not reflect general effects alone but also specific effects of variable nature depending on the particular probe used to construct each scale. In any case, there have been two attempts at establishing a pure solvent general scale over the last decade. Thus, in 1992, Drago14 developed the “unified solvent polarity scale”, also called the “S' scale” by using a least-squares minimization program16 to fit a series of physico-chemical properties (χ) for systems where specific interactions with the solvents were excluded to the equation ∆χ = PS' + W. In 1995, our group reported the solvent polarity-polarizability (SPP) scale, based on UV-Vis measurements of the 2-N,N-dimethyl-7-nitrofluorene/2-fluoro-7-nitrofluorene probe/homomorph pair.15 According to Drago,17 specific interactions can be described as localized donor-acceptor interactions involving specific orbitals in terms of two parameters, viz. E for electrostatic interactions and C for covalent interactions; according to Kamlet and Taft,18a such interactions can be described in terms of hydrogen-bonding acid-base interactions. In fact,
10.3 Solvent effects based on pure solvent scales
585
ever since Lewis unified the acidity and basicity concepts in 1923,19 it has been a constant challenge for chemists to find a single quantifiable property of solvents that could serve as a general basicity indicator. Specially significant among the attempts at finding one are the donor number (DN) of Gutmann et al.,20 the B(MeOD) parameter of Koppel and Palm,21 the pure base calorimetric data of Arnet et al.22 and parameter β of Kamlet and Taft.18b 10.3.2 CHARACTERIZATION OF A MOLECULAR ENVIRONMENT WITH THE AID OF THE PROBE/HOMOMORPH MODEL As a rule, a good solvent probe must possess two energy states such that the energy or intensity of the transition between them will be highly sensitive to the nature of the environment. The transition concerned must thus take place between two states affected in a different manner by the molecular environment within the measurement time scale so that the transition will be strongly modified by a change in the nature of the solvent. For easier quantification, the transition should not overlap with any others of the probe, nor should its spectral profile change over the solvent range of interest. Special care should also be exercised so that the probe chosen will be subject to no structural changes dependent on the nature of the solvent; otherwise, the transition will include this perturbation, which is external to the pure solvent effect to be quantified. All probes that meet the previous requirements are not necessarily good solvent probes, however; in fact, the sensitivity of the probe may be the result of various types of interaction with the solvent and the results difficult to generalize as a consequence. A probe suitable for determining a solvent effect such as polarity, acidity or basicity should be highly sensitive to the interaction concerned but scarcely responsive to other interactions so that any unwanted contributions will be negligible. However, constructing a pure solvent scale also entails offsetting these side effects, which, as shown later on, raises the need to use a homomorph of the probe. The use of a spectroscopic technique (specifically, UV-Vis absorption spectrophotometry) to quantify the solvent effect provides doubtless advantages. Thus, the solute is in its electronic ground state, in thermodynamic equilibrium with its environment, so the transition is vertical and the solvation sphere remains unchanged throughout. These advantages make designing a good environmental probe for any of the previous three effects quite easy. A suitable probe for the general solvent effect must therefore be a polar compound. Around its dipole moment, the solvent molecules will arrange themselves as effectively as possible -in thermal equilibrium- and the interaction between the probe dipole and the solvent molecules that form the cybotactic region must be strongly altered by electronic excitation, which will result in an appropriate shift in charge; the charge will then create a new dipole moment enclosed by the same cybotactic cavity as in the initial state of the transition. The change in the dipole moment of the probe will cause a shift in the electronic transition and reflect the sensitivity of the probe to the solvent polarity and polarizability. A high sensitivity in the solvent is thus usually associated with a large change in the dipole moment by effect of the electronic transition. However, the magnitude of the spectral shift depends not only on the modulus of the dipole moment but also on the potential orientation change. In fact, inappropriate orientation changes are among the sources of contamination of polarity probes with specific effects. For simplicity, orientation changes can be reduced to the following: (a) the dipole moment for the excited state is at a small angle to that for the ground state of the probe, so the orientation change induced by the electronic excitation can
586
Javier Catalán
be neglected; (b) the two dipole moments are at an angle near 180o to each other; and (c) the two dipole moments form an angle in between the previous two. Any transitions where the dipole moments for the states involved in the transition are at angles markedly different from 0 or 180o should be avoided as the Scheme I. spectral shift includes contributions that depend on the cosine of the angle, which hinders its use in practice (see Scheme Ia). This difficulty vanishes at 0 and 180o, which are thus the preferred choices. A transition giving rise to an orientation change by 180o is usually one involving a molecule where charge is strongly localized at a given site and is discharged by delocalization upon electronic excitation; as a result, specific solvating effects in the initial state are strongly altered in the final state, so the transition concerned is strongly contaminated with specific interactions (see Scheme Ib). This situation is also inadvisable with a view to constructing a scale for the general solvent effect. The third possibility, where the transition causes no orientation change in the dipole moment, is the most suitable with a view to establishing the general effect of a solvent; the solvent box, which was the most suitable for solvating the probe, will continue to be so now with an increased dipole moment (see Scheme Ic). In summary, a suitable probe for estimating the general solvent effect must not only meet the above-described requirements but also be able to undergo a transition the dipole moment of which will increase markedly as a result while preserving its orientation. A suitable basicity probe will be one possessing an acid group the acidity of which changes markedly with the electronic transition, so that the transition is affected mostly by solvent basicity. For its acidity to change to an appropriate extent upon excitation, the group concerned should resound with an appropriate electron acceptor and the resonance should increase during the transition by which the effect is probed. Obviously, the presence of this type of group endows the probe with a polar character; in addition, the charge transfer induced by the excitation also produces side effects that must be subtracted in order to obtain measurements contaminated with no extraneous effects such as those due to polarity changes or the acidity of the medium described above. Finding a non-polar probe with a site capable of inducing a specific orientation in the solvent appears to be a difficult task. The best choice in order to be able to subtract the above-mentioned effects appears to be an identical molecular structure exhibiting the same transition but having the acid group blocked. One interesting alternative in this context is the use of a pyrrole N-H group as probe and the corresponding N-Me derivative as homomorph.23 The same approach can be used to design a suitable acidity probe. The previous comments on the basicity probe also apply here; however, there is the added difficulty that the basic site used to probe solvent basicity cannot be blocked so easily. The situation is made much more complex by the fact that, for example, if the basic site used is a ketone group or a pyridine-like nitrogen, then no homomorph can be obtained by blocking the site. The solu-
10.3 Solvent effects based on pure solvent scales
587
tion, as shown below, involves sterically hindering the solvent approach by protecting the basic site of the probe with bulky alkyl (e.g., tert-butyl) groups at adjacent positions.24 One can therefore conclude that constructing an appropriate general or specific solvent scale entails finding a suitable molecular probe to preferentially assess the effect considered and a suitable homomorph to subtract any side effects extraneous to the interaction of interest. Any other type of semi-empirical evaluation will inevitably lead to scales contaminated with other solvent effects, as shown later on. Accordingly, one will hardly be able to provide a global description of solvent effects on the basis of a single-parameter scale. 10.3.3 SINGLE-PARAMETER SOLVENT SCALES: THE Y, G, ET(30), Py , Z, χR, Φ, AND S' SCALES Below are briefly described the most widely used single-parameter solvent scales, with special emphasis on their foundation and use. 10.3.3.1 The solvent ionizing power scale or Y scale In 1948, Grunwald and Winstein25 introduced the concept of “ionizing power of the solvent”, Y, based on the strong influence of the solvent on the solvolysis rate of alkyl Scheme II. halides in general and tert-butyl chloride in particular (see Scheme II). Y is calculated from the following equation: Y = log k tBuCl − log k 0tBuCl
[10.3.1]
where ktBuCl and k0tBuCl are the solvolysis rate constants at 25oC for tert-butyl chloride in the solvent concerned and in an 80% v/v ethanol/water mixture - the latter constant is used as reference for the process. The strong solvent dependence of the solvolysis rate of tert-butyl chloride was examined by Grunwald and Winstein25,26 in the light of the Brönsted equation. They found the logarithmic coefficient of activity for the reactant and transition state to vary linearly in a series of mixtures and the variation to be largely the result of changes in coefficient of activity for the reactant. By contrast, in the more poorly ionizing solvents, changes in k were found to be primarily due to changes in coefficient of activity for the transition state. In a series of papers,25-32 Grunwald and Winstein showed that the solvolysis rate constants for organic halides which exhibit values differing by more than 6 orders of magnitude in this parameter can generally be accurately described by the following equation: log k = mY + log k 0
[10.3.2]
where k and k0 are the rate constants in the solvent concerned and in the 80:20 v/v ethanol/water mixture, and coefficient m denotes the ease of solvolysis of the halide concerned relative to tert-butyl chloride. By grouping logarithms in eq. [10.3.2], one obtains log(k / k 0 ) = mY
[10.3.3]
which is analogous to the Hammett equation:33 log(k / k 0 ) = ρσ
[10.3.4]
588
Javier Catalán
There is thus correspondence between coefficient m and the Hammett reaction constant, ρ, and also between the ionizing power of the solvent, Y, and the Hammett substituent constant, σ. In fact, the solvolysis of tert-butyl chloride is one of the cornerstones of physical organic chemistry.34-36 Thus, some quantitative approaches to the kinetics of spontaneous reactions in various solvents -and, more interesting, solvent mixtures- are based on linear free-energy relations such as that of Grunwald and Winstein25 or its extensions.34,37-41 These equations allow one to interpolate or extrapolate rate constants that cannot be readily measured, and also to derive mechanistically significant information in the process. 10.3.3.2 The G values of Allerhand and Schleyer Allerhand and Schleyer42 found that a proportionality exists between the stretching frequencies of vibrators of the type X=O (with X = C, N, P or S) and the corresponding stretching frequencies in situations involving hydrogen bonds of the X−HLB type in a variety of solvents. They thought these results to be indicative that the solvents studied interacted in a non-specific manner with both types of vibrator, in contradiction with the specific interaction-only approach advocated by Bellamy et al.43-45 They used this information to construct a solvent polarity scale that they called “the G scale”. Allerhand and Schleyer42 used an empirical linear free-energy equation to define G:
(
)
G = v 0 − v s / av 0
[10.3.5]
where v0 and vs are the corresponding stretching frequencies for one such vibrator in the gas phase and in solution, respectively; a is a function of the particular vibrator in a given probe and also a measure of its sensitivity to the solvent; and G is a function of the solvent alone. The scale was initially constructed from 21 solvents; zero was assigned to the gas phase and 100 to dichloromethane. Subsequently, other authors established the G values for additional solvents.46,47 10.3.3.3 The ET(30) scale of Dimroth and Reichardt The ET(30) scale48 is based on the extremely solvatochromic character of 2,6-diphenyl-4-(2,4,6triphenyl-1-pyrido)phenoxide (1) and is defined by the position (in kcal mol-1) of the maximum of the first absorption band for this dye, which has marked intramolecular charge-transfer connotations and gives rise to an excited electronic state that is much less dipolar than the ground state(see scheme III). This results in strong hypsochromism when increasing solvent polarity. On this ET(30) scale, pure solvents take values from 30.7 kcal mol-1 for tetramethylsilane (TMS) to 63.1 kcal mol-1 for water. These data allow one to easily normal-
10.3 Solvent effects based on pure solvent scales
589
Scheme III.
ize the scale by assigning a zero value to tetramethylsilane and a unity value to water, using the following expression: ETN = (ET(solvent) - ET(TMS)) / (ET(water) - ET(TMS)) = (ET(solvent) -30.7) / 32.4
[10.3.6]
The corresponding ETN values (between 0 and 1) allow one to rank all solvents studied. This is no doubt the most comprehensive solvent scale (it encompasses more than 300 solvents) and also the most widely used at present.1,13 The probe (1) exhibits solubility problems in non-polar solvents that the authors have overcome by using a tert-butyl derivative (2). One serious hindrance to the use of this type of probe is its high basicity (pKa = 8.6449), which raises problems with acid solvents. 10.3.3.4 The Py scale of Dong and Winnick The Py scale50 is based on the ratio between the intensity of components (0,0) I1 and (0,2) I3 of the fluorescence of monomeric pyrene (3) in various solvents. It was initially established from 95 solvents50 and spans values from 0.41 for the gas phase to 1.95 for DMSO. This scale is primarily used in biochemical studies, which usually involve fluorescent probes. However, it poses problems arising largely from the difficulty of obtaining precise values of the above-mentioned intensity ratio; this has resulted in divergences among Py values determined by different laboratories.51 One further hindrance is that the mechanism via which low-polar solvents enhance the intensity of symmetry-forbidden vibronic transitions through a reduction in local symmetry is poorly understood.52 10.3.3.5 The Z scale of Kosower The Z scale53,54 is based on the strong solvatochromism of the 1-ethyl-4(methoxycarbonyl)pyridinium iodide ion-pair (4) and defined by the position (in kcal mol-1) of the maximum of its first absorption band, which has marked intermolecular charge-transfer connotations according to Scheme IV (the excited state of the chromophore is much less
590
Javier Catalán
dipolar). This transition undergoes a strong hypsochromic shift as solvent polarity increases, so much so that it occasionally overlaps with strong π → π * bands and results in imprecise localization of the maximum of the charge-transfer band. The Z values for the 20 solvents originally examined by Kosower53 spanned the range from Scheme IV. 64.2 for dichloromethane to 94.6 for water. The scale was subsequently expanded to an overall 61 solvents by Marcus56 and the original range extended to 55.3 kcal mol-1 (for 2-methyltetrahydrofuran). Further expansion of this scale was precluded by the fact that high-polar solvents shift the charge-transfer band at the shortest wavelength to such an extent that it appears above the strong first π → π * transition of the compound, thus hindering measurement; in addition, the probe (4) is scarcely soluble in non-polar solvents. One should also bear in mind that many solvents require using a high concentration of the probe in order to obtain a measurable charge-transfer band; as a result, the position of the band often depends on the probe concentration. One other fact to be considered is that the interest initially aroused by this scale promoted attempts at overcoming the above-mentioned measurement problems by using correlations with other solvent-sensitive processes; as a result, many of the Z values currently in use are not actually measured values but extrapolated values derived from previously established ratios. 10.3.3.6 The χR scale of Brooker In 1951, Brooker57 suggested for the first time that solvatochromic dyes could be used to obtain measures of solvent polarity. This author58 constructed the χR scale on the basis of the solvatochromism of the merocyanine dye (5), the electronic transition of which gives rise to a charge-transfer from the amine nitrogen to a carboxamide group at the other end of the molecule. Hence, the excited status is more dipolar than the ground state, and the resulting band is shifted bathochromically as solvent polarity increases. χR values reflect the position of the maximum of the first band for the chromophore in kcal mol-1. The original scale encompassed 58 solvents spanning χR values from 33.6 kcal mol-1 for m-cresol to 50.9 kcal mol-1 for n-heptane. 10.3.3.7 The Φ scale of Dubois and Bienvenüe Dubois and Bienvenüe59 developed the Φ polarity solvent scale on the basis of the position of the n → π* transition for eight selected aliphatic ketones that were studied in 23 solvents, using n-hexane as reference and the following equation for calculation:
10.3 Solvent effects based on pure solvent scales
(
)
∆v SH = v S − v H = Φ v H − 32637 − 174
591
[10.3.7]
where vS is the absorption wavenumber in solvent S and vH in n-hexane (the reference solvent). The absorption maxima of these ketones depend both on the solvent and on their own structure. The Φ values spanned by the 23 solvents studied range from -0.01 for carbon tetrachloride to 0.65 for formic acid. The values for DMSO and water are 0.115 and 0.545, respectively. 10.3.3.8 The S' scale of Drago Drago60 developed a “universal polarity scale” (the S' scale) from more than three hundred spectral data (electronic transitions, 19F and 15N chemical shifts and RSE coupling constants) for 30 solutes in 31 non-protic solvents from cyclohexane to propylene carbonate. He used the equation ∆χ = S ′P + W
[10.3.8]
where ∆χ is the measured physico-chemical property, S' the solvent polarity, P the solvating susceptibility of the solute and W the value of ∆χ at S' = 0. Drago assigned an arbitrary value of 3.00 to DMSO in order to anchor the scale. The S' scale was constructed from carefully selected data. Thus, it excluded data for (a) all systems involving any contribution from donor-acceptor specific interactions (donor molecules where only measured in donor solvents and data for πsolutes were excluded); (b) concentrated solutions of polar molecules in non-polar solvents (which might result in clustering); and (c) polar solvents occurring as rotamers (each rotamer would be solvated in a different way). In order to extend the S' scale to acid solvents, Drago61 devised an experiment involving separating general interactions from specific interactions using the ET(30) scale. First, the probe [ET(30)] was dissolved in a non-coordinating solvent of weakly basic character -so that it would not compete with the solute for the specific interaction- and slightly polar nature -so that the probe would not cluster at low concentrations of the acceptor solvent-; then, the acceptor solvent was added and the variation of the maximum on the absorption band for the ET(30) probe was plotted as a function of the solvent concentration. By extrapolation to dilute solutions, the non-specific solvation component for the solvent was estimated. An interesting discussion of the solvent scales developed by Drago can be found in his book.62 10.3.4 SOLVENT POLARITY: THE SPP SCALE15 As noted earlier, a suitable probe for assessing solvent polarity should meet various requirements including the following: (a) the modulus of its dipole moment should increase markedly but its orientation remain unaltered in response to electronic excitation: (b) it should undergo no structural changes by effect of electronic excitation or the nature of the solvent; (c) its basicity or acidity should change as little as possible upon electronic excitation so that any changes will be negligible compared to those caused by polarity; (d) the spectral envelope of the electronic transition band used should not change with the nature of the solvent; and (e) its molecular structure should facilitate the construction of a homomorph allowing
592
Javier Catalán
one to offset spurious contributions to the measurements arising from the causes cited in (b) to (d) above. A comprehensive analysis of the literature on the choice of probes for constructing polarity scales led our group to consider the molecular structure of 2-amino-7-nitrofluorene (ANF) (6) for this purpose. This compound had previously been used by Lippert63 to define his well-known equation, which relates the Stokes shift of the chromophore with the change in its dipole moment on passing from the ground state to the excited state. He concluded that, in the first excited state, the dipole moment increased by 18 D on the 5.8 D value in the ground state. This data, together with the dipole moment for the first excited electronic state (23 D), which was obtained by Czekalla et al.64 from electric dichroism measurements, allow one to conclude that the direction of the dipole moment changes very little upon electronic excitation of this chromophore. Baliah and Pillay65 analyzed the dipole moments of a series of fluorene derivatives at positions 2 and 7, and concluded that both positions were strongly resonant and hence an electron-releasing substituent at one and an electron-withdrawing substituent at the other would adopt coplanar positions relative to the fluorene skeleton. In summary, a change by 18 D in dipole moment of a system of these structural features reflects a substantial charge transfer from the donor group at position 2 to the acceptor group at 7 upon electronic excitation. Although ANF seemingly fulfills requirements (a) and (b) above, it appears not to meet requirement (c) (i.e., that its acidity and basicity should not change upon electronic excitation). In fact, electronic excitation will induce a charge transfer from the amino group, so the protons in it will increase in acidity and the transition will be contaminated with specific contributions arising from solvent basicity. In order to avoid this contribution, one may in principle replace the amino group with a dimethylamino group (DMANF, 7), which will exhibit appropriate charge transfer with no significant change in its negligible acidity. The increase in basicity of the nitro group upon electronic excitation (a result of charge transfer from the N,N-dimethyl group), should result in little contamination as this group is scarcely basic and its basicity is bound to hardly change with the amount of charge transferred from the N,N-dimethylamino group at position 2 to the fluorene structure. The analysis of the absorption spectra for DMANF in a broad range of solvents suggests that this probe possesses several interesting spectroscopic properties as regards its first absorption band, which is used to assess the polar properties of solvents. Thus,15 (a) its first absorption band is well resolved from the other electronic bands (an increase in solvent polarity results in no overlap with the other electronic bands in the UV-Vis spectrum for this probe); (b) the position of this band is highly sensitive to solvent polarity and is bathochromically shifted with increase in it (the bathochromic shift in the absorption maximum between perfluorohexane and DMSO is 4130 cm-1); (c) in less polar solvents, where the band appears at lower wavelengths, it is observed at ca. 376 nm (i.e., shifted to the visible region to an extent ensuring that no problems derived from the cut-off of the solvent
10.3 Solvent effects based on pure solvent scales
593
concerned will be encountered); and (d) its first band becomes structured in non-polar solvents (see Figure 1 in ref. 15). In summary, DMANF is a firm candidate for use as a solvent dipolarity/polarizability probe since its absorption is extremely sensitive to changes in the nature of the solvent, largely as a result of the marked increase in its dipole moment on passing from the electronic ground state to the first excited state. In addition, the change does not affect the dipole moment direction, which is of great interest if the compound is to be used as a probe. Because it possesses a large, rigid aromatic structure, DMANF is highly polarizable; consequently, its first electronic transition occurs at energies where no appreciable interferences with the cut-offs of ordinary solvents are to be expected. However, the change in structure of the first absorption band for DMANF in passing from non-polar solvents to polar solvents and the potential contaminating effect of solvent acidity on the position of this band entails introducing a homomorph for the probe in order to offset the detrimental effects of these factors on the estimation of solvent polarities. The homomorph to be used should essentially possess the same structure as the probe, viz. a nitro group at position 7 ensuring the occurrence of the same type of interaction with the solvents and an electron-releasing group at position 2 ensuring similar, through weaker, interactions with the nitro function at 7 in order to obtain a lower dipole moment relative to DMANF). The most suitable replacement for the -NMe2 function in this context is a fluorine atom, which poses no structural problems and is inert to solvents. The homomorph chosen was thus 2-fluoro-7-nitrofluorene (FNF) (9). The analysis of the absorption spectra for FNF in a broad range of solvents clearly revealed that its first absorption band behaves identically with that for DMANF (its structure changes in passing from non-polar solvents to polar ones). However, the bathochromic shift in this band with increase in solvent polarity is much smaller than that in DMANF (see Figure 1 in ref. 15). Obviously, the difference between the solvatochromism of DMANF and FNF will cancel many of the spurious effects involved in measurements of solvent polarity. Because the envelopes of the first absorption bands for FNF and DMANF are identical (see Figure 1 in ref. 15), one of the most common sources of error in polarity scales is thus avoided. The polarity of a solvent on the SPP scale is given by the difference between the solvatochromism of the probe DMANF and its homomorph FNF [∆v(solvent) = vFNF vDMANF] and can be evaluated on a fixed scale from 0 for the gas phase (i.e., the absence of solvent) to 1 for DMSO, using the following equation: SPP(solvent) = [∆v(solvent) - ∆v(gas)] / [∆v(DMSO) - ∆v(gas)] [10.3.9] Table 10.3.1 gives the SPP values for a broad range of solvents, ranked in increasing order of polarity. Data were all obtained from measurements made at our laboratory and have largely been reported elsewhere15,66-70 -some, however, are published here for the first time.
594
Javier Catalán
Table 10.3.1. The property parameters of solvents: Polarity/Polarizability SPP, Basicity SB, Acidity SA Solvents
SPP
SB
SA
0
0
0a
perfluoro-n-hexane
0.214
0.057
0a
2-methylbutane
0.479
0.053
0a
petroleum ether
0.493
0.043
0a
n-pentane
0.507
0.073
0a
n-hexane
0.519
0.056
0a
n-heptane
0.526
0.083
0a
isooctane
0.533
0.044
0a
cyclopentane
0.535
0.063
0a
n-octane
0.542
0.079
0a
ethylcyclohexane
0.548
0.074
0a
n-nonane
0.552
0.053
0a
cyclohexane
0.557
0.073
0a
n-decane
0.562
0.066
0a
n-undecane
0.563
0.080
0a
methylcyclohexane
0.563
0.078
0a
butylcyclohexane
0.570
0.073
0a
propylcyclohexane
0.571
0.074
0a
n-dodecane
0.571
0.086
0a
decahydronaphthalene
0.574
0.056
0a
mesitylene
0.576
0.190
0a
n-pentadecane
0.578
0.068
0a
n-hexadecane
0.578
0.086
0a
cycloheptane
0.582
0.069
0a
tert-butylcyclohexane
0.585
0.074
0a
cyclooctane
0.590
0.077
0a
1,2,3,5-tetramethylbenzene
0.592
0.186
0a
cis-decahydronaphthalene
0.601
0.056
0a
tripropylamine
0.612
0.844
0s
m-xylene
0.616
0.162
0a
triethylamine
0.617
0.885
0s
gas phase
10.3 Solvent effects based on pure solvent scales
Solvents
595
SPP
SB
SA
p-xylene
0.617
0.160
0a
1-methylpiperidine
0.622
0.836
0s
tributylamine
0.624
0.854
0s
1,4-dimethylpiperazine
0.627
0.832
0s
hexafluorobenzene
0.629
0.119
0a
trimethylacetic acid
0.630
0.130
0.471
dibutylamine
0.630
0.991
0s
di-n-hexyl ether
0.630
0.618
0a
1-methylpyrrolidine
0.631
0.918
0s
tetrachloromethane
0.632
0.044
0a
di-n-pentyl ether
0.636
0.629
0a
butylbenzene
0.639
0.149
0a
o-xylene
0.641
0.157
0a
isobutyric acid
0.643
0.281
0.515
isovaleric acid
0.647
0.405
0.538
ethylbenzene
0.650
0.138
0a
di-n-butyl ether
0.652
0.637
0a
hexanoic acid
0.656
0.304
0.456
toluene
0.655
0.128
0a
propylbenzene
0.655
0.144
0a
tert-butylbenzene
0.657
0.171
0a
N-methylbutylamine
0.661
0.960
0s
heptanoic acid
0.662
0.328
0.445
di-isopropyl ether
0.663
0.657
0a
N-methylcyclohexylamine
0.664
0.925
0s
N,N-dimethylcyclohexylamine
0.667
0.998
0s
benzene
0.667
0.124
0a
1,2,3,4-tetrahydronaphthlene
0.668
0.180
0a
ethylenediamine
0.674
0.843
0.047
di-n-propyl ether
0.676
0.666
0a
propionic acid
0.690
0.377
0.608
tert-butyl methyl ether
0.687
0.567
0a
diethyl ether
0.694
0.562
0a
596
Javier Catalán
Solvents
SPP
SB
SA
2-methylbutyric acid
0.695
0.250
0.439
butyl methyl ether
0.695
0.505
0a
pentafluoropyridine
0.697
0.144
0a
1,4-dioxane
0.701
0.444
0.0
2-methyltetrahydrofuran
0.717
0.584
0.0
1-methylnaphthalene
0.726
0.156
0a
butylamine
0.730
0.944
0.0
ethoxybenzene
0.739
0.295
0a
piperidine
0.740
0.933
0.0
1-undecanol
0.748
0.909
0.257
isoamyl acetate
0.752
0.481
0.0
1-decanol
0.765
0.912
0.259
fluorobenzene
0.769
0.113
0a
1-nonanol
0.770
0.906
0.270
tetrahydropyran
0.778
0.591
0.0
acetic acid
0.781
0.390
0.689
n-propyl acetate
0.782
0.548
0.0
n-butyl acetate
0.784
0.525
0.0
methyl acetate
0.785
0.527
0.0
1-octanol
0.785
0.923
0.299
chloroform
0.786
0.071
0.047
2-octanol
0.786
0.963
0.088
cyclohexylamine
0.787
0.959
0.0
trimethyl orthoformate
0.787
0.528
0.0
1,2-dimethoxyethane
0.788
0.636
0.0
pyrrolidine
0.794
0.990
0.0
ethyl acetate
0.795
0.542
0.0
1-heptanol
0.795
0.912
0.302
N,N,-dimethylaniline
0.797
0.308
0.0
methyl formate
0.804
0.422
0.0
2-buthoxyethanol
0.807
0.714
0.292
1-hexanol
0.810
0.879
0.315
3-methyl-1-butanol
0.814
0.858
0.315
10.3 Solvent effects based on pure solvent scales
Solvents
597
SPP
SB
SA
propyl formate
0.815
0.549
0.0
1-pentanol
0.817
0.860
0.319
dibenzyl ether
0.819
0.330
0a
methoxybenzene
0.823
0.299
0.084
chlorobenzene
0.824
0.182
0a
bromobenzene
0.824
0.191
0a
cyclooctanol
0.827
0.919
0.137
2,6-dimethylpyridine
0.829
0.708
0.0
2-methyl-2-propanol
0.829
0.928
0.145
2-pentanol
0.830
0.916
0.204
2-methyl-2-butanol
0.831
0.941
0.096
2-methyl-1-propanol
0.832
0.828
0.311
formamide
0.833
0.414
0.674
2,4,6-trimethylpyridine
0.833
0.748
0.0
3-hexanol
0.833
0.979
0.140
iodobenzene
0.835
0.158
0a
ethyl benzoate
0.835
0.417
0.0
dibutyl oxalate
0.835
0.549
0.0
methyl benzoate
0.836
0.378
0.0
1-chlorobutane
0.837
0.138
0a
1-butanol
0.837
0.809
0.341
2-methyl-1-butanol
0.838
0.900
0.289
tetrahydrofuran
0.838
0.591
0.0
pyrrole
0.838
0.179
0.387
cycloheptanol
0.841
0.911
0.183
2-butanol
0.842
0.888
0.221
1,3-dioxolane
0.843
0.398
0.0
3-methyl-2-butanol
0.843
0.893
0.196
2-hexanol
0.847
0.966
0.140
cyclohexanol
0.847
0.854
0.258
1-propanol
0.847
0.727
0.367
2-propanol
0.848
0.762
0.283
ethanol
0.853
0.658
0.400
598
Javier Catalán
Solvents
SPP
SB
SA
1,2-dimethoxybenzene
0.854
0.340
0.0
2-methoxyethyl ether
0.855
0.623
0.0
methanol
0.857
0.545
0.605
methyl methoxyacetate
0.858
0.484
0.0
3-pentanol
0.863
0.950
0.100
cyclopentanol
0.865
0.836
0.258
cyclohexanone
0.874
0.482
0.0
propionitrile
0.875
0.365
0.030
allyl alcohol
0.875
0.585
0.415
propiophenone
0.875
0.382
0a
triacetin
0.875
0.416
0.023
dichloromethane
0.876
0.178
0.040
2-methylpyridine
0.880
0.629
0.0
2-butanone
0.881
0.520
0.0
acetone
0.881
0.475
0.0
2-methoxyethanol
0.882
0.560
0.355
3-pentanone
0.883
0.557
0.0
benzyl alcohol
0.886
0.461
0.409
4-methyl-2-pentanone
0.887
0.540
0.0
trimethyl phosphate
0.889
0.522
0.0
1-methylpyrrole
0.890
0.244
0.0
1,2-dichloroethane
0.890
0.126
0.030
2-phenylethanol
0.890
0.523
0.376
2-chloroethanol
0.893
0.377
0.563
nitroethane
0.894
0.234
0.0
acetonitrile
0.895
0.286
0.044
chloroacetonitrile
0.896
0.184
0.445
N-methylacetamide
0.897
0.735
0.328
1,2-butanediol
0.899
0.668
0.466
valeronitrile
0.900
0.408
0.0
N-methylaniline
0.902
0.212
0.073
2,3-butanediol
0.904
0.652
0.461
acetophenone
0.904
0.365
0.044
10.3 Solvent effects based on pure solvent scales
Solvents
599
SPP
SB
SA
nitromethane
0.907
0.236
0.078
cyclopentanone
0.908
0.465
0.0
triethyl phosphate
0.908
0.614
0.0
1,2-dichlorobenzene
0.911
0.144
0.033
tetrahydrothiophene
0.912
0.436
0a
2,2,2-trifluoroethanol
0.912
0.107
0.893
butyronitrile
0.915
0.384
0.0
N-methylformamide
0.920
0.590
0.444
isobutyronitrile
0.920
0.430
0.0
pyridine
0.922
0.581
0.033
2-methyl-1,3-propanediol
0.924
0.615
0.451
1,2-propanediol
0.926
0.598
0.475
propylen carbonate
0.930
0.341
0.106
N,N-diethylacetamide
0.930
0.660
0.0
hexamethylphosphoric acid triamide
0.932
0.813
0s
1,2-ethanediol
0.932
0.534
0.565
N,N-diethylformamide
0.939
0.614
0.0
1,3-butanediol
0.944
0.610
0.424
1,2,3-propanetriol
0.948
0.309
0.618
1-methylimidazole
0.950
0.668
0.069
1,4-butanediol
0.950
0.598
0.424
1,3-propanediol
0.951
0.514
0.486
tetramethylurea
0.952
0.624
0.0
N,N-dimethylformamide
0.954
0.613
0.031
2-pyrrolidinone
0.956
0.597
0.347
benzonitrile
0.960
0.281
0.047
2,2,2-trichloroethanol
0.968
0.186
0.588
nitrobenzene
0.968
0.240
0.056
1-methyl-2-pyrrolidinone
0.970
0.613
0.024
N,N-dimethylacetamide
0.970
0.650
0.028
water
0.962
0.025
1.062
γ-butyrolactone
0.987
0.399
0.057
dimethyl sulfoxide
1.000
0.647
0.072
600
Javier Catalán
Solvents
SPP
SB
SA
sulfolane
1.003
0.365
0.052
m-cresol
1.000
0.192
0.697
1,1,1,3,3,3-hexafluoro-2-propanol
1.014
0.014
1.00
α α α-trifluoro-m-cresol
1.085
0.051
0.763
a
Assumed value because it is considered non acid solvent. Assumed value because the first band of the TBSB spectra exhibits vibronic structure.
a
A brief analysis of these SPP data allows one to draw several interesting conclusions from structural effects on solvent polarity, namely: • Cyclohexane is used as the non-polar reference in many scales. In the SPP scale, the polarity gap between cyclohexane and the gas phase (0.557 SPP units) is as wide as that between cyclohexane and the highest polarity (0.443 SPP units). • Alkanes span a wide range of SPP values (e.g., 0.214 for perfluoro-n-hexane, 0.479 for 2-methylbutane and 0.601 for decalin). • Unsaturation increases polarity in alcohols. Thus, the SPP values for n-propanol, allyl alcohol and propargyl alcohol are 0.847, 0.875 and 0.915, respectively. 10.3.5 SOLVENT BASICITY: THE SB SCALE72 As stated above, for a probe of solvent basicity to be usable in UV-Vis spectroscopy, it should meet a series of requirements. One is that it should be acidic enough in its electronic ground state to allow characterization of the basicity of its environment. In addition, its acidity should increase upon electronic excitation such that its electronic transitions will be sensitive to the basicity of the medium. This behavior will result in a bathochromic shift in the absorption band the magnitude of which will increase with increasing basicity of the environment. The probe should also be free of potential conformational changes that might influence the electronic transition to be evaluated. Finally, its molecular structure should be readily converted into a homomorph lacking the acid site without any side effects that might affect the resulting solvent basicity. 5-Nitroindoline (NI, 9) possesses the above-described electronic and structural features. It is an N-H acid with a single acid site borne by a donor group whose free rotation is hindered by an ethylene bridge on the ring. However, if charge transfer endows the compound with appropriate acid properties that increase with electronic excitation, the basicity of the acceptor (nitro) group will also increase and the polarity of the compound will be altered as a result. Both effects will influence the electronic transition of the probe that is to be used to evaluate basicity. Replacement of the acid proton (N-H) by a methyl group in this molecular structure has the same side effects; as a result, the compound will be similarly sensitive to the polarity and acidity of the medium, but not to its basicity owing to the absence of an acid site. Consequently, 1-methyl-5-nitroindoline (MNI, 10) possesses the required properties for use as a homomorph of 5-nitroindoline in order to construct our solvent basicity scale (SB). The suitability of this probe/homomorph couple is consistent with theoretical MP2/6-31G** data. Thus, both the probe and its homomorph exhibit the same sensitivity to solvent polarity/polarizability because of their similar dipole moments (µNI = 7.13, µMNI = 7.31 D) and
10.3 Solvent effects based on pure solvent scales
601
polarizabilities (20.38 and 22.78 α/J-1C2m2 for NI and MNI, respectively). Both also have the same sensitivity to solvent acidity as their surface electrostatic potential minima are virtually coincident [VS,min(NI) = -47.49 kcal mol-1, VS,min(MNI) = -47.58 kcal mol-1]; according to Politzer et al.,71 this means that they exhibit the same hydrogen-bonding basicity. In addition, the electrostatic potential surface in both molecules suggests that sole electrophilic sites are those on the oxygen atoms of the nitro group. Obviously, the difference in solvatochromism between NI and MNI [∆v(solvent) = vNI - vMNI] will cancel many spurious effects accompanying the basicity effect of the solvent. In addition, NI and MNI possess several advantageous spectral features; thus, they exhibit a sharp first absorption band that overlaps with no higher-energy band in any solvent, and both have the same spectral envelope in such a band (see Figure 1 in ref. 72), which facilitates comparison between the two spectra and the precise establishment of the basicity parameter, SB. The basicity of a solvent (SB) on a scale encompassing values between zero for the gas phase -the absence of solvent- and unity for tetramethylguanidine (TMG) can be directly obtained from the following equation: SB(solvent) = [∆v(solvent) - ∆v(gas)] / [∆v(TMG) - ∆v(gas)]
[10.3.10]
Table 10.3.1 lists the SB values for a wide range of solvents. All data were obtained from measurements made in our laboratory; most have been reported elsewhere70,72 but some are published here for the first time. A brief analysis of these SB data allows one to draw several interesting conclusions as regards structural effects on solvent basicity. Thus: • Appropriate substitution in compound families such as amines and alcohols allows the entire range of the solvent basicity scale to be spanned with substances from such families. Thus, perfluorotriethylamine can be considered non-basic (SB=0.082), whereas N,N-dimethylcyclohexylamine is at the top of the scale (SB=0.998). Similarly, hexafluoro-2-propanol is non-basic (SB=0.014), whereas 2-octanol is very near the top (SB=0.963). • The basicity of n-alkanols increases significantly with increase in chain length and levels off beyond octanol. • Cyclization hardly influences solvent basicity. Thus, there is little difference in basicity between n-pentane (SB=0.073) and cyclopentane (SB=0.063) or between n-pentanol (SB=0.869) and cyclopentanol (SB=0.836). • Aromatization decreases solvent basicity by a factor of 3.5-5.5, as illustrated by the following couples: pyrrolidine/pyrrole (0.99/0.18), N-methylpyrrolidine/ N-ethylpyrrole (0.92/0.22), tetrahydrofuran/furan (0.59/0.11), 2-methyltetrahydrofuran/2-methylfuran (0.56/0.16), cyclohexylamine/aniline (0.96/0.26), N-methylcyclohexylamine/N-methylaniline (0.92/0.21), N,N-dimethylcyclohexylamine/N,N-dimethylaniline (0.99/0.30) and a structurally less similar couple such as piperidine/pyridine (0.93/0.58). The potential family-dependence of our SB scale was examined and discarded elsewhere.73 10.3.6 SOLVENT ACIDITY: THE SA SCALE74 As stated above, a probe for solvent acidity should possess a number of features to be usable in UV-Vis spectroscopy. One is that it should be basic enough in its electronic ground state
602
Javier Catalán
to be able to characterize the acidity of its environment. In addition, its basicity should increase upon electronic excitation such that its electronic transitions will be sensitive to the acidity of the medium. This behavior will result in a bathochromic shift in the absorption band the magnitude of which will increase with increasing acidity of the environment. The probe should also be free of potential conformational changes that might influence the electronic transition to be evaluated. Finally, its molecular structure should be easily converted into a homomorph lacking the basic site without any side effects potentially affecting the resulting solvent acidity. Our group has shown24,75 that the extremely strong negative solvatochromism of the chromophore stilbazolium betaine dye (11), about 6500 cm-1, is not a result of a change in the non-specific effect of the solvent but rather of change in acidity. This was confirmed75 by a study of the solvatochromic effect of a derivative of (11), o,o'-di-tert-butylstilbazolium betaine dye (DTBSB) (12) in the same series of solvents. In this compound, the basic site in the betaine dye (its oxygen atom) is protected on both sides by bulky tert-butyl groups; the specific effect of the solvent is hindered and the compound exhibits only a small solvatochromic effect that can be ascribed to non-specific solvent effects.75 Since the carbonyl group of stilbazolium betaine possesses two lone pairs in the plane of the quinoid ring, one can have two (11), one (TBSB, 13) or no channels (DTBSB) to approach hydrogen bond-donor solvents, depending on the number of o-tert-butyl groups present in the molecular structure of the (11) derivative concerned. Our group24 has also shown that, for at least 20 alkanols, the wavenumber difference between the maximum of the first absorption bands for (11) and TBSB, and for TBSB and DTBSB, is virtually identical. These results show that an oxygen lone pair in (11) is basic enough, as a result of hydrogen bonding, for (11) to be used as an acidity probe in UV-Vis spectroscopy; thus, the absorption maximum for TBSB shifts by about 1000 cm-1 from 1-decanol to ethanol,24 whereas that for DTBSB shifts by only 300 cm-1. This sensitivity to acidity, and the structural likeness of the probe and its homomorph -which must endow them with a similar sensitivity to solvent basicity and dipolarity/polarizability-, suggest that the two compounds make an appropriate probe-homomorph couple for developing a pure solvent acidity scale. For this purpose, the first visible absorption band for these compounds exhibits quite an in-
10.3 Solvent effects based on pure solvent scales
603
teresting spectroscopic behavior. Thus, in a non-HBD solvent (i.e., one that cannot interact directly with the oxygen lone pair), the band is structured; on the other hand, the band loses its structure upon interaction with an HBD solvent.75 The SA acidity scale was established by comparing the solvatochromism of the probes TBSB and DTBSB using the method of Kamlet and Taft.18a In this method, the solvatochromism of DTBSB in a solvent is used as the reference for zero acidity. Consequently, non-acidic solvents obey the equation vTBSB = 1.409vDTBSB - 6288.7
[10.3.11]
with n = 50, r = 0.961 and sd = 43.34 cm-1. Based on eq. [10.3.11] for non-acidic solvents, the SA value for a given solvent can be obtained from the following expression, where a value of 0.4 is assigned to ethanol, the acid solvatochromic behavior of which is exhibited at 1299.8 cm-1: SA = [[vTBSB - (1.409vDTBSB - 6288.7)] / 1299.8]0.4
[10.3.12]
Table 10.3.1 gives the SA values for a wide variety of C-H, N-H and O-H acid solvents. Data were all obtained from measurements made in our laboratory.74 The homomorph was insoluble in some solvents, which therefore could not be measured, so they where assumed not to interact specifically and assigned a zero SA value if the probe exhibited a structured spectrum in them. Evaluating the acidity of weakly acidic solvents entails using a probe basic enough to afford measurement of such an acidity; as a result, the probe is usually protonated by strongly acidic solvents and useless for the intended purpose. Stilbazolium betaines are subject to this constraint as they have pK values of about 10 -(11) and DTBSB have a pKa of 8.5776 and ca. 10,77 respectively, which makes them unsuitable for the evaluation of solvents more acidic than methanol. This forced us to find a suitable probe with a view to expanding the SA scale to more acidic solvents. This problem is not exclusive to our scale; in fact, it affects all acidity scales, which usually provide little information about strongly acidic solvents. The problem was solved thanks to the exceptional behavior of 3,6-diethyltetrazine70 (DETZ, 14). This compound possesses two lone electron pairs located on opposite sites of its hexagonal ring; the strong interaction between the two pairs results in one antibonding n orbital in the compound lying at an anomalously high energy level; this, as shown below, has special spectroscopic implications. The π-electron system of DETZ is similar to that of benzene. Thus, the first π → π * transition in the two systems appears at λmax = 252 nm in DETZ and at 260 nm in benzene. On the other hand, the π → π * transition in DETZ is strongly shifted to the visible region (λmax ≅ 550 nm), which is the origin of its deep color. This unique feature makes this compound a firm candidate for use as an environmental probe as the wide energy gap between the two transitions excludes potential overlap. Problems such as protonation of the DETZ probe by acid solvents can be ruled out since, according to Mason,78 this type of substance exhibits pK values below zero. The sensitivity of the n → π* transition in DETZ to the solvent is clearly reflected in its UV/Vis spectra in methylcyclohexane, methanol, trifluoromethanol,
604
Javier Catalán
hexafluoro-2-propanol, acetic acid and trifluoroacetic acid, with λmax values of 550, 536, 524, 517, 534 and 507 nm, respectively. In addition, the spectral envelope undergoes no significant change in the solvents studied, so a potential protonation of the probe -even a partial one- can be discarded. The shift in the n → π* band is largely caused by solvent acidity; however, an n → π* transition is also obviously sensitive to the polarity of the medium, the contribution of which must be subtracted if acidity is to be accurately determined. To this end, our group used the solvatochromic method of Kamlet and Taft18a to plot the frequency of the absorption maximum for the n → π* transition against solvent polarity (the SPP value). As expected, non-acidic solvents exhibited a linear dependence [eq. (10.3.13)], whereas acidic solvents departed from this behavior -the more acidic the greater the divergence. This departure from the linear behavior described by eq. [10.3.13] is quantified by ∆vDETZ, vDETZ = 1.015SPP + 17.51
[10.3.13]
with n = 13, r = 0.983 and sd = 23 cm-1. The SA value of an acid solvent as determined using the DETZ probe is calculated using the expression established from the information provided by a series of solvents of increased acidity measurable by the TBSB/DTBSB probe/homomorph couple (viz. 1,2-butanediol, 1,3,-butanediol, ethanol, methanol and hexafluoro-2-propanol); this gives rise to eq. [10.3.14], which, in principle, should only be used to determine SA for solvents with ∆vDETZ values above 100 cm-1: SA = 0.833∆vDETZ + 0.339
[10.3.14]
with n = 6 and r = 0.987. Table 10.3.1 gives the SA values for a wide range of highly acidic solvents. Data were largely obtained from measurements made in our laboratory;70 some, however, are reported here for the first time. A brief analysis of these SA data allow one to draw several interesting conclusions from structural effects on solvent acidity. Thus: • Substitution into alcoholic compounds allows the entire range of the solvent acidity scale to be spanned. Thus, 2-octanol can be considered to be scarcely HBD (SA=0.088) and ethylene glycol to be highly HBD (SA=0.717). Perfluoroalkanols are even more highly HBD. • The acidity of 1-alkanols and carboxylic acids decreases significantly with increasing chain length but levels off beyond nonanol and octanoic acid, respectively. • Cyclization decreases solvent acidity. Thus, SA decreases from 0.318 to 0.257 between 1-pentanol and cyclopentanol, and from 0.298 to 0.136 between 1-octanol and cyclooctanol. • Aromatic rings make aniline (SA=0.131) and pyrrole (SA=0.386) more HBD than their saturated homologs: cyclohexylamine and pyrrolidine, respectively, both of which are non-HBD (SA=0). • The acidity of alkanols decreases dramatically with increasing alkylation at the atom that bears the hydroxyl group [e.g., from 1-butanol (SA=0.340) to 2-butanol (SA=0.221) to tert-butyl alcohol (SA=0.146)].
10.3 Solvent effects based on pure solvent scales
605
10.3.7 APPLICATIONS OF THE PURE SPP, SA AND SB SCALES The SPP general solvent scale, and the SA and SB specific solvent scales, are orthogonal to one another, as can be inferred from the small correlation coefficients obtained in mutual fittings involving the 200 solvents listed in Table 10.3.1 [r2(SPP vs. SA) = 0.13, r2(SPP vs. SB) = 0.10 and r2(SA vs. SB) = 0.01]. These results support the use of these scales for the multi-parameter analysis of other solvent scales or data sets sensitive to the solvent effect on the basis of the following equation: P = aSPP + bSA + c SB + P0
[10.3.15]
where P is the quantity to be described in a given solvent; SPP, SA and SB are the corresponding polarity/polarizability, acidity and basicity values for such a solvent; coefficients a, b and c denote the sensitivity of P to such effects; and P0 is the P value in the absence of solvent (i.e., the gas phase, which is given a zero value in our scales). Each of these scales (SPP, SB and SA) has previously been compared with other reported solvent scales and found to be pure scales for the respective effects. However, eq. [10.3.15] is used below to perform a multi-parameter analysis of various reported scales and experimental data sets of interest. 10.3.7.1 Other reported solvents scales Reported scales for describing solvent polarity [f(ε,n),21,79 π*80 and S'60], basicity (DN20 and β18) and acidity (AN81 and α18) were previously analyzed against our SPP, SA and SB scales in the originating references, so no further comment is made here. Rather, this section analyses the behavior of reported single-parameter scales developed to describe the global behavior of solvents [viz. the Z, χR, Py, Φ, G and ET(30) scales] against the SPP, SB and SA scales on the basis of the 200 solvents listed in Table 10.3.1. The Y scale, established to describe the behavior of solvolysis-like kinetics, is dealt with in the section devoted to the description of kinetic data. The data used in this analysis were taken from the following sources: those for the Z scale from the recent review paper by Marcus;56 those for the χR, Φ and G scales from the compilation in Table 7.2 of Reichardt’s book;1 those for parameter Py from the paper by Dong and Winnik;50 and those for ET(30) from the recent review by Reichardt13 or Table 7.3 in his book.1 The Z value for the 51 solvents in Table 10.3.1 reported by Marcus allow one to establish the following equation: Z = 22.37(± 4.85)SPP + 7.68(± 1.64)SB + 31.27(± 1.81)SA + 41.03(± 4.15)
[10.3.16]
with r = 0.944 and sd = 3.03 kcal mol-1. This equation reveals that Z is largely the result of solvent acidity and polarity, and also, to a lesser extent, of solvent basicity. The strong delocalization of charge in the structure of the probe (4) upon electronic excitation, see scheme IV, accounts for the fact that the electronic transition of this probe is hypsochromically shifted by solvent polarity and acidity. The shift is a result of the increased stabilizing effect of the solvent -exerted via general and specific interactions- in the electronic ground state being partly lost upon electronic excitation through delocalization of the charge in the electronic excited state, which gives rise to decreased polarity and specific solvation. On the other hand, the small contribution of solvent basicity appears to have no immediate explanation.
606
Javier Catalán
Figure 10.3.1. Plot of Kosower’s Z values vs. the predicted Z values according to eq. [10.3.16].
Figure 10.3.2. Plot of Brooker’s χR values vs. the predicted χR values according to [eq. 10.3.17].
Scheme V.
It should be noted that the three solvent effects (polarity, acidity and basicity) increase Z; also, as can be seen from Figure 10.3.1, the greatest imprecision in this parameter corresponds to solvents with Z values below 70 kcal mol-1. The 28 solvents in Table 10.3.1 with known values of χR reveal that this parameter reflects solvent polarity and, to a lesser extent, also solvent acidity: χR = -15.81(±1.19)SPP - 4.74(±0.90)SA + 58.84(±0.94)
[10.3.17]
with r = 0.958 and sd = 0.78 kcal mol-1. It should be noted that the two solvent effects involved (polarity and acidity) contribute to decreasing χR. As can be seen from Figure 10.3.2, the precision in χR is similarly good throughout the studied range. The structure of the probe used to construct this scale (5) reveals that its electron-donor site (an amino group) is not accessible for the solvent, so the charge transfer involved in the electronic excitation does not alter its solvation; rather, it increases its polarity and localization of the charge on the acceptor sites of the compound (carbonyl groups), see scheme V, so increasing solvent polarity and acidity will result in a bathochromically shifted electronic transition by effect of the increased stabilization of the excited state through general and specific interactions. The results of the multi-parameter analysis of Py suggests that its values are highly imprecise; even if the value for the gas phase is removed, this parameter only reflects -in an im-
10.3 Solvent effects based on pure solvent scales
607
Figure 10.3.3. Plot of Dong and Winnck’s Py values vs. the predicted Py values according to eq. [10.3.18] (Point for gas phase, , is included in the plot, but not in the regression equation [10.3.18]).
Figure 10.3.4. Plot of Dubois and Bienvenüe’s Φ values vs. the predicted Φ values according to eq.[10.3.19].
precise manner- solvent polarity, as is clearly apparent from eq. [10.3.18] (see Figure 10.3.3): Py = 2.30(±0.20)SPP - 0.57(±0.16)
[10.3.18]
with r = 0.839 and sd = 0.20. The difficulty of understanding the performance of the probe precludes rationalizing its spectral behavior. The 19 Φ values examined indicate that, based on eq. [10.3.19] and Figure 10.3.4, this parameter reflects mostly solvent acidity and, to a lesser extent, solvent polarity: Φ = 0.18(±0.11)SPP + 0.38(±0.04)SA - 0.07(±0.08)
[10.3.19]
with r = 0.941 and sd = 0.05. The Φ scale is based on the solvatochromic behavior of an n → π * electronic transition, so it reflects a high hypsochromic sensitivity to solvent acidity (specific solvation between the lone pair involved in the electronic transition and solvent acidity is lost when one electron in the pair is electronically excited). An n → π * electronic transition is also known to decrease the polarity of the chromophore concerned,82 so an increase in SPP for the solvent will cause a hypsochromic shift in the electronic transition. Based on Figure 10.3.5 and the following equation, the values of G parameters are also dictated by solvent polarity and, to a lesser extent, solvent acidity: G = 102.96(±9.02)SPP + 78.20(±27.78)SA - 0.35(±6.28) [10.3.20] with r = 0.965 and sd = 7.6. It should be noted that the independent term for the gas phase is very small and highly imprecise in both the G scale and the Φ scale.
608
Javier Catalán
Figure 10.3.5. Plot of Allerhand and Schleyer’s G values vs. the predicted G values according to eq. [10.3.20].
Figure 10.3.6. Plot of Reichardt’s ET(30) values vs. the predicted ET(30) values according to eq. [10.3.21].
For the 138 solvents with a known ET(30) value listed in Table 10.3.1, this parameter reflects mostly solvent acidity and polarity, and, to a lesser degree, also solvent basicity (see Figure 10.3.6): ET(30) = 0.62(±0.04)SPP + 0.12(±0.02)SB + 0.77(±0.02)SA -0.31(±0.03) [10.3.21]
with r = 0.965 and sd = 0.06. It should be noted that the three solvent effects (polarity, acidity and basicity) contribute to increasing ET(30), i.e., to a hypsochromic shift in the transition of the probe (see Figure 10.3.6). The spectroscopic behavior of the probe used to construct this scale (1) is typical of a structure exhibiting highly localized charge on its carbonyl group in the electronic ground state, see scheme III, and hence strong stabilization by effect of increased solvent polarity and acidity. The electronic transition delocalizes the charge and results in an excited state that is much less markedly stabilized by increased polarity or acidity in the solvent. The decreased dipole moment associated to the electronic transition in this probe also contributes to the hypsochromic shift. The small contribution of solvent basicity to the transition of the probe (1) is not so clear, however. 10.3.7.2 Treatment of the solvent effect in: 10.3.7.2.1 Spectroscopy Analyzing the SPP, SB and SA scales in the light of spectroscopic data that are sensitive to the nature of the solvent poses no special problem thanks to the vertical nature of the transitions, where the cybotactic region surrounding the chromophore is hardly altered. The papers where the SPP, SA and SB scales were reported discuss large sets of spectroscopic data in terms of the nature of the solvent. Some additional comments are made below. Recently, Fawcet and Kloss,83 analyzed the S=O stretching frequencies of dimethyl sulfoxide (DMSO) with a view to elucidating the behavior of 20 solvents and the gas phase.
10.3 Solvent effects based on pure solvent scales
609
They found the frequency of this vibration in DMSO to be shifted by more than 150 cm-1 and to be correlated to the Gutmann acceptor number (AN) for the solvents. Based on our analysis, the frequency shift reflects the acidity and, to a lesser extent, the polarity of the solvent, according to the following equation: vS=O = 38.90(±7.50)SPP - 89.67(±4.60)SA + 1099.1
[10.3.22]
with n = 21, r = 0.979 and sd = 6.8 cm-1. This fit is very good, taking into account that it encompasses highly polar solvents such as DMSO itself, highly acidic solvents such as trifluoroacetic acid (SPP=1.016, SA=1.307), and highly non-polar and non-acidic solvents such as the gas phase. Giam and Lyle84 determined the solvent sensitivity of the 19F NMR shifts on 4-fluoropyridine relative to benzene as an internal reference in 31 of the solvents listed in Table 10.3.1. If DMF is excluded on the grounds of its odd value, the chemical shifts for the remaining 30 solvents are accurately described by the following function of solvent acidity and polarity: ∆4F = -1.66(±0.57)SPP - 6.84(±0.36)SA - 6.50(±0.45)
[10.3.23]
with r = 0.971 and sd = 0.34 ppm. Clearly, the acidity of the medium is the dominant factor in the chemical shift measured for this compound, which reveals the central role played by the lone electron pair in pyridine. Both absorption and emission electronic transitions are acceptably described by our scales, as shown by a recent study85 on the solvation of a series of probes containing an intramolecular hydrogen bond, so no further comment is made here other than the following: even if one is only interested in evaluating the change in dipole moment upon electronic excitation via solvatochromic analysis, a multi-parameter analysis must be conducted in order to isolate the shifts corresponding to the pure dipolar effect of the solvent.85 Lagalante et al.86 proposed the use of 4-nitropyridine N-oxide as a suitable solvatochromic indicator of solvent acidity. The hypsochromic shifts determined by these authors for 43 of the solvents in Table 10.3.1 are due largely to the acidity of solvent and, to a lesser extent, also to its basicity; based on the following equation, however, solvent polarity induces a bathochromic shift in the band: vmax = -0.92(±0.44)SPP + 0.68(±0.17)SB + 3.63(±0.23)SA +29.10(±0.31) [10.3.24]
with r = 0.943 and sd = 0.31 kK. As can be seen from Figure 10.3.7, this probe classifies solvents in groups encompassing non-acidic solvents (below 29 kK), moderately acidic solvents (at about 30 kK) and highly acidic solvents (between 31 and 32 kK). Because the solvatochromism does not change gradually with increase in solvent acidity, the probe appears to be unsuitable for quantifying this effect. Davis87 determined the solvatochromism of a charge-transfer complex formed by tetra-n-hexylammonium iodide-nitrobenzene in 23 different solvents and, using the Z and ET(30) scales, observed a bilinear behavior in scarcely polar and highly polar solvents. A
610
Javier Catalán
Figure 10.3.7. Plot of the experimental UV/Vis absorption maxima,vmax, of 4-nitropyridine N-oxide vs. the predicted vmax values according to eq. [10.3.24].
Figure 10.3.8. Plot of the fluorescence maxima, vmax(em), for the Neutral Red in different solvents vs. the solvent polarity function ∆f [∆f = (ε -1)/(2ε +1)-(n2 -1) / ( 2n2 +1)].
multi-parameter fit of the data for 22 of the solvents revealed that the solvatochromism is seemingly sensitive to solvent acidity only: vc.t. = 11.23(±0.75)SA + 21.41(±0.184)
[10.3.25]
with n = 22, r = 0.959 and sd = 0.66 kK. An interesting situation is encountered in the analysis of electronic transitions that results from chromophores which, assisted by solvent polarity, can undergo a change in the electronic state responsible for the fluorescence emission; this is a frequent occurrence in systems involving a TICT mechanism.68,88-90 In this situation, a plot of Stokes shift against solvent polarity is a bilinear curve depending on whether the solvents are non-polar (where the fluorescence is emitted from the normal excited state of the chromophore) or polar enough for the transition to take place from the more polar state. Unless the solvents used are carefully selected, it is venturesome to assume that the electronic states will be inverted simply because the variation of solvatochromism with a function of solvent polarity is bilinear. Recently, Sapre et al.91 showed that a plot of the maximum fluorescence of Neutral Red (NR) against the solvent polarity function ∆f is clearly bilinear (see Figure 10.3.8). Accordingly, they concluded that, in solvents with ∆f > 0.37, the emitting state changes to a much more polar, ICT state. The analysis of this spectroscopic data in the light of our scales reveals that, in fact, the solvatochromism of NR is normal, albeit dependent not only on the polarity of the solvent (SPP), but also on its acidity (SA) (see Figure 10.3.9): NR = -4.31(±0.57)SPP - 1.78(±0.19)SA + 21.72(±0.47) v max
with r = 0.974 and sd = 0.21 kK.
[10.3.26]
10.3 Solvent effects based on pure solvent scales
611
Figure 10.3.9. Plot of the fluorescence maxima, vmax(em), Figure 10.3.10. Plot of Grunwald and Wistein’s log for the Neutral Red in different solvents vs. the pre- ktBuCl values vs. the predicted log ktBuCl values according to eq. [10.3.29]. dicted, vmax(em), values according to eq. [10.3.26].
10.3.7.2.2 Kinetics Kinetics so closely related to the solvent effect as those of the Menschutkin reaction between triethylamine and ethyl iodide [eq. (10.3.27)], the solvolysis of tert-butyl chloride [eq. (10.3.28)] or the decarboxylation of 3-carboxybenzisoxazole [eq. (10.3.29)], are acceptably described by our scales.15,92,93 The equation for the Menschutkin kinetics is log ks/kHex = 8.84(±0.66)SPP + 1.90(±1.37)SA - 4.07(±0.53) [10.3.27] with n = 27, r = 0.947 and sd = 0.41. The rate of this reaction between triethylamine and ethyl iodide, which varies by five orders of magnitude from n-hexane (1.35×10-8 l mol-1 s-1) to DMSO (8.78×10-4 l mol-1 s-1), is accurately described by solvent polarity and acidity -the sensitivity to the latter is somewhat imprecise. The equation for the kinetics of solvolysis of tert-butyl chloride is: log k = 10.02(± 1.14)SPP + 1.84(± 0.99)SB + 8.03(± 0.69)SA - 19.85(± 0.70)
[10.3.28]
with n = 19, r = 0.985 and sd = 0.80. Based on eq. [10.3.28], all solvent effects increase the rate of solvolysis. However, the strongest contribution is that of polarity and the weakest one that of acidity. Although much less significant, the contribution of solvent basicity is especially interesting as it confirms that nucleophilicity also assists in the solvolytic process. Taking into account that it encompasses data spanning 18 orders of magnitude (from log k = -1.54 for water to log k = -19.3 for the gas phase), the fit is very good (see Figure 10.3.10). The equation for the kinetics of decarboxylation of 3-carboxybenzisoxazole is: log k = 10.37(± 1.47)SPP + 2.59(± 0.74)SB - 5.93(± 0.58)SA - 9.74(± 1.17)
with n = 24, r = 0.951 and sd = 0.73.
[10.3.29]
612
Javier Catalán
Equation [10.3.29] reproduces acceptably well the sensitivity of the decarboxylation rate of 3-carboxybenzisoxazole in pure solvents observed by Kemp and Paul.7 In addition, it clearly shows that such a rate increases dramatically with increasing polarity and, also, to a lesser degree, with solvent basicity. By contrast, it decreases markedly with increasing solvent acidity. This behavior is consistent with the accepted scheme for this decarboxylation reaction.93 10.3.7.2.3 Electrochemistry Gritzner94 examined the solvent effect on half-wave potentials and found those of K+ relative to bis(biphenyl)chromium(I)/(0), designated E1/2(BCr)K+, to be related to the Gutmann donor number (DN) for the solvents, so he concluded that K+ behaves as a Lewis acid against basic solvents. Based on the following equation, the behavior of E1/2(BCr)K+ is dictated largely by the basicity of the solvent but it is also dependent, however weakly, on its acidity: E1/2(BCr)K+ = -0.51(±0.05)SB + 0.12(±0.04)SA -1.06(±0.03)
[10.3.30]
with n = 17, r = 0.941 and sd = 0.03 V. It should be noted that solvent basicity increases the half-wave potentials of K+ whereas solvent acidity decreases it. 10.3.7.2.4 Thermodynamics In order to compare the shifts of the conformational equilibrium position with solvent effects it is advisable to select species exhibiting negligible cavity effects on the equilibrium position. Two firm candidates in this respect are the Scheme VI. conformational equilibria of 1,2,2-trichloroethane and the equilibrium between the equatorial and axial forms of 2-chlorocyclohexanone; both are accurately described by our scales.15 Of special interest is also the equilibrium between the keto and enol tautomers of pentane-2,4-dione (see the scheme VI). The ∆G values for this equilibrium in 21 solvents reported by Emsley and Freeman95 are accurately reproduced by solvent polarity and acidity according to ∆G = 18.45(±2.41)SPP + 4.22(±0.86)SA - 18.67(±1.96)
[10.3.31]
with n = 21, r = 0.924 and sd = 1.34 kJ mol-1. The equilibrium position is mainly dictated by solvent polarity; however, there is clearly a specific contribution of solvent acidity from the carbonyl groups of the dione. 10.3.7.3 Mixtures of solvents. Understanding the preferential solvation model The solvation of a solute, whether ionic or neutral, in a mixture of solvents is even more complex than in a pure solvent.1 This is assumed to be so largely because a solvent mixture
10.3 Solvent effects based on pure solvent scales
613
involves interactions not only between the solute and solvent but also among different molecules present in the mixture; the latter type of contribution also plays a central role in the solvation process. Among others, it results in significant deviations of the vapor pressure of a mixture with respect to the ideal behavior established by Raoult’s law. Solvation studies of solutes in mixed solvents have led to the conclusion that the above-mentioned divergences may arise from the fact that the proportion of solvent components may be significantly different around the solute and in the bulk solution. This would be the case if the solute were preferentially surrounded by one of the mixture components, which would lead to a more negative Gibbs energy of solvation.1,96 Consequently, the solvent shell around the solute would have a composition other than the macroscopic ratio. This phenomenon is known as “preferential solvation”, a term that indicates that the solute induces a change with respect to the bulk solvent in its environment; however, such a change takes place via either non-specific solute-solvent interactions called “dielectric enrichment” or specific solute-solvent association (e.g. hydrogen bonding). Preferential solvation has been studied in the light of various methods, most of which are based on conductance and transference measurements96, NMR measurements of the chemical shift of a nucleus in the solute97 or measurements of the solvatochromism of a solute in the IR98 or UV-Vis spectral region.99 Plots of the data obtained from such measurements against the composition of the bulk solvent (usually as a mole fraction) depart clearly from the ideal behavior and the deviation is ascribed to the presence of preferential solvation. Several reported methods aim to quantify preferential solvation;96-100 none, however, provides an acceptable characterization facilitating a clear understanding of the phenomenon. If preferential solvation is so strongly dictated by the polar or ionic character of the solute, then characterizing a mixture of solvents by using a molecular probe will be utterly impossible since any conclusions reached could only be extrapolated to solutes of identical nature as regards not only polarity and charge, but also molecular size and shape. However, the experimental evidence presented below allows one to conclude that this is not the case and that solvent mixtures can in fact be characterized in as simple and precise terms as can a pure a solvent. In the light of the previous reasoning, describing the solvolysis of tert-butyl chloride or the decarboxylation kinetics of 3-carboxybenzisoxazole in mixed solvents in terms of SPP, SB and SA for the mixtures appeared to be rather difficult owing to the differences between the processes concerned and the solvatochromism upon which the scales were constructed. However, the results are categorical as judged by the following facts: (a) The solvolysis rate of tert-butyl chloride in 27 pure solvents and 120 binary mixtures of water with methanol (31 mixtures), ethanol (31), isopropyl alcohol (1), trifluoroethanol (8), dioxane (13), acetone (27) and acetic acid (9), in addition to the datum for the gas phase, all conform to the following equation:92 log k = 10.62(±0.44)SPP + 1.71(±0.22)SB + 7.89(±0.17)SA - 20.07(±0.34) [10.3.32]
with n = 148, r = 0.99 and sd = 0.40. (b) The decarboxylation rate of 3-carboxybenzisoxazole in 24 pure solvents and 36 mixtures of DMSO with diglyme (4 mixtures), acetonitrile (4), benzene (7), dichloro-
614
Javier Catalán
methane (6), chloroform (6) and methanol (9) are accurately described by the following expression: log k = 10.03(±1.05)SPP + 2.41(±0.49)SB - 5.73(±0.40)SA - 9.58(±0.83) [10.3.33]
with n = 60, r = 0.990 and sd = 0.60. Other evidence obtained in our laboratory using solvent mixtures and probes as disparate in size and properties as the cation Na+ and Reichardt’s ET(30), also confirm that solvent mixtures are no more difficult to characterize than pure solvents. REFERENCES C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, 2nd edn., VCH Publishers, Weinheim, 1988. 2 M. Berthelot, L. Péan de Saint-Gilles, Ann. Chim. Phys., 65, 385 (1862); 66, 5 (1862); 68, 225 (1863). 3 N.A. Menschutkin, Z. Phys. Chem., 5, 589 (1890). 4 E. Grunwald, S. Wistein, J. Am. Chem. Soc., 70, 846 (1948). 5 S. Wistein, A.H. Fainberg, J. Am. Chem. Soc., 79, 5937 (1957). 6 M.H. Abraham, R.M. Dogerty, M.J. Kamlet, J.M. Harris, R.W. Taft, J. Chem. Soc., Perkin Trans. 2, 913 (1987). 7 D.S. Kemp, K.G. Paul, J. Am. Chem. Soc., 97, 7305 (1975). 8 J.G. Kirkwood, J. Chem. Phys., 2, 351 (1934); 7, 911 (1939). 9 L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). 10 H. Block, S.M. Walker, Chem. Phys. Lett., 19, 363 (1973). 11 J.E. Brady, P.W. Carr, J. Phys. Chem., 89, 5759 (1985). 12 D.V. Matyushov, R. Schmid, B.M. Landansyi, J. Phys. Chem. B, 101, 1035 (1997). 13 C. Reichardt, Chem. Rev., 94, 2319 (1994). 14 R.S. Drago, J. Chem. Soc., Perkin Trans. 2, 1827 (1992). 15 J. Catalán, V. López, P. Pérez, R. Martin-Villamil, J.G. Rodriguez, Liebigs Ann., 241 (1995). 16 G. Taddei, E. Castelluci, F.D. Verderame, J. Chem. Phys., 53, 2407 (1970). 17. T.D. Epley, R.S. Drago, J. Am. Chem. Soc., 89, 5770 (1967); R.S. Drago, T.D. Epley, J. Am. Chem. Soc., 91, 2883 (1969); G.C. Vogel, R.S. Drago, J. Am. Chem. Soc., 92, 3924 (1970); R.S. Drago, L.B. Parr, C.S- Chamberlain, J. Am. Chem. Soc., 99, 3202 (1977); R.S. Drago, K.F. Purcell, Prog. Inorg. Chem., 6, 271 (1964); R.S. Drago, Coord. Chem. Rev., 33, 251 (1980); R.S. Drago, G.C. Vogel, J. Am. Chem. Soc., 114, 9527 (1992); R.S. Drago, Inorg. Chem., 32, 2473 (1993). 18 a) M.J. Kamlet, R.W. Taft, J. Am. Chem. Soc., 98, 377 (1976); b) R.W. Taft, M.J. Kamlet, J. Am. Chem. Soc., 98, 2886 (1976). 19 G.N. Lewis, Valence and Structure of Atoms and Molecules, The Chemical Catalog Co., 1923, p 142. 20 V. Gutmann, E. Vychera, Inorg. Nucl. Chem. Lett., 2, 257 (1966). 21 I.A. Koppel, V.A. Palm in Advances in Linear Free Energy Relationships (Eds. N.B. Chapman, J. Shorter) Chapter 5, Plenum Press, London, 1972, p 204. 22 E.M. Arnett, L. Joris, E. Michell, T.S.S.R. Murty, T.M. Gorrie, P. v R. Scheyer, J. Am. Chem. Soc., 96, 3875 (1974). 23 J. Catalán, J. Gómez, A. Couto, J. Laynez, J. Am. Chem. Soc., 112, 1678 (1990). 24 J. Catalán, P. Pérez, J. Elguero, W. Meutermans, Chem. Ber., 126, 2445 (1993). 25 E. Grunwald, S. Winstein, J. Am. Chem. Soc., 70, 846 (1948). 26 S. Winstein, A.H. Fainberg, J. Am. Chem. Soc., 79, 5937 (1957). 27 S. Winstein, E. Grunwald, H.W. Jones, J. Am. Chem. Soc., 73, 2700 (1951) 28 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 78, 2770 (1956). 29 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 79, 1597 (1957). 30 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 79, 1602 (1957). 31 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 79, 1608 (1957). 32 S. Winstein, A.H. Fainberg, E. Grunwald, J. Am. Chem. Soc., 79, 4146 (1957). 33 L.P. Hammett, J. Am. Chem. Soc., 59, 96(1937); and in Physical Organic Chemistry, 2nd Edition, McGraw-Hill, New York 1970. 34 A. Streitwieser Jr., Chem. Rev., 56, 617 (1956); Solvolytic Displacement Reactions, McGraw-Hill, New York 1962. 1
10.3 Solvent effects based on pure solvent scales
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87
615
T.W. Bentley, P.v R. Schleyer, Adv. Phys. Org. Chem., 14, 32 (1977). T.W. Bentley, G. Llewellyn, Prog. Phys. Org. Chem., 17, 121 (1990). C.G. Swain, R.B. Mosely, D.E. Bown, J. Am. Chem. Soc., 77, 3731 (1955). T.W. Bentley, F.L. Schad, P.v R. Schleyer, J. Am. Chem. Soc., 94, 992 (1972). F.L. Schad, T.W. Bentley, P.v R. Schleyer, J. Am. Chem. Soc., 98, 7667 (1976). T.W. Bentley, G.E. Carter, J. Am. Chem. Soc., 104, 5741 (1982). I.A. Koppel, V.A. Palm in Advances in linear Free Energy Relationships, N.B. Chapman, J. Shorter Eds., Plenum Press, London, 1972, p 203 and references therein. G.A. Allerhand, P.V. Schleyer, J. Am. Chem. Soc., 85, 374 (1963). L.J. Bellamy, H.E. Hallam, Trans. Faraday Soc., 55, 220 (1959). L.J. Bellamy, H.E. Halam, R.L. Williams, Trans. Faraday Soc., 54, 1120 (1958). L.J. Bellamy, R.L. Williams, Proc. Roy. Soc.(London), A255, 22 (1960). C. Walling, P.J. Wagner, J. Am. Chem. Soc., 86, 3368 (1964). C. Somolinos, I. Rodriguez, M.I. Redondo, M.V. Garcia, J. Mol. Struct., 143, 301 (1986). K. Dimroth, C. Reichardt, T. Siepmann, F. Bohlmann, Liebigs Ann. Chem., 661, 1 (1963). M.A. Kessler, O.S. Wolfbeis, Chem. Phys. Lipids, 50, 51 (1989); C.J. Drummond, F. Grieser, T.W. Healy, Faraday Discuss. Chem. Soc., 81, 95 (1986). D.C. Dong , M.A. Winnick, Can. J. Chem., 62, 2560 (1984). K.W. Street, W E. Acree, Analyst, 111, 1197 (1986). I. Kristjánsson, J. Ulstrup, Chem. Scripta, 25, 49 (1985). E.M. Kosower, J. Am. Chem. Soc., 80, 3253 (1958). E.M. Kosower, J.A. Skorez, J. Am. Chem. Soc., 82, 2188 (1960). T.R. Griffiths, D.C. Pugh, Coord. Chem. Rev., 29, 129 (1979). Y. Marcus, Chem. Soc. Rev., 409 (1993). L.G.S. Brooker, G.H. Keyes, D.W. Heseltine, J. Am. Chem. Soc., 73, 5350-54 (1951). L.G.S. Brooker, A.C. Craig, D.W. Heseltine, P.W. Jenkins, L.L. Lincoln, J.Am. Chem. Soc., 87, 2443 (1965). J.E. Dubois, A. Bienvenue, J.Chim. Phys., 65, 1259 (1968). R.S. Drago, J. Chem. Soc., Perkin Trans II, 1827 (1992). R.S. Drago, M.S. Hirsch, D.C. Ferris, C.W. Chronister, J. Chem. Soc., Perkin Trans II, 219 (1994). R.S. Drago, Apllications of Electrostatic-Covalent Models in Chemistry, Surface Scientific Publishers, Gainesville, 1994. E. Lippert, Z. Elektrochem., 61, 962 (1957). J. Czekalla, W. Liptay, K.O. Meyer, Z. Elektrochem., 67, 465 (1963). V. Baliah, M.K. Pillay, Indian J. Chem., 9, 845 (1971). J. Catalán, V. López, P. Pérez, Liebigs Ann., 793 (1993). J. Catalán, J. Org. Chem., 60, 8315 (1995). J. Catalán. New. J. Chem., 19, 1233 (1995). J. Catalán, C. Díaz, V. López, P. Pérez, R.M. Claramunt, J. Phys..Chem., 100, 18392 (1996). J. Catalán, C. Díaz, Eur. J. Org. Chem., 885 (1999). J.S. Murray, S. Rauganathan, P. Politzer, J. Org. Chem., 56, 3734 (1991). J. Catalán, C. Díaz, V. López, P. Pérez, J.L.G. de Paz, J.G. Rodríguez, Liebigs Ann., 1785 (1996). J. Catalán, J. Palomar, C. Díaz, J.L.G. de Paz, J. Phys. Chem., 101, 5183 (1997). J. Catalán, C. Díaz, Liebigs Ann., 1942 (1997). J. Catalán, E. Mena, W. Meutermans, J. Elguero, J. Phys. Chem., 96, 3615 (1992). J.E. Kuder, D. Wychick, Chem. Phys. Lett., 24, 69 (1974). I. Guda, F. Bolduc, J. Org. Chem., 49, 3300 (1984). S.F. Mason, J.Chem. Soc., 1240 (1959). J.A. Paéz, N. Campillo, J. Elguero, Gazz. Chim. Ital., 126, 307 (1996). M.J. Kamlet, J.L.M. Aboud, R.W. Taft, J.Am. Chem. Soc., 99, 6027 (1977); ibid, 99, 8325 (1977). U. Mayer, V. Gutman, W. Gerger, Monasth. Chem., 106, 1235 (1975). V.I. Minkin, O.A. Osipov, Yu A. Zhadanov, Dipole Moments in Organic Chemistry, Plenum Press, New York, 1970. W.R. Fawcet, A.A. Kloss, J. Phys. Chem., 100, 2019 (1996). C.S. Giam, J.L. Lyle, J. Am. Chem. Soc., 95, 3235 (1973). J. Catalán, J.C. del Valle, C. Díaz, J. Palomar, J.L.G. de Paz, M. Kasha, Internat.J. Quam. Chem., 72, 421 (1999). A.F. Lagalante, R.J. Jacobson, T.J. Bruno, J. Org. Chem., 61, 6404 (1996). K.M.C. Davis, J. Chem. Soc. (B), 1128 (1967).
616
Victor Cherginets
F. Schneider, E. Lippert, Ber.Bunsen-Ges. Phys. Chem., 72, 1155 (1968). J. Catalán, V. López, P. Pérez, J. Fluorescence, 6, 15 (1996). J. Catalán, C. Díaz, V. López, P. Pérez, R.M. Claramunt, Eur. J. Org. Chem., 1697 (1998). M.K. Singh, H. Pal, A.C. Bhasikuttan , A.V. Sapre, Photochem & Photobiol., 68, 32 (1998). J. Catalán, C. Díaz, F. Garcia-Blanco, J. Org. Chem., 64, 6512 (1999). J. Catalán, C. Díaz, F. Garcia-Blanco, J. Org. Chem., in press. G. Gritzner, J. Phys. Chem., 90, 5478 (1986). J. Emsley, N.J. Freeman, J. Mol. Struc., 161, 193 (1987). H. Scheneider, in Solute-Solvent Interactions, J.F. Coetzee and C.D. Ritchie (Eds), Dekker, New York, Vol 1 p 301 (1969). 97 L. S. Frankel, C.H. Langford, T.R. Stengle, J. Phys. Chem., 74, 1376 (1970). J.F. Hinton, E.S. Amis, Chem. Rev:, 67, 367 (1967). 98 A.I. Popov, in Soluto-Solvent Interactions, J.F. Coetzee and C.D.Ritchie(eds) Dekker, New York, 1976, Vol 2 pg 271. 99 K. Dimroth , C. Reichardt, Z. Anal. Chem., 215, 344 (1966) J. G. Dawber, J.Ward, R. A. Williams, J. Chem. Soc., Faraday Trans. 1, 84, 713 (1988). 100 J. Midwinter, P. Suppan, Spectrochim. Acta, 25A, 953 (1969); P. Suppan, J. Chem. Soc., Faraday Trans.,1, 83, 495 (1987); M. W. Muanda, J.B. Nagy, O.B. Nagy, Tetrahedron Lett., 38, 3424 (1974); O.B. Nagy, M.W. Muanda, J.B. Nagy, J. Phys. Chem., 83, 1961 (1979): H. Strehlow, H. Schneider, Pure Appl. Chem., 25, 327 (1971); M.S. Greenberg, A.I. Popov, Spectrochim. Acta, 31A, 697 (1975); H. Langhals, Angew. Chem. Int. Ed. Engl., 21, 724 (1982); A. Ben-Nain, J. Phys. Chem., 93, 3809 (1989); P. Chatterjee, S. Bagchi, J. Chem. Soc., Faraday Trans, 87, 587 (1991); W.E. Acree, Jr., S. A Tucker, D. C. Wilkins, J. Phys. Chem., 97, 11199 (1993); R.D. Skwierczynski, K.A. Connors, J. Chem. Soc. Perkin Trans 2, 467(1994) M. Rosés, C. Ráfols, J. Ortega, E. Bosch, J. Chem. Soc. Perkin Trans 2, 1607 (1995); W. E. Acree Jr., J. R. Powell, S. A. Tucker, J. Chem. Soc. Perkin Trans 2, 529 (1995). 88 89 90 91 92 93 94 95 96
10.4 ACID-BASE EQUILIBRIA IN IONIC SOLVENTS (IONIC MELTS) Victor Cherginets Institute for Single Crystals, Kharkov, Ukraine
Ionic melts are widely used in the science and engineering as media for performing different processes such as electrolysis, electrochemical synthesis, single crystals growing, etc. Practically complete dissociation of ionic media to the constituent ions creates high current densities at electrolysis. The absence of oxidants, similar to H+, makes it possible to obtain products, which cannot be obtained from aqueous solvents (i.e., alkaline and alkaline earth metals, sub-ions, etc.). From the ecological standpoint, molten ionic media are especially available as technological solvents since their employment does not cause the accumulation of liquid wastes because cooling to the room temperature transforms ionic liquids into a solid state. Processes taking place in ionic melt-solvents are considerably affected by impurities contained in the initial components of the melt or formed during preparation (mainly, melting) of solvents due to the high-temperature hydrolysis of melts or their interactions with container materials (Al2O3, SiO2, etc.) or active components of atmosphere (O2, CO2, etc.). The list of these impurities is wide enough and includes multivalent cations of transition metals, different complex anions (oxo- or halide anions). The effect of the mentioned admixtures on the processes in ionic melts depends mainly on the degree of their donor-acceptor interactions with constituent parts of the melt.
10.4 Acid-base equilibria in ionic solvents
617
10.4.1 ACID-BASE DEFINITIONS USED FOR THE DESCRIPTION OF DONOR-ACCEPTOR INTERACTIONS IN IONIC MEDIA Donor-acceptor interactions in ionic media are often described as acid-base interactions according to Lewis1 and Lux-Flood2-5 definitions. The classic variant of the former definition considers acids as acceptors of electron pairs and bases as their donors. In modern variant of this definition, acids are the electron pair (or anion) acceptors or cation (proton) donors, bases are the electron pair (anion) donors or cation (proton) acceptors. Lux-Flood definition considers bases as donors of oxide-ion, O2-, its acceptors are Lux acids. 10.4.1.1 The Lewis definition The Lewis acid-base process may be described by the following scheme: A + :B = A:B, K
[10.4.1]
where: A B K
an acid, a base, the equilibrium constant (used below pK ≡ -log K).
This definition can be used for the description of interactions in ionic melts, containing complex anions undergoing heterolytic dissociation. Processes reverse to [10.4.1], i.e., the acid-base dissociation of A:B adducts (ions) in some solvents are considered as acid-base equilibria of these solvents. Alkaline chloroaluminate melts containing the excess of AlCl3 vs. the stoichiometry 3+ 6,7 + the excess of are promising solvents for preparing sub-ions, e.g., Cd+(Cd 2+ 2 ), Bi , Bi 5 , 8 acid (AlCl3) favors their formation. Tremillon and Letisse, Torsi and Mamantov9,10 studied the acid-base properties of molten mixtures AlCl3-MCl (M=Li, Na, K, Cs) with AlCl3 concentration exceeding 50 mol% in the temperature range 175-400oC. Equilibrium molarities of Cl- were determined by a potentiometric method with the use of a chloride-reversible electrode. The solvents undergo the acid-base dissociation according to the following equation:
(
)
2AlCl 4− = Al 2 Cl 7− AlCl 3 ⋅ AlCl 4− + Cl − ,
pK
[10.4.2]
In the melts, chloride-ion donors were bases and substances increasing AlCl3 concentration were acids. The pK values decrease with the temperature (from 7.1 to 5.0 at temperatures 175 and 400oC, respectively, Na-based melt9) elevation and from Cs to Li (at 400oC pK were 3.8, 5.0, 5.8, 7.4 for Li-, Na-, K- and Cs-based melts, respectively10). The latter effect may be explained from the point of view of “hard” and “soft” acids and bases11-13 - Li-Cl complexes should be more stable than Cs-Cl ones as formed by “hard” base (Cl-) and “more hard” acid (Li+). Dioum, Vedel and Tremillon14 investigated molten KGaX4 (X=Cl, I) as background for acid-base processes. The following acid-base equilibria exist in the pure solvents:
(
)
2GaCl 4− = Ga 2 Cl 7− GaCl 3 ⋅ GaCl 4− + Cl − ,
pK=4.25±0.05
[10.4.3]
GaI 4− = GaI 3 + I − ,
pK=2.6±0.05
[10.4.4]
618
Victor Cherginets
Wagner15 reported an electrochemical study of molten NaBF4 at 420oC. The range of acid-base properties varying according to reaction: BF4− = BF3 + F − ,
pK=1.8
[10.4.5]
has been estimated as ~2pF units. The acid-base ranges in ionic melts containing halide complexes are relatively narrow, it means that varying their halide basicity (i.e., equilibrium concentration of the halide ions) may be performed in 2-4 logarithmic units, although such changes are enough to obtain some unstable compounds with the intermediate oxidation degrees.6,7 10.4.1.2 The Lux-Flood definition Oxygen-containing impurities in molten salts are most usual, their effect on technological processes is mainly negative and consists of bonding acidic reagents with the formation of insoluble (suspensions) or slightly dissociated products. Similar reactions result in retarding the main processes caused by the decrease of the equilibrium concentrations of initial reagents, forming oxide inclusions in metals obtained by the electrolysis of melts, inclusions of insoluble particles in single crystals, considerable corrosion of container for crystal growth, etc. Therefore, quantitative studies of reactions with participation of oxide ions in ionic melts are of importance for scientific and engineering purposes. Reactions with the transfer of oxide ions in ionic media are mainly considered as oxoacidity or acid-base equilibria by Lux-Flood:2-5 A + O-2 = B, K
[10.4.6]
The addition of acids (bases) in a melt as it follows from [10.4.6] leads to changes of the oxide ion activity, for the characterization of melt acidities (basicities) Lux2 proposed the oxygen index, or pO, which was similar to pOH (i.e., basicity index) in aqueous solutions: pO ≡ − log aO 2 −
( ≡ − logm ) O2−
[10.4.7]
where: a O 2 - , m O 2 - activity and molarity of oxide ions in the melt, respectively
The measurements of oxygen indices during various reactions in molten media define their thermodynamic characteristics - dissociation constants and solubility products. Most oxoacidity studies were made in nitrate melts, while there were much less communications dealing with the similar studies in higher-melting alkaline chlorides, molten bromides and iodides. There is a considerable scatter of obtained experimental results, which are often in conflict one with another. There were no serious attempts to generalize these data or consider them from the common point of view. 10.4.2 THE FEATURES OF IONIC MELTS AS MEDIA FOR ACID-BASE INTERACTIONS Ionic melts as media for Lux-Flood acid-base reactions may be divided into two types on the base of constitutional (i.e., being a part of main components of the melt) oxygen ions: oxygen-less and oxygen-containing ones. Let us consider some features of acid-base interactions in the mentioned melts.16
10.4 Acid-base equilibria in ionic solvents
619
10.4.2.1 Oxygen-less media There are no oxide ions in the composition of these melts, therefore the pure melts cannot possess oxide-donor properties. Real oxygen-less melts contain small amounts of O2- owing to inevitable ingress of oxygen-containing impurities into the melt, but oxide ion concentration in the “pure” melts is variable, depending on the concentration and acid-base character of impurities. For example, even considerable amounts of sulfates in melts do not create appreciable O2- concentration, carbonate ion dissociation is substantially stronger, hydroxide ions may be referred to most strongly dissociated Lux bases. In “pure” oxygen-less melts oxygen index pO is usually in the range 3 to 4.5. The employment of strongest purifying agents (HCl, CCl4, etc.) does not allow to decrease this concentration essentially, the latter being in “an unavoidable harm” causing errors at the quantitative investigations, especially when small amounts of acid or bases are studied. But, in some cases, oxygen index for such a “pure” melt has been used as the internal standard for construction of acidity scales.17-20 Quantitative studies of different Lux-Flood acids and bases in ionic melts were performed by two main ways: the construction of empirical acidity scales to estimate relative acidic strength of the substances and the determination of acid-base equilibria constants using potentiometric titration techniques. The first approach has been proposed in the classic work of Lux2 who obtained the acidity scale for the equimolar mixture of potassium and sodium sulfates. Although this work was the basis for series of later studies, the results cannot be considered as undoubted ones. Addition of acids (bases) to the mentioned melt led to increasing (decreasing) e.m.f. of cell with the oxide-selective electrode vs. the corresponding magnitude in the neutral melt. Then, the e.m.f. value shifted progressively to that for the neutral melt because of SO3 (Na2O) evaporation from the acidic (basic) melt. Therefore, Lux extrapolated values of e.m.f. to the point of acid (base) addition to the melt. This resulted in the decrease of the data2 accuracy. The empirical acidity scales17-20 give some information about strength of the acids and bases in melts studied. The principal error related to the term “acidity scale” in the case of oxygen-less melts is because it is not connected with the melt properties. Therefore, the values obtained could not be considered as the quantitative characteristics of the oxygen-less melts. The acidity scale length in a solvent is believed to be the interval (measured in acidity index units) between standard solutions of strong acid and base. If the solvent possesses its own acid-base autodissociation equilibrium then a substance creating unit concentration of acid (base) of solvent in the standard solution should be considered as the strongest acid (base). Addition of stronger acids (bases) should not result in extension of the acidity scale because of the known phenomenon of leveling acidic and basic properties by the solvent.21 Oxygen-less melts do not possess acid-base equilibrium, therefore, values of the basicity index, in the “pure” melts may have any reasonable value. 10.4.2.2 Oxygen-containing melts Acid-base processes in oxygen-containing melts are more complex than those in oxygen-less ones, since they are accompanied by competitive equilibria of own acid-base autodissociation of the melt-solvent. The coexistence of acidity and basicity “carriers” into melt is the characteristic feature of oxygen-containing melts making them similar to low-temperature molecular solvents with own acid-base equilibrium. But, there exist some principal features due to relatively high temperatures of the liquid state. Own acids of the melts are often unstable or volatile,2,22 therefore, acidic solutions,
620
Victor Cherginets
as a rule, lose acids because of their evaporation2 or decomposition,22 the oxygen index of such melts is shifted progressively to that of the neutral melt. Hence, in oxygen-containing melts it is possible to observe not only the leveling of acidic properties but also an upper limit of acidity. CrO3 and MoO3 have been found23 to breakdown the nitrate melt KNO3 with the evaporation of NO2. It should be noted, that the temperature elevation leads to the efficient shrinkage of the acidic region in the melt. The interactions between the oxides and the melt have been assumed to result in the formation of nitrogen (V) oxide, N2O5, and corresponding oxoanions. Since the acids studied (CrO3, MoO3) may be referred to as very strong acids, it may be supposed that reaction: 2CrO 3 + NO 3− = Cr2 O 72 − + NO +2
[10.4.8]
should be completely shifted to the right and the strength of the acid is determined by acidic properties of nitronium cation, NO2+. The stability of the latter is low and after achieving a certain concentration its reaction with the melt anions becomes intensive enough: 1 NO +2 + NO 3− = 2NO 2 ↑ + O 2 2
[10.4.9]
The so-called kinetic methods of melt acidity determination, which will be considered below, are based just on this reaction. The reaction [10.4.9] leads to the decrease of nitronium concentration and the rate of NO2 emission decreases until essentially constant acid concentration is determined by a sequence of consecutive measurements. This concentration is the “upper limit” of acidity of the melt. The thermal dependence of upper limit of acidity23 in nitrate melts can be easily explained on the basis of the increase of the melt temperature, which leads not only to the reduction of nitronium stability but also to the elevation of process [10.4.9] rate. Sulphate melts have the upper limit of acidity too, it seems connected with limited and low solubility of SO3 at elevated temperatures, such assumption may be confirmed by results.2 10.4.2.3 The effect of the ionic solvent composition on acid-base equilibria The equilibrium parameters of Lux acid-base reactions in ionic media (solubility products of oxides and acid-base equilibrium constants) are essentially affected by the acidic properties of the molten alkaline halide mixtures, i.e., they are dependent on the constituent cation acidities. Therefore, one should consider the reverse problem - the estimation of the basicity indices of ionic melts on the basis of the calculated equilibrium constants. The default acid-base processes are actually a superimposed effect of interactions between oxide-ions formed and the most acidic cations of the melt: iMe m+ + O 2 − ↔ Me i O i⋅ m − 2 ,
Kl,i
[10.4.10]
where: Mem+
the most acidic cation of the solvent
The increase of melt acidity (Kl,i) leads the shift of interactions [10.4.10] to the right. The distribution of oxide ions added to the melt between different complexes with the melt cations may be presented by the following equation (N is mole fraction, m - molarity):
10.4 Acid-base equilibria in ionic solvents
621
n i NO0 2 − = NO 2 − 1 + ∑ K l, i NMe + i =1
[10.4.11]
It may be seen that the ratio “free oxide-ion/total oxide-ion” is the constant which may be designated as Il:24 I l = NO 2 − / NO0 2 −
(= m
O
2−
/ mO0 2 −
) = 1/ 1+ ∑ K n
i =1
l, i
i NMe +
[10.4.12]
Now let us to estimate the equilibrium molarity of the constituent acidic cations in the melt, e.g., the eutectic KCl-LiCl melt (0.4:0.6) contains ~8.5 mole of Li+ per 1 kg. Usually the ionic complexes in melts are characterized by the coordination number ~ 4-6.21 For the solution of O2- of the 0.1 mole/kg concentration, the maximum possible quantity of fixed Li+ concentration may be estimated as 0.4-0.6 mole/kg, i.e., efficiently lower than 8.5. In this case the change of actual Li+ concentration is approximately equal to 5-7% and mMe+ in this case may be suggested as constant. Therefore, for each melt the sum in the denominator of [10.4.12] is the constant reflecting its acidic properties. So, pIl= -logIl is a measure of melt acidities and may be denoted as the “oxobasicity index” of the melt. Since the determination of the “absolute” concentration of free O2- is practically impossible one should choose the “standard melt”, for which Il is conditionally equal to 1 and pIl=0. It is reasonable to choose the equimolar mixture KCl-NaCl as the “standard melt”, since this melt is most frequently investigated. Further, one should choose “standard equilibria” and formulate the non-thermodynamic assumptions which usually postulate that the constant of the “standard equilibrium” calculated using “absolute” oxide ion concentrations remains the same for all other melts. Now let us consider possible variants of such “standard equilibria” and conditions of their use. One of the first attempts to define and estimate the oxoacidity parameters of ionic (chloride-based) melts was connected with the studies of the equilibrium H2 O gas + 2Cl − ↔ 2HCl gas + O 2 −
[10.4.13]
in different chloride mixtures.25,26 Here the partial pressures of H2O and HCl are known and the oxide ion concentration is calculated on the base of the calibration data as m 0O 2 - ( N0O 2 - ). The equilibrium constant of [10.4.13] may be represented in terms of the “absolute” mole fractions of O2- and Il by: K1 =
(
2 p HCl NO 2 − I l−1
p H 2O N
2 Cl −
) =K
−1 KCl − NaCl l
I
[10.4.14]
and, consequently: pK l = pK KCl − NaCl − pI l Combes et al.25 introduced the oxo-acidity function, Ω:
[10.4.15]
622
Victor Cherginets
Ω = 14 − pK l + pO 2 −
[10.4.16]
taking into account [10.4.15] eq. [10.4.16] Ω = (14 − pK KCl − NaCl ) + pI l + pO 2 −
[10.4.17]
where (14 - pKKCl-NaCl) is the Ω value in KCl-NaCl standard solution (pO=0) and pO2- is the instrumental pO in the solvent studied. Therefore, the shift of pO scales vs. KCl-NaCl depends only on the Il value. On the basis of [10.4.13] it has been estimated as +7-8 log units (Figure 325) for KCl-LiCl at 1000K. Homogeneous Lux acid-base equilibria of type of [10.4.6] have not been earlier considered as available for estimations of oxoacidity indices. However, pK for these reactions may be written as pK1 = − log
(
NB
NA NO 2 − I l−1
)
= pK KCl − NaCl + pI l
[10.4.18]
For the use of this type of equilibrium it is necessary to make the following non-thermodynamic assumption: γB γA
γB = l γ A KCl − NaCl
[10.4.19]
where: γ
corresponding activity coefficients of the acid and the base in solvent “l” and KCl-NaCl
The use of [10.4.16] where A and B are anion acid and base for estimations of the oxobasicity indices may be justified because the acid and the conjugated base are negatively charged (in the pair Cr2O7-2/CrO4-2 they are of the same charge). Finally, let us consider the usability of the oxide solubility data MeOs = Me 2+ + O 2 −
[10.4.20]
for pIl estimations. The solubility product value, P, may be presented as
(
)
PMeO, l = NMe 2 + NO 2 − I l−1 = PMeO, KCl − NaCl I l−1
[10.4.21]
pPMeO, l = pPMeO, KCl − NaCl − pI l
[10.4.22]
or
Metal cations in molten halides form halide complexes. The reaction [10.4.20] for chloride melts (NCl- =1) suggests that the distribution of Me2+ between different complexes MeCln2-n remains unchanged (similarly to eq. [10.4.10]-[10.4.12] for the oxide ion distribution). It is clear that the solubility data may be used for estimations of the oxoacidity indices only in melts with the same anion composition. The anion changes cause errors since Me2+ in eq. [10.4.21] is referred to essentially different acids, e.g., halide complexes MeCl42- and
10.4 Acid-base equilibria in ionic solvents
623
MeBr42- in the case of chloride and bromide melts. Therefore, regardless of the cation composition, the oxide solubilities are essentially affected by anion composition of a melt.27 Equations [10.4.14], [10.4.18] and [10.4.22] estimate pIKCl-LiCl vs. KCl-NaCl as 7-8,25 24 3.7 and 3.424, respectively. This estimation is considerably larger than the estimation based on eq. [10.4.18] and [10.4.22]. For Ca2+ based chloride melts, the similar estimation made on the basis of [10.4.14] gave pI~1025 vs. ~4 determined on the basis of [10.4.22].28 So, the oxobasicity indices calculated according to HCl/H2O equilibrium constant are essentially different from those obtained using reactions without reactive gas atmosphere. Probably, this discrepancy may be explained by the features of the water behavior in aprotic ionic melts. It is known that, similar to basic properties, water possesses oxoacidic properties according to the reaction:
(
H2 O + O 2 − ↔ 2OH − or (OH) 2
−2
)
[10.4.23]
Such polynuclear complexes are stable according to spectral data29 because of the H-bonds formation. Three acids (HCl + Me+ + H2O) exist in ionic melts saturated with water and HCl and consequently reaction [10.4.23] depends on the partial pressure of water. In Liand Ca-based chloride melts, retention of water30 and the solubility of H2O is appreciably higher than in e.g. KCl-NaCl. Apparently, even the simplest reaction [10.4.13] is actually complicated by additional interactions favoring the fixation of “free” oxide ions. Values of oxobasicity indices calculated from eq. [10.4.13] are thus somewhat overestimated. The oxobasicity index for KCl-LiCl (0.4:0.6) compared with KCl-NaCl at 700oC lies within the range 3.4-3.7 log units. 10.4.3 METHODS FOR ESTIMATIONS OF ACIDITIES OF SOLUTIONS BASED ON IONIC MELTS The literature data show that many methods were used for oxoacidity studies and estimation of the acidic properties of melts. One simplest method31 involves indicator. Acid-base indicators usually employed in aqueous solutions for protic acidity measurements have been used for acidity studies in molten KNO3-LiNO3 at 210oC and KSCN at 200oC. The color of indicator solution, relative to acidity, changes during titration of bases (sodium hydroxide or peroxide) by potassium pyrosulfate, K2S2O7: 2HInd + O 2 − = 2Ind − + H 2 O
[10.4.24]
where: HInd Ind-
the protonized form of the acid-base indicator the anionic form of the said indicator
In molten nitrates such transition was observed only for phenolphthalein (yellow-purple), other indicators seemingly were oxidized by the melt (the conclusion made31 was “insoluble in the melt”). In molten KSCN not possessing oxidizing properties color transitions were observed for all indicators used (methyl red, thymolphthalein, etc.). The employment of indicators for acidity estimation is limited mainly by their thermal instability and tendency to oxidize at high temperatures. Since ionic melts, as a rule, have no constitutional water, the reverse transition of indicator into the protonized form [10.4.24] is hardly possible (the solution of phenolphthalein became yellow at the reverse transition “base-acid”31).
624
Victor Cherginets
Therefore, the use of indicator method is possible only at low-temperatures (250oC) and in non-oxidizing ionic melts. Spectral methods were also used for the melt basicity estimation.32,33 The scale of “optical” basicity for metallurgical slugs with respect to CaO (standard) has been constructed.32 The “optical” basicity, Λ, was estimated from spectral line shifts (the transition 1S0 → 3P1) for Tl+, Pb2+, Bi3+ cations in the basic medium. Ionic melts Na2O-B2O3-Al2O3, Na2O-B2O3SiO2, Na2O-SiO2-Al2O3 have been investigated by X-ray fluorescent spectroscopy,32 relative oxide acidities decrease in the sequence B2O3>SiO2>Al2O3. The use of spectral methods is based on studies of cooled (quenched) samples, such a routine may distort results because of inconsistency between the solution temperature and that of acidity estimations. Data on the melt basicity estimations by the determination of acidic gases (SO3, CO2, H2O) solubilities are presented elsewhere.34,35 Sulphur trioxide solubility in molten sodium phosphate was determined by thermogravimetric analysis,34 the correlation between the melt basicity and the SO3 solubility was found. The use of CO2 and H2O for basicity estimation was described elsewhere.35 The similar methods may be used only for basic melts since acids displace the acidic gas from the melt. Furthermore, the interaction products of the acidic gas with the melt may be relatively stable, especially in basic solutions, and their formation leads to irreversible changes of the melt properties. A “kinetic method” was used for Lux acidity studies in molten nitrates. The interaction between the nitrate melts KNO3-NaNO3 and potassium pyrosulfate was studied.36,37 S 2 O 72 − + NO 3− = 2SO 42 − + NO 2+
[10.4.25]
The formed nitronium cation, NO2+, reacted with NO3- according to [10.4.9]. The latter stage is considerably slower than [10.4.25], hence, its rate allowed to estimate the melt acidity, which can be presented by the following sum: TA = [S2O72-] + [NO2+].36 The acid-base interactions between Cr2O72- and chlorate ions, ClO3-, in molten KNO3-NaNO3 are described by the following equations.38,39 Cr2 O 72 − + ClO 3− = 2CrO 42 − + ClO +2
[10.4.26]
ClO +2 + Cl − → Cl 2 ↑ +O 2 ↑
[10.4.27]
From two above processes, reaction [10.4.27] is the limiting stage, the rate of the chlorine evolution is proportional to the total acidity of the melt: TA = [Cr2O72-] + [ClO2+]. The reaction of bromate ions, BrO3-, with potassium dichromate in the nitrate melt has been studied.38 Several studies36-40 gave estimate of relative acidities of the oxo-cations. The acidities increased in the sequence BrO2+
[10.4.28]
10.4 Acid-base equilibria in ionic solvents
Me2+
625
Cu2+ or Co2+
Metal cation acidities are comparable with acidic properties of dichromate-ion and mentioned above36-40 cations. A potentiometric method, or more accurately, method of e.m.f. measurements with liquid junction in cells with oxide-reversible electrodes was most advantageous in oxoacidity studies of various melts. The principal construction of electrochemical cells for oxoacidity studies may be presented by the following scheme: Reference electrode || Melt+O2-| Oxide-reversible electrode [10.4.29] In all known studies, the potential of liquid junction was zero. The known reference electrodes were metallic (i.e., metal immersed into the solution with the definite concentration of the corresponding cation) or gas oxygen electrodes immersed into solutions with definite oxide ion concentrations (concentration cells). 10.4.4 ON STUDIES OF THE HOMOGENEOUS ACID-BASE REACTIONS IN IONIC MELTS 10.4.4.1 Nitrate melts Traditionally, nitrate melts are considered among most comprehensively studied ones. Relatively low melting temperatures both of individual nitrates and their mixtures allowed to use simpler experimental techniques. The potentiometric studies gave the acid-base equilibrium constants. Most studies were on oxoacidity reactions in molten KNO3 at 350oC.23,43-67 Constants of acid-base reactions of Group V highest oxides (P2O5, As2O5, V2O5) with Lux bases were also investigated.43-48 All oxides breakdown the melt, the first stage is the formation of corresponding meta- acid salts and nitronium cation: R2 O 5 + NO 3− = 2RO 3− + NO +2
[10.4.30]
where: R
the designation of Group V element,
the second step is the redox interaction of NO2+ with NO3- according to [10.4.9], which is shifted to the right, NO2 evolution took place until e.m.f. values reached down to ~0.5 V, this magnitude corresponded to the upper limit of oxo-acidity in molten KNO3. The oxygen index value for this limit can be estimated as ~10-16.43-48 Addition of Na2O2 as a Lux base led to the following neutralization steps 2RO 3− + O 2 − = R2 O 74 −
[10.4.31]
R2 O 74 − + O 2 − = 2RO 43 −
[10.4.32]
The acidic properties of oxides increased in the sequence V2O5
[10.4.33]
2HRO 42 − + O 2 − = 2RO 43 − + H2 O ↑
[10.4.34]
626
Victor Cherginets
Since even the pure salts containing mentioned hydroanions undergo the decomposition at temperatures much lower then 350oC with the formation of corresponding meta- or pyro- salts,68 equilibria [10.4.33] and [10.4.34] hardly take place in the absence of water vapor in atmosphere over the melt. The acidic properties of highest oxides of VI Group elements (Cr, Mo, W) were investigated.23,48,49 Similarly to Group V oxides43-48 CrO3 and MoO3 destructed the nitrate melt with formation of Cr2O72- and Mo3O102-, respectively. After the complete evolution of NO2, e.m.f. values were close to 0.5 V, what was in a good agreement other studies.43-48 The potentiometric titration curves of the formed products contained only one drop of e.m.f. (or pO). Chromate and molybdate ions were the final products of the neutralization reactions. WO3 was weaker acid and did not destruct the nitrate melt. Its titration was the one-stage process resulting in the tungstate formation. The acidities of oxides increased in the order: WO3<MoO3
[10.4.35]
The cations studied were arranged in the following sequence: K~Na>Li~Ba>Sr>Ca>Pb of basicity decrease. The cation sequence was in a good correlation with the electronegativity of metals. The proposed method of investigation had an essential problem since the acids used for titration fixed oxide ions with the simultaneous destruction of carbonate and formed insoluble chromates and phosphates with the majority of cations studied. As it has been shown,36,37 the addition of Ca2+ into nitrate melts leads to the shift of the interaction Cr2O72- - NO3- to formation of nitronium cation and the melt decomposition although the solution without Ca2+ is stable. Such shift is explained by formation of slightly insoluble CaCrO4 in nitrate melt and has no relation to Ca2+ acidity. It is referred to the barium position in the sequence that gives evidence of the said complexes or precipitates formation, since their acidic properties are substantially weaker than those of Li+. The formation of the BaCrO4 precipitate leads to approaching of Ba2+ acidity to Li+ acidity. Probably, the same reason effected arrangement of others alkaline earth metals in this sequence. The relative strength of different Lux bases in nitrate melts was investigated.54,55 The bases studied were divided into two groups: the oxide ion group (OH-, O2-, O22-) and the carbonate ion group (CO32-, HCO3-, (COO)22-, CH3COO-, HCOO-). The bases belonging to the first group have been assumed54,55 to be completely transformed into O2-.The members of the first group of bases were weaker than the second group bases. From the thermal data it may be seen that at experimental temperature (350oC) all the second group bases are completely broken down to carbonate. The differences between single bases may be attributed to reduction properties - organic salts reduce nitrate ions to more basic nitrite ones and favor the accumulation of oxide and carbonate ions in the melt over the stoichiometric ratio.
10.4 Acid-base equilibria in ionic solvents
627
Hence, the use of a majority of proposed bases for quantitative studies of oxoacidity is not possible, especially for acids having apparent oxidizing properties (chromates, vanadates, etc.). Oxide ions in molten nitrates exist as the so-called pyronitrate ions N2O74- which was a strong base with the basicity approximately equal to those for CO32-.55 The modern point of view excludes the existence of nitrogen compounds with the coordination number exceeding 3 by oxygen, hence pyronitrate was hardly existing in the mentioned conditions. Indeed, following oxoanion studies51 have demonstrated the absence of the mentioned ion in basic nitrate solutions. A possible way for the stabilization of O2- in nitrate melts is the formation of silicates due to reactions with walls of the container (usually it is pyrex). The anomalous behavior of the electrode Pt(O2) has been noted and explained by “peroxide” function of the gas oxygen electrodes (of the first type).57,58 On the contrary to gas electrodes, metal-oxide electrodes (of the second type) were reversible with the slope corresponding to the reaction: 1 / 2O 2 + 2e = O 2 −
[10.4.36]
The acid-base equilibria in complex solutions based on nitrates containing phosphates and molybdates were attributed59,60 to the formation of heteropolyacids in molten nitrates. This is confirmed by cryoscopic studies.61 Some features of potentiometric titration techniques in molten salts have been discussed.62,63 The automatic potentiometric titration described in details62 consists of the following. A rod, prepared as frozen K2Cr2O7 solution in the melt, was immersed into the melt containing a base with the constant rate. e.m.f. data were recorded by the recording potentiometer. Such a routine may be used only for the detection of equivalence points in melts (with accuracy of ~3.5 %). The effect of acidic oxides on the thermal stability of molten NaNO3 has been investigated.64 The increase of oxide acidity results in the decrease of the melt decomposition temperature. This is explained by the reduction of the activation energy of the decomposition owing to fixing oxide ions by acids. But there was also another explanation: decreasing temperature of melt decomposition is caused by the equilibrium shift towards the oxide-ions and nitrogen oxide formation under the action of the acid. Acidic properties of oxides decrease in the sequence SiO2>TiO2>ZrO2>Al2O3>MgO. Potentiometric method65-67 used metal-oxide electrodes (Nb|Nb2O5, Ta|Ta2O5, Zr|ZrO2) for oxoacidity studies in the nitrate melt. The empirical acidity scale in molten potassium nitrate at 350oC was constructed: NH4VO3>NaPO3>NaH2PO4>K2Cr2O7>K2HPO4> Na4P2O7>NaHAsO4>K2CO3>Na2O2. This sequence includes some acids seldom existing in molten salts. NH4VO3 at 350oC should be completely transformed into V2O5 with a partial reduction of the latter by ammonia. Acidic salts mentioned were decomposed to corresponding pyro- and meta-salts. Therefore the scale obtained contains a number of errors. 10.4.4.2 Sulphate melts Sulphate melts are referred to melts with the own acid-base dissociation equilibrium: SO 42 − = SO 3, l ↑ +O l2 −
[10.4.37]
628
Victor Cherginets
There are rather few contributions devoted to the Lux acidity studies in the similar melts. This may be explained by relatively high melting points both of single sulfates and their eutectic mixtures. The weak stability of acidic solutions in sulphate melts can also be considered as a reason why these melts are studied insufficiently. Indeed, reaction [10.4.37] should result in SO3 formation in the melt. Reference data,68,69 however, show that decomposition temperatures of pyrosulfates (i.e., complexes SO3(SO42-) do not exceed 460oC, in dissolved state stability of solvated SO3 should be even lower. The absence of SO3 partial pressure over the melt gives rise to removing SO3 from the latter and, since the equilibrium state of [10.4.37] is not attained, SO3 should be removed completely. Among sulphate melts the ternary eutectic mixture K2SO4-Li2SO4-Na2SO4 (0.135:0.78:0.085) with the lowest melting point 512oC70 may be most promising for the acidity studies since it is possible to perform studies at the temperatures significantly lower than in the classic work of Lux.2 Under these conditions stability of acidic and basic solutions increases. Lux2 reported the oxygen electrode reversibility investigation and construction of empirical acidity scale in the molten eutectic mixture K2SO4-Na2SO4 at 800oC. The thermal stability of pyrosulfates has been shown by Flood3 to decrease with melt cation acidity increasing. Kaneko and Kojima71,72 investigated acid-base properties of different solutions in the molten K2SO4-Li2SO4-Na2SO4 at 550oC; the empirical acidity scale length was approximately 10 pO units between 0.01 mole/kg solutions of S2O72- and O22- ions. From the latter data, the acid-base interval of this solvent can be easily estimated as the length of acidity scale between standard solution - ~14 pO units. The pO range of VO2+ stability has been determined to be 8.5 to 10.6. To complete the consideration of investigations in sulphate melts the work73 should be mentioned, it is concerned with potentiometric studies of the acidic properties of MoO3, Cr2O72-, PO3-, P2O74-, V2O5 in molten K2SO4-Li2SO4-Na2SO4 at 625oC. Acids mentioned were neutralized by sodium carbonate. The potential (pO) drop in equivalence points was observed at “acid-base” ratio 1:3 for vanadium oxide and 1:1 for all other Lux acids. Values of equilibrium constants were not presented. 10.4.4.3 Silicate melts Studies performed in molten silicates are of importance for applied purposes, since systems under consideration are widely used in various industries (slugs, glasses, etc.). Flood et al.74-76 studied the acidity dependence of silicate melts upon their composition. A system PbO-SiO2 have been studied in the range of SiO2 concentrations of NSiO 2 from 0 to 0.6 at 1100-1200oC. The coexistence of (SiO3)36- and (SiO2.5)66- polyanions together with the ordinary orthosilicate ions, SiO44- has been shown. The basicities of molten glasses vs. Na2O/SiO2 ratios were studied75 by the potentiometric method with the use of the gas oxygen electrode Ag(O2) in concentration cells. The decrease of the mentioned ratio has been stated to cause a reduction of the basicity, while magnesium oxide has been found to have no effect on the melt basicity. The gas oxygen electrode reversibility in molten PbO-SiO2, Na2O-CaO-SiO2, MeO-PbO-SiO2, where Me2+ - alkaline earth cation, has been investigated.77 The electrode Pt(O2) has been shown to be reversible to O2-. The use of a membrane oxygen electrode Pt(O2)|ZrO2 for molten silicate acidity control during glass making has been described.78 Studies of basicities in molten Na2O-Al2O3-SiO2 using a potentiometric method79 showed that the acidic properties of Al2O3 were weaker, then SiO2 ones. Processes of glass corrosion in molten alkaline earth nitrates have been studied for glass compositions Na2O⋅xAl2O3⋅2SiO2, where x was in the range from 0 to 0.4.80 The degree of interaction between the melt and immersed glass in-
10.4 Acid-base equilibria in ionic solvents
629
creased with the reduction of the alkaline earth cation radius. The effect of the cation charge and the radius on the basicity of lead meta-silicate melt at 800, 850, 900oC by additions of Tl2O, PbO, CdO, ZnO and Bi2O3 has been studied by potentiometry.81 10.4.4.4 The equimolar mixture KCl-NaCl This melt is the most investigated among chloride based melts. Some oxoacidity studies were performed by Shapoval et al.82-87 Main purposes of these works were to investigate oxygen electrode reversibility and to obtain equilibria constants for acid-base reactions including oxo-compounds of CrVI, MoVI and WVI at 700oC. Equilibrium constants of acid-base interactions for CrO3 and PO3- have been determined.82-84 The titration of the former substance proceeds in two stages: −1
. 88 ×10 . 4 ×10 → Cr2 O 72 − 1 → CrO 42 − CrO 3 2 3
[10.4.38]
Sodium metaphosphate is two-basic acid too: 2. 88±1. 2 ) ×10 2. 5±1. 4 ) ×10 → P2 O 74 − ( → PO 43− PO 3− ( 3
1
[10.4.39]
The excess of titrant, O2-, has been found to cause the formation of basic phosphates with assumed composition PO3-⋅2O2-. When Na2CO3 was used as titrant, there was no formation of the mentioned basic products. Ba2+ and Li+ cations have been shown to possess appreciable acidic properties, corresponding constant values were estimated as 8.1×101 and 3.53×102. An investigation of acidic properties of MoO3 was performed:85 ×10 ±880 MoO 3 5 → Mo 2 O 72 − 880 → MoO 42 − 4
[10.4.40]
Acidic properties of oxo-compounds of WVI have been studied.86,87 The titration of WO3 in molten chlorides was a two-stage process: ±390 ±880 → W 2 O 72 − 880 → WO 42 − WO 3 1460
[10.4.41]
Results formed the base for the following electrochemical studies of electroreduction processes of Group VI metals.88-90 These studies developed theoretical bases and principles to control electrochemical processes of metals and their compounds (carbides, borides, silicides) deposition from ionic melts.81 The acid-base equilibria in the scheelite (CaWO4) solutions in the said chloride mixture at 1000K were investigated.92 CaO solubility in molten KCl-NaCl has been determined to be about 0.084 mol%, that of scheelite was 10-3.5 mole/kg. Equilibrium constants for acid-base reactions with WO3 participation have been determined, too: → WO 42 − WO 3 10 10
[10.4.42]
The reverse titration has been assumed by authors92 to result in the formation of polyanion W3O102-, pK=12.7. The titration curve was irreversible. This result can be easily explained taking into account that WO3 is unstable in strong acidic chloride melts and removed from as WO2Cl2. Therefore, additions of WO3 into the acidic melt should lead to un-
630
Victor Cherginets
controlled titrant loss affecting experimental results. Ditto referred to all results obtained by the reverse titration by MoO3 and CrO3 as the equivalence point. Acidic properties of PV oxocompounds in the chloride melt at 700oC were investigated using a membrane oxygen electrode Ni,NiO|ZrO2.93 Polyphosphates of compositions with Na2O-P2O5 ratio from 1.67 to 3 formed in the acidic region: Pn O 3−n2+(1n+2 ) + PO 43 − = Pn+1O 3−n( +n5+5 )
[10.4.43]
The corresponding constants were estimated. Polyphosphate solubilities were determined by a cryoscopic method. The titration of V2O5 proceed in two stages:94 = −9. 3±0. 3 = −8. 3±0. 3 V2 O 5 pK → VO 3− pK → V2 O 74 −
[10.4.44]
Some results of oxoacidity studies in molten KCl-NaCl were obtained in our works,95-99 they are presented in Table 10.4.1. Table 10.4.1. The Lux acid-base equilibrium constants in molten KCl-NaCl and NaI (at the confidence level 0.95) [After references 95,96,98] -pK
Equilibrium
m
N
mol%
8.01±0.1
10.67
6.67
5.93±0.1
7.26
5.26
KCl-NaCl 2PO + O 3
− 3
PO + O
2−
2−
− 3
Cr2 O
2− 7
2CrO
= PO
= [ PO ⋅ O ]
7.24±0.1
-
-
2−
2− 4
3− 4
8.41±0.1
9.77
7.77
7.18±0.1
7.18
7.18
1.60±0.2
4.26
0.26
8.32±0.2
9.65
7.65
9.71±0.3
11.37
7.37
9.31±0.2
10.64
8.64
10.67±0.5
13.33
9.33
4.82±0.1
2.16
4.16
2.37±0.2
3.17
1.70
6.95±0.2
6.95
6.95
12.23±0.1
14.89
10.89
12.30±0.1
14.96
10.96
13.88±0.5
19.20
11.20
4.18±0.4
6.84
2.84
= CrO
+O
2− 4
2−
+O
MoO3 + 2O 2−
2− 7
2−
B4 O − 2
+O
BO + O
2−
V2 O5 + O
2−
V2 O5 + 3O
2−
V2 O5 + 5O
2−
2GeO2 + O
2−
2− 4
2−
2− 4
2− 4 2−
= [WO ⋅ O ] = 4BO
− 2
3− 3
= 2VO
V2 O5 + 2O
2−
2− 4
= BO
2−
2− 4
= [MoO ⋅ O ]
= WO
WO3 + 2O
2− 4
= [2CrO ⋅ O ]
= MoO
2−
2−
2−
= 2CrO
2−
2−
MoO3 + O WO3 + O
3− 4
2−
PO + 2O CrO3 + O
= P2 O
4− 7
− 3
= V2 O
4− 7
= 2VO
3− 4 3− 4
2−
= 2[VO ⋅ O ] = Ge2 O
2− 5
10.4 Acid-base equilibria in ionic solvents
Equilibrium
631
-pK m
N
mol%
NaI − 3
2VO + O
2−
4− 7
2−
V2 O
B4 O
2− 7
+O
+O
2−
= V2 O
4− 7
= 2VO
5.40±0.3
7.10
4.10
3− 4
1.68±0.3
1.68
1.68
− 2
5.02±0.8
3.37
5.37
= 4BO
The results included in the Table 10.4.1 are in good agreement with the similar studies, performed using a membrane oxygen electrode Ni,NiO|ZrO2. The use of the gas platinum-oxygen electrode resulted in significant underrating of constant values (of order of 3-4 pK units). 10.4.4.5 Other alkaline halide melts Some studies of acid-base equilibria in molten alkaline halides have been performed by Rybkin et al.17-20 The empirical acidity scales have been constructed on the base of potentiometric studies in molten KCl17,18 and CsI.19 The practical significance of the mentioned works consists in the proposition to use some buffer solutions - SO42-/S2O82-, WO42-/W2O72-, PO43-/P2O74- as reference standards for the indicator electrode calibration and determinations of pO in molten alkaline halides. Rybkin and Banik20 obtained the empirical acidity scale in molten NaI at 700oC. The basicities of CO32- and OH- were close which is incorrect since 0.01 mole/kg solutions were used for the scale constructing while equations 1 1 OH − = H2 O + O 2 − 2 2
[10.4.45]
CO 32 − = CO 2 ↑ + O 2 −
[10.4.46]
showed that these additions were not equivalent. CO32- concentration recalculated to oxide ions was twice as high as OH concentration. Indeed, in the later work these authors100,101 showed that hydroxide ion was stronger base than carbonate ion. We have performed a study of acidic properties of boron (III) and vanadium (V) oxo-compounds in molten NaI at 700oC98 (see Table 10.4.1). A comparative study of strength of Lux bases OH-, CO32-, SO42- was determined100,101 by potentiometric titration using sodium pyrophosphate as acid. Two moles of the first base may be neutralized by 1 mole of P4O72-, while two other moles react with the acid in ratio 1:1. On the base of e.m.f. drop magnitude at the equivalence point, the bases have been arranged in sequence OH->CO32->SO42- of basicity decrease. The equilibrium constants were not estimated.100,101 There is no correlation17,19 between substances entered into melt and those really existing in it. E.g., potassium nitrate and nitrite17,18 are decomposed to K2O at temperatures considerably lower than the temperature of the experiment (700oC). Ditto17,18 referred to K2S2O7 and Na2S2O8 as acidic phosphates.19 In the latter work it has been found that pyrophosphate acidity is larger than that for metaphosphate. However, titration of phosphorus (V) oxo-compounds in melts runs according to the scheme [10.4.39]. The acidity decreases,
632
Victor Cherginets
and, hence, pyrophosphate acidity should be lower, since in pair “metaphosphate-pyrophosphate” the latter is the conjugated base. The absence of correlation between pH in aqueous solutions and pO in melts for the same substance is often noted , this conclusion is made mainly from data for oxocompounds of PV. This fact is due to the following reason. The stages of phosphoric acid neutralisation in aqueous solutions are: −
−
−
→ H2 PO 4− OH → HPO 42 − OH → PO 43 − H3 PO 4 OH
[10.4.47]
In melts, a similar sequence can be presented as 1 1 / 2O 2 − / 2O 2 − / 2O 2 − P2 O 5 1 → PO 3− 1 → P2 O 74 − 1 → PO 43− 2 2
[10.4.48]
Hence, correlations should be found between the anion in aqueous solution and corresponding anhydro-acid in melt. NaPO3 and Na4P2O7 may be easily obtained by the calcination of the corresponding acidic salts, but reverse processes do not take place in aqueous media and pyro- and metaphosphate exist in water as salts of stable acids. For arsenates which are more prone to hydrolysis similar correlation should take place. The absence of the correlation is caused mainly by kinetic limitations. 10.4.5 REACTIONS OF MELTS WITH GASEOUS ACIDS AND BASES The gases present in the atmosphere over molten salts can react both with their principal components and with impurities. In reactions of the first kind, which include, in particular, hydrolysis, the gas (water) behaves as a Lux base since its action results in the increase of O2- concentration. In reactions of the second type, which are usually employed for the purification of melts, the gas has acidic properties. 10.4.5.1 High-temperature hydrolysis of molten halides These reactions were investigated mainly for alkaline halides and their mixtures. Hanf and Sole102 studied the hydrolysis of solid and molten NaCl in the temperature range 600-950oC by so-called “dynamic method” which consisted of passing inert gas (N2) containing water vapor through a layer of solid or fused NaCl. The developed routine determined the equilibrium constant of the following reaction: NaCl s,1 + H2 O gas = NaOHs,1 + HCl gas
[10.4.49]
The values of the equilibrium constants (log K) of [10.4.49] for solid and liquid sodium chloride were in the ranges -8 to -6 (650-800oC) and -6 to -5 (800-900oC). The hydrolysis of NaCl occurs to a slight extent, although the rate of HCl evolution exceeded the expected value because of dissolution of NaOH formed in NaCl. An evident disadvantage of the above method102 is assumption that the partial pressure of HCl in the gas phase is equal to the equilibrium pressure. The same method was used to study the high-temperature hydrolysis and oxidation of sodium iodide.103 According to the estimates, the heats of the reactions: NaI + H2 O = NaOH + HI
[10.4.50]
10.4 Acid-base equilibria in ionic solvents
NaI + 1 / 2 O 2 = 1 / 2 Na 2 O 2 + 1 / 2 I 2
633
[10.4.51]
are 32.8±2 and 13.9±1 kJ/mole for solid NaI and 30.6±2 and 9.3±2 kJ/mole for liquid one. These values suggest that NaI has a greater tendency to oxidation than to hydrolysis. This conclusions cannot be adopted without stipulation, because it is based on values ∆H and not ∆G, although only the latter values determine the direction of a reversible reactions at constant p and T. A simple but unquestionable method has been used to investigate high-temperature hydrolysis of molten KCl-NaCl.104 A mixture of HCl and H2O obtained by passing an inert gas through aqueous solutions at a definite concentration was passed into the melt. Measurements of the equilibrium O2- concentration by potentiometric method allowed to calculate the equilibrium constants of [10.4.13] as pK=55.3×103T-1-40.2. At 1000, pK=15.1, it means that equilibrium [10.4.13] in molten KCl-NaCl is displaced to the left. A similar method has been used to study the hydrolysis of the eutectic KCl-LiCl at 500oC (pK=9.77±0.4).105 The hydrolysis of the chloride melt is thermodynamically unfavorable and is completely suppressed in the presence of bases (even at a concentration of ~10-3 mole/kg). The potentiometric cells:
(H2 O + H2 )Au|N MeOH + (1 − N)MeX ||MeX |C(X 2 )
[10.4.52]
were used to investigate the hydrolysis of individual alkaline halide melts (with the exception of the lithium salts).106 The e.m.f. of cell [10.4.52] is related to the concentration of the reactants by the expression: E =E 0 +
1/ 2 1/ 2 RT PH 2 PX 2 ln a MeOH F PH 2O
[10.4.53]
where PH1/22 and PX1/22 are partial pressures of hydrogen and halogen over the melt, respectively. The logarithms of the equilibrium constants 0 PH1/22 PX1/22 ∆G HX F ln K = + ln E− a MeOH RT RT PH 2O
[10.4.54]
are negative for all the melts studied, which indicates that the hydrolysis of alkali metal halide melts is thermodynamically unfavorable. The trend towards hydrolysis of melts of individual alkaline halides diminishes with the increase in the radii of both cation and anion. According to the results of Smirnov et al.106 the solutions of alkali metal hydroxides in the corresponding individual halides are close to ideality. 10.4.5.2 The processes of removal of oxide admixtures from melts The studies of the elimination of oxygen containing impurities from ionic melts are of an applied character and are performed in melts having industrial applications. Various
634
Victor Cherginets
halogenating agents, usually hydrogen halides or the halogens themselves, are used to eliminate the impurities:107-111 2HX ↑ +O 2 − = H2 O ↑ +2X −
[10.4.55]
X 2 ↑ + O 2 − = 1 / 2O 2 ↑ + 2X −
[10.4.56]
However, the water formed in reaction [10.4.55] exhibits the properties of Lux acid owing to process opposite to [10.4.45] and this retards the purification process. The cessation of the HX effect may lead to partial hydrolysis, because reactions of type of [10.4.55] are reversible. The shift of [10.4.55] to the left is also promoted by the thermal instability of HBr and HI. The effectiveness of the purification of melts based on lithium salts by hydrogen halides is much lower. Water dissolves in the KCl-LiCl melt in appreciable amounts and is firmly retained at temperatures up to 400oC.30 When dry HCl is passed for 1 h, removal of H2O is incomplete. The effectiveness of the purification by the halogens depends on their oxidation-reduction potentials and decreases from chlorine to iodine. In addition, the latter can disproportionate: 3I 2 ↑ +O 2 − = 5I − + IO 3−
[10.4.57]
However, the addition of NH4X is recommended only for the drying and heat treatment of the initial components and its use for industrial purification of fused salts is undesirable because it is much easier to ensure the continuous supply of the corresponding hydrogen halide in the melt being purified. An additional lack of this purifying agent results in evolution of gaseous ammonia as one of the finishing products. It is believed that the most convenient method of the elimination of oxygen-containing impurities from alkali metal halide melts is carbohalogenation.112-118 The reduction of O2- concentration under action of CCl4, Cl2, C (C is acetylene carbon black)+Cl2 and COCl2 may be explained by the occurrence of the following reactions: CCl 4 ↑ +2O 2 − = 4Cl − + CO 2 ↑
[10.4.58]
C + Cl 2 ↑ +O 2 − = 2Cl − + CO ↑
[10.4.59]
COCl 2 ↑ +O 2 − 2Cl − + CO 2 ↑
[10.4.60]
Thermodynamic analysis of processes [10.4.58]-[10.4.60] showed that the effectiveness of the purification is in all cases approximately the same,113 but reaction [10.4.58] has the advantage that highly toxic reagents (Cl2, COCl2) are not used in it. For this reason, CCl4 has been used in virtually all subsequent studies involving the carbochlorination of chloride melts.114-118 It has been suggested that CHBr3, CBr4, C2H5Br and CHI3 may be used for the removal of oxygen-containing impurities from bromide and iodide melts.119 A significant advantage of carbohalogenation is that it leads to an appreciable decrease in the concentration of transition metal cations in the melt being purified. However, in the purification of melts by halo-
10.4 Acid-base equilibria in ionic solvents
635
gen-substituted compounds with a relatively low halogen content, carbon (carbon black) accumulates in the melt and is displaced by the crystallization front when single crystals are grown.112 The presence of a suspension of carbon in alkali metal halide melts is not always desirable and this method cannot therefore be recognized as universal. The carbohalogenation of melts results in formation of CO2 as one of products of [10.4.58]-[10.4.60] processes, carbon dioxide reacts with remaining oxide ions and induces the accumulation of fairly stable carbonate-ions. The latter is the main form in which oxygen-containing impurities exist in melts purified in this way. The purification threshold of carbohalogenation is estimated as 1-2×10-4 mole/kg of O2-.24 The use of halo-derivatives of silicon for the purification of iodide melts120 is based on the reaction: SiX 4 ↑ +2O 2 − = 4X − + SiO 2 ↓
[10.4.61]
the product of [10.4.61] is silicon dioxide, which is separated from the melt as a consequence of the difference between the densities. Thermodynamic analysis showed121 that the effectiveness of the purification of iodide melts diminishes in the sequence SiI4>HI>I2. The use of silicon halides for the purification of the melts used in the growing single crystals does not lead to the appearance of additional impurities, because the processes are carried out in quartz (i.e., SiO2) containers. Although the methods of purification described are fairly effective in most cases, some kind of oxygen-containing and cationic impurities, characteristic of each method, always remains in the melt. The method of purification must therefore be selected depending on the aims of the subsequent application of the melt. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
G.N. Lewis, Valence and the structure of atoms and molecules, S.n., New York, 1923. H. Lux, Z. Elektrochem., 45, 303(1939). H. Flood and T. Forland, Acta Chem.Scand., 1, 592 (1947). H. Flood and T. Forland, Acta Chem.Scand., 1, 781 (1947). H. Flood, T. Forland and B. Roald, Acta Chem.Scand., 1, 790 (1947). N.J. Bjerrum, C.R. Roston and G.P. Smith, Inorg.Chem., 6, 1162 (1967). N.J. Bjerrum and G.P. Smith, Inorg.Chem., 7, 2528 (1968). B. Tremillon and G. Letisse, J. Electroanal. Chem., 17, 371 (1968). G. Torsi and G. Mamantov, J. Electrocanal. Chem., 30, 193 (1971). G. Torsi and G. Mamantov, Inorg. Chem., 11, 1439 (1972). R. G. Pearson, J. Amer. Chem. Soc., 85, 3533 (1963). R. G. Pearson, J. Chem. Educ., 45, 581 (1968). R. G. Pearson, J. Chem. Educ., 45, 643 (1968). I. G. Dioum, J. Vedel, and B. Tremillon, J. Electroanal. Chem., 137, 219 (1982). J. F. Wagner, Electrochemical study in molten sodium fluoroborate at 420oC, Rep. CEA-N-2350, 15(2), Abstr. No15:005221 (1983). V. L. Cherginets, Rus. Chem. Rev., 66, 597 (1997). Yu. F. Rybkin and A. S. Seredenko, Ukr. Khim. Zhurn., 36, 133 (1970). Yu. F. Rybkin and A. S. Seredenko, Ukr. Khim. Zhurn., 40, 137 (1974). Yu. F. Rybkin and V. V. Banik, In Single Crystals and Engineering , No. 2(9) issue, Inst. Single Crystals, Kharkov, 152 (1973). N. N. Ovsyannikova and Yu. F. Rybkin, Ukr. Khim. Zhurn., 42, 151 (1976). J. E. Huheey, Inorganic chemistry. Principles of structure and reactivity, Harper and Row Publishers, New York, 1983. M. Dratovsky and D. Havlichek, Electrochim. Acta, 28, 1761 (1983).
636
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
Victor Cherginets A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 9, 349 (1965). V. L. Cherginets and T. P. Rebrova, Electrochim. Acta, 45, 469 (1999). R. Combes, M. N. Levelut and B. Tremillon, Electrochim. Acta, 23, 1291 (1978). R. Combes, B. Tremillon, F. De Andrade, M. Lopes and H. Ferreira, Anal. Lett., 15A, 1585 (1982). V. L. Cherginets, Electrochim. Acta, 42, 3619 (1997). Y. Castrillejo, A. M. Martinez, G. M. Haarberg, B. Borresen, K. S. Osen and R. Tunold, Electrochim. Acta, 42, 1489 (1997). A. I. Novozhilov and E. I. Ptchelina, Zhurn. neorg. Khim., 22, 2057 (1977). W. J. Burkhardt and J. D. Corbett, J. Amer. Chem. Soc., 79, 6361 (1957). B. J. Brough, D. H. Kerridge and M. Mosley, J. Chem. Soc. (A), N11, 1556 (1966). J. A. Duffy, M. D. Ingram and I. D. Sommerville, J. Chem. Soc. Faraday Trans., 74, 1410 (1978). T. Maekawa and T. Yokokawa, Nippon Kagaku, N6, 900 (1982). A. Kato R. Nishibashi, M. Hagano and I. Mochida, J. Amer. Ceram. Soc., 55, 183 (1972). N. Iwamoto, Youen, N21(3), 287 (1978). F. R. Duke and S. Yamamoto, J. Amer. Chem. Soc., 80, 5061 (1958). F. R. Duke and S. Yamamoto, J. Amer. Chem. Soc., 81, 6378 (1959). J. M. Shlegel, J. Phys. Chem., 69, 3638 (1965). J. M. Shlegel, J. Phys. Chem., 71, 1520 (1967). F. R. Duke, J. M. Shlegel, J. Phys. Chem., 67, 2487 (1963). I. Slama, Coll. Czechoslov. Chem. Commun., 28, 985 (1963). I. Slama, Coll. Czechoslov. Chem. Commun., 28, 1069 (1963). A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 8, 312 (1964). A. M. Shams El Din and A. A. A. Gerges, J. Inorg. Nucl. Chem., 26, 1537 (1963). A. M. Shams El Din and A. A. A. Gerges, Electrochim. Acta, 9, 123 (1964). A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 7, 464 (1964). A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 17, 238 (1968). A. M. Shams El Din, A. A. El Hosary and A. A. A. Gerges, J. Electroanal. Chem., 6, 131 (1963). A. M. Shams El Din, Electrochim. Acta, 7, 285 (1962). A. M. Shams El Din and A. A. A. Gerges, J. Electroanal. Chem., 4, 309 (1962). A. M. Shams El Din and, A. A. El Hosary, J. Electroanal. Chem., 16, 551 (1968). A. M. Shams El Din, A. A. El Hosary and H. D. Taki El Din, Electrochim. Acta, 13, 407 (1968). A. A. El Hosary, M. E. Ibrahim and A. M. Shams El Din, Electrochim. Acta, 24, 645 (1979). A. M. Shams El Din and A. A. El Hosary, Electrochim. Acta, 13, 135 (1968). A. M. Shams El Din and A. A. El Hosary, Electrochim. Acta, 12, 1665 (1967). J. D. Burke and D. H. Kerridge, Electrochim. Acta, 19, 251 (1974). A. M. Shams El Din and A. A. A. Gerges, In Electrochemistry, Pergamon Press, 1964 pp. 562-577. A. M. Shams El Din and A. A. A. Gerges, Electrochim. Acta., 9, 613 (1964). N. Coumert and M. Porthault, J. -C. Merlin, Bull. Soc. Chim. France, 33, 910 (1965). N. Coumert and M. Porthault, J. -C. Merlin, Bull. Soc. Chim. France, 35, 332 (1967). M. Hassanein and N. S. Youssep, Indian J. Chem., A21, 72 (1982). C. Liteanu, E. Cordos and L. Margineanu, Rev. Roumaine Chim., 15, 583 (1970). J. M. Schlegel, J. Chem. Educ., 43, 362 (1966). Y. Hoshino, T. Utsunomiya and O. Abe, Bull. Chem. Soc. Jpn., 54, 135 (1981). A. Baraka, A. Abdel-Razik and A. J. Abdel-Rohman, Surface Technol., 25, 31 (1985). A. Baraka, A. J. Abdel-Rohman and E. A. El-Taher, Mater. Chem. Phys., 9, 447 (1983). A. Baraka, A. J. Abdel-Rohman and E. A. El-Taher, Mater. Chem. Phys., 9, 583 (1983). A. I. Efimov, Properties of inorganic compounds. Handbook, Khimiya, Leningrad, 1983. I. T. Goronovskiy, Yu. P. Nazarenko and E. F. Nekryach, Short handbook on chemistry, Naukova Dumka, Kiev, 1987. G. J. Janz, Molten salts handbook, Academic Press, New York, 1967. Y. Kaneko and H. Kojima, Int. Symp. Molten Salts Chem. Technol., No. 1, 441 (1983) cited Chem. Abstr. 102, Abstr. No. 122309c (1988. ) Y. Kaneko and H. Kojima, Youen, 28, 109 (1985). A. Rahmel, J. Electroanal. Chem., 61, 333 (1975). H. Flood and W. Knapp, J. Amer. Ceram. Soc., 46, 61 (1963). T. Forland and M. Tashiro, Glass Ind., 37, 381 (1956). T. Forland, Glastekn. Tidskr., 17, 35 (1962).
10.4 Acid-base equilibria in ionic solvents
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120
637
V. I. Minenko,C. M. Petrov and N. S. Ivanova, Izv. vysshikh utchebnykh zavedeniy. Chernaya metallurgiya, N7, 10 (1960). US Patent 4,313,799; Chem. Abstr., 96, 115179g (1982). H. Itoh and T. Yokokawa, Trans. Jpn. Inst. Met., 25, 879 (1984). T. T. Bobrova, V. V. Moiseev, T. V. Permyakova and G. E. Sheshukova, Izv. AN SSSR. Neorg. mater, 9, 1416 (1973). R. Didtschenko and E. G. Rochow, J. Amer. Chem. Soc., 76, 3291 (1954). V. I. Shapoval and O. G. Tsiklauri, Conf. "Phys. chemistry and electrochemistry of molten salts and solid electrolytes", Sverdlovsk, June 5-7, 1973, Part II, Sverdlovsk, 1973 pp. 32-33. V. I. Shapoval, A. S. Avaliani and O. G. Tsiklauri, Soobsch. AN Gruz. SSR, 72, 585 (1973). Yu. K. Delimarsky, V. I. Shapoval,O. G. Tsiklauri and V. A. Vasilenko, Ukr. Khim. Zhurn., 40, 8 (1974). V. I. Shapoval, A. S. Avaliani and N. A. Gasviani, Soobsch. AN Gruz. SSR, 72, 105 (1973). V. I. Shapoval, V. F. Grischenko and L. I. Zarubitskaya, Ukr. Khim. Zhurn., 39, 867 (1973). V. I. Shapoval, V. F. Grischenko and L. I. Zarubitskaya, Ukr. Khim. Zhurn., 38, 1088 (1972). V. I. Shapoval, Yu. K. Delimarsky and V. F. Grischenko, in Ionic melts, Naukova Dumka, Kiev. 1974, pp. 222-241. O. G. Tsiklauri and N. A. Gasviani, In Materials of Conference of Young Scientists. Inst. of chemistry and electrochemistry of Academy of Sciences of Georgian SSR, Tbilisi, 1976, p. 63. V. I. Shapoval,O. G. Tsiklauri and N. A. Gasviani, Soobsch. AN Gruz. SSR, 88, 609 (1977). Kh. B. Kushkhov and V. I. Shapoval, X All-Union Conference on Physical Chemistry and Electrochemistry of Ionic Melts and Solid Electrolytes, Ekaterinbourg, October 27-29,1992, Ekaterinbourg, 1992, p. 3. R. Combes and B. Tremillon, J. Electroanal. Chem., 83, 297 (1977). M. Tazika, S. Mizoe,M. Nagano and A. Kato, Denki kagaku., 46, 37 (1978). R. Combes, F. De Andrade and L. Carvalho, C. r. Acad. Sci., C285, 137 (1977). V. L. Cherginets and V. V. Banik, Rasplavy, N6, 92 (1990). V. L. Cherginets and V. V. Banik, Rasplavy, N2, 118 (1991). T. P. Boyarchuk, E. G. Khailova and V. L. Cherginets, Ukr. Khim. Zhurn., 58, 758 (1992). V. L. Cherginets and V. V. Banik, Zhurn. Fiz. Khim., 68, 145 (1994). V. L. Cherginets, Potentiometric studies of acidic properties of niobium (V) and germanium (IV) oxides in chloride melts, Kharkov,1991;Dep in ONIITEKhim (Cherkassy),N 241 (1991). Yu. F. Rybkin, V. V. Banik, In Methods of obtaining and investigations of single crystals and scintillators, Inst. Single Crystals,Kharkov,1980, p. 121-125 (1980). Yu. F. Rybkin, V. V. Banik, In Single crystals and engineering, Inst. Single Crystals,Kharkov, 1974, p. 111-114. N. V. Hanf and M. J. Sole, Trans. Faraday Soc., 66, 3065 (1970). Y. F. Rybkin and Y. A. Nesterenko, Zhurn. Fiz. Khim., 50, 781 (1976). R. Combes, J. Vedel and B. Tremillon, Electrochim. Acta, 20, 191 (1975). V. L. Cherginets and V. V. Banik, Rasplavy, (4), 98 (1991). M. V. Smirnov, I. V. Korzun and V. A. Oleynikova, Electrochim. Acta, 33, 781 (1988). C. Butler, J. R. Russel, R. B. Quincy, J. Chem. Phys., 45, 968 (1966). F. Rosenberger, in Ultrapurity methods and Techniques, Dekker Inc, New York, 1972, pp. 3-70. D. Ecklin, Helv. Chim. Acta, 50, 1107 (1967). J. M. Peech, D. A. Bower, R. O. Rohl, J. Appl. Phys., 38, 2166 (1967). U. Gross, Mat. Res. Bull., 5, 117 (1970). M. Lebl and J. Trnka, Z. Phys.,186, 128 (1965). O. V. Demirskaya and Y. A. Nesterenko, In Physics and Chemistry of Crystals, Inst. Single Crystals, Kharkov, 1977, pp. 155-159. Y. F. Rybkin and O. V. Demirskaya, In Single Crystals and Engineering, Inst. Single Crystals, Kharkov, 1974, No. 1(10), pp. 115-118. R. C. Pastor and A. C. Pastor, Mat. Res. Bull., 10, 117 (1975). R. C. Pastor and A. C. Pastor, Mat. Res. Bull., 11, 1043 (1976). A. I. Agulyanskii and P. T. Stangrit, Zhurn. Prikl. Khim., 10, 1201 (1977). V. V. Banik and V. L. Cherginets, VII All-Union Conf. Chem. and Technology of Rare Alkali Elements, Apatity, June 5-10, 1988, Cola Division of Acad. Sci. of USSR, Apatity,1988, p. 116. V. V. Banik, N. I. Davidenko and Y. A. Nesterenko, In Physics and Chemistry of Solids, Inst. Single Crystals, Kharkov, 1983, No. 10, pp. 139-141. J. Ekstein, U. Gross and G. Rubinova, Krist. Tech., 3, 583 (1968).
638
Victor Cherginets
121 N. N. Smirnov and V. R. Lyubinskii, In Single Crystals and Engineering, Inst. Single Crystals, Kharkov, 1971, No. 5, pp. 95-101. 122 O. B. Babushkina and S. V. Volkov, XIV Ukraine Conf. Inorg. Chem., Kiev, September 10-12, 1996, Kiev National University, Kiev, 1996, p. 33.
11
Electronic and Electrical Effects of Solvents 11.1 THEORETICAL TREATMENT OF SOLVENT EFFECTS ON ELECTRONIC AND VIBRATIONAL SPECTRA OF COMPOUNDS IN CONDENSED MEDIA Mati Karelson Department of Chemistry, University of Tartu, Tartu, Estonia
11.1.1 INTRODUCTION The electronic-vibrational spectra of molecules can be substantially influenced by the surrounding condensed medium. The resultant effects arise from a variety of intermolecular interactions between the chromophoric solute and the solvent molecules in such media. Experimentally, these effects can be observed as • the shifts of the spectral maxima (solvatochromic shifts) • the change in the intensity of the spectral line or band • the change of the shape and width of the spectral band Each of those, so-called solvent effects can be described theoretically using different model approaches. The solvatochromic spectral shifts are expected to arise from the difference in the solvation of the ground and the excited states of the molecule. As a result of the spectroscopic excitation, the charge distribution of the molecule changes and thus the interaction will be different in the ground state and in the excited state of the molecule. The direction and size of the respective spectral shift depends directly on the difference in the solvation energy of the molecule in those two states. The larger solvation energy of the ground state (S0), as compared to that of the excited state (S1), results in the negative solvatochromic shift (blue shift) of the spectral maximum (cf. Figure 11.1.1a). Alternatively, the stronger solvation of the excited state, as compared to the solvation of the ground state, leads to the decrease of the excitation energy and is reflected by the positive solvatochromic shift (red shift) in the spectrum of the compound (Figure 11.1.1b). In each case, the direction and the size of the shift depends on the nature and electronic structure of the ground and excited state. For example, in the case of the n → π * transition in acetone (Scheme 11.1.1) an electron from the n-orbital (lone pair) is transferred to the Scheme 11.1.1. antibonding π* orbital localized on the C=O double bond. In
640
Mati Karelson
Figure 11.1.1. The origin of two types of solvatochromic shifts in the spectra of chromophoric compounds.
the ground state, the acetone molecule has a significant dipole moment (2.7 D) arising from the polarity of the C=O bond. Because of the difference in the electronegativity of bonded atoms, the electron distribution is shifted towards the oxygen that could be characterized by a negative partial charge on this atom. However, in the excited state the electron cloud is shifted from the oxygen atom to the bond and, correspondingly, Figure 11.1.2. The schematic representation of the the dipole moment of the molecule is subchange of the solvation energy of the ground (S0) and ex- stantially reduced. In result, the interaction cited state (S1) of acetone moving from a non-polar sol- of the dipole of the solute (acetone) with the vent (1) to a polar non-hydrogen bonded solvent (2) to a surrounding medium in more polar solvents polar hydrogen-bonded solvent (3). is larger in the ground state as compared to the excited state (Figure 11.1.2). Thus, the energy of the ground state is lowered more by the electrostatic solvation than the energy of the excited state. Consequently, the excitation energy increases and the respective spectral maximum is shifted towards the blue end of spectrum (negative solvatochromic shift). In the hydrogen-bonding solvents, the ground state of acetone is ad-
11.1 Theoretical treatment of solvent effects
641
ditionally stabilized by the hydrogen bonding of the oxygen lone pair by the solvent that leads to further increase of the excitation energy (Figure 11.1.2) and the respective blue shift of the spectral maximum. Notably, the formation of this hydrogen bonding is impossible in the S1(nπ*) excited state of the acetone because of the electron transfer from the oxygen lone pair to the antibonding π* orbital. In many cases, the dipole moment increases in the excited state. For instance, in the nitrobenzene (Scheme 11.1.2) the π → π* transition leads to the Scheme 11.1.2. substantial redistribution of the electronic charge reflected by the shift of negative charge on the nitro-group. The real excited state is given by a combination of the La and Lb states. Nevertheless, the dipole moment of nitrobenzene is substantially increased in the first excited state (S1). Because of the substantial charge redistribution in such excited states, they are often called the charge-transfer (CT) states. The much larger dipole of the nitrobenzene in the S1 (ππ∗ ) state is additionally stabilized by a more polar solvents that leads to the decrease in the excitation energy (Figure 3) and to the corresponding solvatochromic red shift of the spectral maximum. In Table 11.1.1, the solvatochromic Figure 11.1.3. The relative energies of the ground (S0) and the first excited state (S1(ππ*)) of nitrobenzene in shifts characterizing various positively and negatively solvatochromic compounds are non-polar (1) and polar (2) solvents. listed. In most cases, the theoretical treatment of the solvatochromic shifts has been based on the calculation of the solvation energies of the chromophoric molecule in the ground and excited states, respectively. Table 11.1.1. The solvatochromic shifts for various positively and negatively solvatochromic compounds Compound
νmax (non-polar solvent), cm-1
νmax (polar solvent), cm-1
∆νmax, cm-1
Ref.
(Scheme 3)
30000 (hexane)
25760 (CF3CH2OH)
+4240
(a)
(Scheme 4)
30170 (hexane)
26140 (water)
+4030
(b)
(Scheme 5)
20640 (hexane)
16860 (water)
+3780
(c)
(Scheme 6)
22620 (hexane)
19920 (DMSO)
+2700
(d)
(Scheme 7)
20730 (hexane)
18410 (methanol)
+2320
(e)
(Scheme 8)
27400 (cyclohexane)
23230 (water)
+4170
(f)
(Scheme 9)
43370 (isooctane)
41220 (water)
+2150
(g)
(Scheme 10)
14600 (toluene)
24100 (water)
-9500
(h)
642
Mati Karelson
Compound
νmax (non-polar solvent), cm-1
νmax (polar solvent), cm-1
∆νmax, cm-1
Ref.
(Scheme 11)
15480 (tetrahydrofuran)
24450 (water)
-8970
(i)
(Scheme 12)
20410 (chloroform)
22080 (water)
-1670
(j)
(Scheme 13)
20160 (chloroform)
22370 (water)
-2210
(j)
(Scheme 14)
16080 (chloroform)
22170 (water)
-6090
(k)
(Scheme 15)
35870 (CCl4)
37735 (CF3CH2OH)
-1865
(l)
(Scheme 16)
16390 (pyridine)
21280 (water)
-4890
(m)
(Scheme 17)
19560 (chloroform)
22060 (water)
-2500
(n)
S. Spange, D. Keutel, Liebig’s Ann. Chim., 423 (1992). S. Dähne, F. Shob, K.-D. Nolte, R. Radeglia, Ukr. Khim. Zh., 41, 1170 (1975). J.F. Deye, T.A. Berger, A.G. Anderson, Anal. Chem., 62, 615 (1990). D.-M. Shin, K.S. Schanze, D.G. Whitten, J. Am. Chem. Soc., 111, 8494 (1989). E. Buncel, S. Rajagopal, J. Org. Chem., 54, 798 (1989). M.J. Kamlet, E.G. Kayser, J.W. Eastes, W.H. Gilligan, J. Am. Chem. Soc., 95, 5210 (1973). E.M. Kosower, J. Am. Chem. Soc., 80, 3261 (1958). L.S.G. Brooker, A.C. Craig, D.W. Heseltine, P.W. Jenkins, L.L. Lincoln, J. Am. Chem. Soc., 87, 2443 (1965). (i) M.A. Kessler, O.S. Wolfbeis, Chem. Phys. Liquids, 50, 51 (1989). (j) H. Ephardt, P. Fromherz, J. Phys. Chem., 97, 4540 (1993). (k) H.W. Gibson, F.C. Bailey, Tetrahedron, 30, 2043 (1974). (l) G.E. Bennett, K.P. Johnston, J. Phys. Chem., 98, 441 (1994). (m) N.S. Bayliss, E.G. McRae, J. Am. Chem. Soc., 74, 5803 (1952). (n) E.M. Kosower, B.G. Ramsay, J. Am. Chem. Soc., 81, 856 (1959). (a) (b) (c) (d) (e) (f) (g) (h)
Scheme 3
Scheme 4
Scheme 5
Scheme 6
Scheme 7
Scheme 8
11.1 Theoretical treatment of solvent effects
Scheme 9
643
Scheme 10
Scheme 11
Scheme 12
Scheme 13
Scheme 14
Scheme 15
Scheme 16
Scheme 17
The solvent-induced broadening of the spectral lines and bands arises primarily from the variation of the local environment of the chromophoric solute molecule in the condensed medium caused by the thermal motion of the surrounding solvent molecules. At any given instant of time, there is a distribution of differently solvated solute molecules, each of which has characteristic transition energy to the excited state. The respective distribution of the transition energies leads to the broadening of the spectral band. It has to be kept in mind, however, that the broadening of spectral lines and bands can be also originated from adjoin-
644
Mati Karelson
ing the rotational and vibrational energy levels in the polyatomic molecule or from the Doppler and natural broadening of spectral lines. Those are more significant in the case of atoms and small molecules. The theoretical assessment of the solvent-induced spectral broadening has thus to rely on a proper statistical treatment of the solvent distribution around the chromophoric solute molecule, both in the ground and in the excited state of the latter. The surrounding solvent can also influence the intensity of the spectral transition (absorption or emission). The intensity of the spectral transition is usually characterized by the oscillator strength f defined as follows 8πm ν 2 f = | M | 3he 2
[11.1.1]
where m and e are the electron mass and the electron charge, respectively, h is the Planck’s constant, M is the transition moment and ν is the mean absorption wavenumber. Following the last equation, the intensity of the spectrum is proportionally related to transition energy, provided that the transition moment M is independent of the surrounding medium (solvent). This may, however, be not the case. The definition of the transition moment1 M = ∑ ψ 0 q i r i ψ 1*
[11.1.2]
i
includes, apart from the charges (qi) and their position-vectors (ri) in the molecule, the wave function of the molecule in the ground state (ψ0) and in the excited state (ψ *1 ), respectively. Therefore, whenever the solvent affects the wavefunction of the molecule either in the ground state or in the excited state, the intensity of spectral transition is further influenced by the change of the respective transition moment. The analysis of the solvatochromic effects on molecular absorption and emission (fluorescence and phosphorescence) spectra is further complicated by the variation of time scales for the solvent relaxation after the spectral excitation of the solute molecule. The spectral transition is a very fast process that takes place within approximately 10-16 s. Thus, during this short period of time the atomic nuclei do not practically move. The excited state reached by the respective vertical transition is often called the Franck-Condon state (Figure 11.1.4). The lifetime of the fluorescent excited state may be long enough (10-7 - 10-9 s) to allow in addition to the intramolecular nuclear relaxation (10-12 s), also the solvent orientational relaxation. The latter, which is characterized by the relaxation times ranging from 10-10 s up to infinity (in the case of Figure 11.1.4. The Franck-Condon transitions during the solids) may bring up the additional, solexcitation and the de-excitation of the molecule.
11.1 Theoretical treatment of solvent effects
645
vent-induced stabilization of the relaxed excited state as compared to the Franck-Condon state. Thus, as a rule, the solvatochromic shifts in the absorption and fluorescence spectra are not equal. The theoretical treatment of the time-dependent effects on molecular spectra in condensed phases is extremely complicated.2 In most cases, it is assumed that only the electronic polarization of the solvent contributes to the solvation energy of the Franck-Condon state (S1 in the case of absorption and S0 in the case of emission). In the case of long-living states, i.e., the ground state and the relaxed excited state, a full relaxation of the solvent is assumed in the field of the solute molecule. The solvation energy of different states at different degrees of relaxation will thus be rather different that may result in rather different dependence of the absorption and emission transition energies on the polarity of the solvent. Some examples of solvatofluorochromical compounds are given in Table 11.1.2. Table 11.1.2. The solvatofluorochromic shifts for various positively and negatively solvatochromic compounds Compound
νmax (non-polar solvent), cm-1
νmax (polar solvent), cm-1
∆νmax, cm-1
Ref.
(Scheme 18)
24400 (hexane)
16500 (water)
+7900
(a)
(Scheme 19)
21980 (cyclohexane)
18210 (water)
+3770
(b)
(Scheme 20)
~20000 (cyclohexane)
~17000 (CH3CN)
+3000
(c)
(Scheme 21)
23530 (pentane)
22730 (water)
+800
(d)
(Scheme 22)
24150 (hexane)
21850 (methanol)
+2300
(e)
(a) (b) (c) (d) (e)
I.A. Zhmyreva, V.V. Zelinskii, V.P. Kolobkov, N.D. Krasnitskaya, Dokl. Akad. Nauk SSSR, Ser. Khim., 129, 1089 (1959). M. Maroncelli, G.R. Fleming, J. Chem. Phys., 86, 6221 (1987). A. Safarzadeh-Amini, M. Thompson, U.J. Krall, J. Photochem. Photobiol., Part A., 49, 151 (1989). M.S.A. Abdel-Mottaleb, F.M. El-Zawawi, M.S. Antonious, M.M. Abo-Aly, M. El-Feky, J. Photochem. Photobiol., Part A., 46, 99 (1989). J. Catalán, C. Díaz, V. López, P. Pérez, J. Phys. Chem., 100, 18392 (1996).
Scheme 18
Scheme 20
Scheme 19
646
Mati Karelson
Scheme 21
Scheme 22
All theoretical treatments of solvatochromic shifts proceed from modelling the solvational interactions in the liquids and solutions. Theoretically, the interaction potential between a solute molecule and the surrounding solvent molecules Φ is given by the following integral ∞
Φ = C ∫ ϕ (R) g (R) R 2 dR
[11.1.3]
0
where ϕ (R) and g(R) are the pair interaction potential between the solute and the solvent molecule and the solvent radial distribution function around the solute molecule, respectively, and C is a constant depending on the density of the system. The integration in the last formula is carried out over the distance between the solute and the solvent molecule, R. The equation [11.1.3] is derived proceeding from the assumption that the intermolecular forces in the condensed medium are additive. This assumption may be, however, violated because of possible three- and many-body interactions between the molecules in the solution. For most of the real systems, the application of Eq. [11.1.3] directly is rather impractical because of the low precision of ϕ (R) and g(R), particularly in the case of many-atomic molecules. Moreover, this equation will be strictly valid only for the solute-solvent systems in thermodynamic equilibrium and thus not applicable for the Franck-Condon states. Thus, almost all theoretical calculations of solvatochromic effects proceed from different physical models describing the intermolecular interactions in liquids and solutions. Traditionally, the solvation energy of a molecule Esolv in a given solvent can be divided into the following terms3 E solv = E cav + E disp + E elst + E H − bond
[11.1.4]
each of which corresponds to a certain type of intermolecular interaction in the condensed media. Thus, Ecav denotes the energy of the cavity formation for the solute in the solvent, Edisp is the dispersion energy and Eelst the electrostatic energy of the interaction of the solute with the surrounding solvent molecules. The term EH-bond accounts for the energy of the hydrogen bond formation between the solute and solvent molecules. The value of each of the above terms will change as a result of the Franck-Condon excitation of the solute molecule. First, the size of the molecule increases, as a rule, during the excitation. However, as the excitation process is practically instantaneous, the position and orientation of the solvent molecules in the solvation sheath of the chromophoric solute will not change. This means that
11.1 Theoretical treatment of solvent effects
647
the average distance between the surface of the solute and the solvent molecules will decrease in the Franck-Condon excited state of the former that normally causes the enhanced solute-solvent repulsion in that state. At the same time, the dispersion energy that stabilizes the solute-solvent system will also increase in the absolute value, but to the opposite direction. In consequence, both effects may cancel each other and the net effect will be close to zero. For the polar solutes, both in the ground and in the excited state, the electrostatic solvation energy is therefore often considered as the most important term in Eq. [11.1.4]. During the excitation or de-excitation of the molecule, the molecular electronic wavefunction and the electron distribution may change significantly. In result, substantial differences are expected in the electrostatic and dispersion solvation energies of the ground and the excited state, respectively. In addition, the hydrogen bonding between the solute and solvent molecules may be affected by the excitation of the solute molecule that will be reflected as another contribution to the difference in the solvation energy of solute in the ground and in the excited state, respectively. In the following, we proceed with the systematic presentation of the theoretical methods developed for the description of the solvatochromic effects on molecular electronic and vibrational spectra in condensed disordered media (liquids, solutions, glasses etc.). 11.1.2 THEORETICAL TREATMENT OF SOLVENT CAVITY EFFECTS ON ELECTRONIC-VIBRATIONAL SPECTRA OF MOLECULES As described above, the change (increase) in the size of the molecule during the excitation will result in increased van-der-Waals repulsion between the electron clouds of the chromophoric solute and the solvent molecules. Alternatively, the size of the molecule is expected to shrink as a result of the de-excitation of the molecule back to the ground state. In such case, the repulsion between the solute and solvent molecules will be reduced correspondingly. The respective energetic effect may be modeled as the difference in the cavity formation energies for the solute molecule in two states. The dependence of the cavity formation energy on the cavity size has been derived using several different model concepts. The simplest approach is based on the concept of microscopic surface tension on the boundary between the solute cavity and the solvent. Within this approach, the free energy of cavity formation is assumed simply proportional to the surface of the solute cavity, SM: ∆G cav = σS M
[11.1.5]
where σ is the surface tension of the solvent. This formula has been applied for the evaluation of the free energy of transfer of electroneutral solutes between different solvents.4 It has been extended to account for the size of the solvent molecule as follows: ∆G cav = σS M − RT ln(1 − Vs ns )
[11.1.6]
where VS is the intrinsic volume of a solvent molecule and nS is the number density of the solvent. In order to account for the chemical individuality of constituent atoms, it has been suggested to use different surface tension values σ i for different atomic types in the solute molecule.5 Thus, ∆G cav = C + ∑ σ i Ai i
[11.1.7]
648
Mati Karelson
where Ai are the solvent-accessible surface areas of different atoms in the solute molecule and C is an empirically adjustable parameter. The quality of the description of experimental data has been, however, not significantly improved by the introduction of individual atomic surface tension characteristics. Another theoretical approach for the calculation of the free energy of cavity formation proceeds from the theory of microscopic curved surfaces. According to this theory,6 V ∆G cav = ksg S VM
σS
[11.1.8]
where S is the area of the cavity and k gs (VS / VM ) is a correction factor, characteristic of a given solvent and depending on the ratio of molecular volumes of the solvent and solute. This factor has been approximated by the following formula V ksg S VM
V = 1 + S VM
2/3
[k
g s
(1) − 1]
[11.1.9]
where k gs (1) is estimated from the solubility of a given solute in a given solvent. The main deficiency of this approach is connected with the introduction of additional empirical information, often not readily available. The free energy of cavity formation has been also estimated from the data on isothermal compressibility, βT, as follows7 ∆G cav =
Vcav +C βT
[11.1.10]
where Vcav is the volume of the cavity and C is a constant term. However, the microscopic isothermal compressibility of water, calculated from the slope of Eq. [11.1.10], was found to be about an order higher than the respective experimental value for water (βT(calc) = 23.5 vs. βT(exp) = 3.14). Therefore, the use of the macroscopic surface tension or compressibility of the solvent for the respective microscopic model quantities is questionable. An entropic approach to the calculation of the free energy of cavity formation proceeds from the scaled particle theory (SPT).8,9 The free energy of the formation of a spherical cavity in a solvent, ∆Gcav, can be calculated proceeding within the framework of SPT as follows 2 9 y aM 3y a M 3y ∆Gcav = RT 1− ln(1− y ) + + + 1− y a S 1− y 2 1− y a S
2
[11.1.11]
where y=
4πρas2 3
[11.1.12]
11.1 Theoretical treatment of solvent effects
649
is the reduced number density of the solvent. In the two last equations, aM and aS denote the intrinsic radii of the solute and solvent molecules, respectively, and ρ is the number density of the solvent. In the case of an ellipsoidal solute cavity, the SPT cavity formation energy has been given by the following equation10 2 αy a βy y a ∆Gcav = RT 1− ln(1− y ) + + + γ 1− y a S 1− y 1− y a S
2
[11.1.13]
where α , β and γ denote the geometrical coefficients and a is the characteristic length of the ellipsoid (the major semi-axis). The scaled particle theory has been extended to dilute solutions of arbitrary shaped solutes and has been successfully applied for the calculation of the solvation free energy of hydrocarbons in aqueous solutions.11 For most practical applications that involve the lowest excited states of the molecules, the increase in the cavity size during the excitation of the solute molecule would not be accompanied with a significant energetic effect. However, it may be important to account for the so-called Pauli repulsion between the solute electronic system and the surrounding medium. This interaction will force the solute electrons to stay inside the cavity and not to penetrate into the dielectric continuum (consisted of electrons, too) that surrounds it. The Pauli repulsion has been modeled by the respective model potentials, e.g., by expanding the potential in spherical Gaussian shells as follows:12
[
VPR = ∑ b i exp −β i (r − ro, i ) i
2
]
[11.1.14]
where bi are the weight factors, βi the exponents and r0,i the radii of spherical shell functions. In general, the electrons in the solvent cavity could be treated as confined many-electron systems.13 11.1.3 THEORETICAL TREATMENT OF SOLVENT ELECTROSTATIC POLARIZATION ON ELECTRONIC-VIBRATIONAL SPECTRA OF MOLECULES The origin of the solvatochromic shifts in the electronic spectra is related to the change in the electrostatic and dispersion forces between the solvent and the chromophoric solute molecule in the ground and in the excited state, respectively. The semiclassical approach to the treatment of the respective effects is based on the assumption that the solute and the solvent molecules are sufficiently separated to neglect the overlap between the electron distribution of these two molecular systems. The wave function for the whole system can then be approximated as the product of the wavefunctions of each individual system, i.e., the solute and individual solvent molecules: Ψ = ψ s0(1) ψ s0( 2 ) L ψ s0( n ) ψ (aν )
[11.1.15]
0 0 , etc. are the wavefunctions of the respective solvent molecules in where ψ 0s(1) , ψ s(2) ,L ψ s(n) the ground state and ψ (aν ) is the wavefunction of the solute molecule in the ν-th state. The antisymmetry of the total electronic wavefunction is ignored as the individual molecules are assumed separated enough not to allow the electron exchange. This approximation may not be valid in the case of strong semichemical interactions between the solute and the solvent
650
Mati Karelson
molecules such as the hydrogen bonding or the formation of charge-transfer complexes. In such cases, the system consisting of the central solute molecule and the adjacent solvent molecules has to be treated as a supermolecule. In the absence of strong semichemical interactions between the solute and solvent molecules, the interaction energy between them can be derived using the perturbation theory.14 In the first approximation, the interaction between the nonionic molecules can be reduced to the dipole-dipole interaction between the molecules. The following perturbation operator can describe this interaction $ µ$ µ$ (αv ) Θ ai s ( i ) $ ′ H =∑ Rai3 i
[11.1.16]
where µ$ (αν ) and µ$ s(i) are the dipole moment operators for the solute a in the ν-th state and for the i-th solvent molecule in the ground state, respectively, Rai is the distance between the charge centroids of the interacting molecules and →
→
$ = 1$ − 3 R ai R ai Θ ai
[11.1.17]
is the angular term describing the relative orientation of these two molecules in the space. The subsequent application of the perturbation theory to derive the energy of interaction between a pair of a solute and a set of N solvent molecules gives the following result15 N
µ (aν ) Θ ai µ s ( i )
i =1
Rai3
∆E a = ∑ −
−
(ν ) (ν ) 1 N µ a Θ ai α s Θ ai µ a − ∑ 2 i =1 Rai6
( ν) µ ( 0 p ) Θ µ ( λν)µ ( λν) Θaj µ (s0( ip) ) 1 N N µ s ( i ) Θai α a Θaj µ s ( j ) N − ∑ ∑ ∑ s ( i ) ai a 6 a ∑ ∑ 3 3 Rai Raj Rai 2 i =1 j =1 i =1 p ≠ 0 λ≠ ν
[11.1.18]
with the following notations: µ (aν ) = ψ (aν ) µ$ a ψ (aν )
[11.1.19]
is the dipole moment of the solute in the ν-th state, µ s ( i ) = ψ s0( i ) µ$ s ψ s0( i )
[11.1.20]
is the dipole moment of the i-th solvent molecule in the ground state, α
(ν ) a
= 2∑ λ≠ν
ψ (aν ) µ$ a ψ (aλ ) ψ a( λ ) µ$ a ψ a( ν ) Eν −Eλ
is the polarizability tensor of the solute molecule in the ν-th state,
[11.1.21]
11.1 Theoretical treatment of solvent effects
α s = 2∑
p≠0
ψ s0 µ$ s ψ s( p ) ψ s( p ) µ$ s ψ s0 E 0s − E ps
651
[11.1.22]
is the polarizability tensor of the solvent molecule in the ground state, respectively, and µ s( 0( ip) ) = ψ s0( i ) µ$ a ψ s( p( i))
[11.1.23]
µ (aλν ) = ψ (aλ ) µ$ a ψ (aν )
[11.1.24]
and
are the transition dipoles between the two states (0 → p) in the solvent and in the solute (λ → ν) molecules, respectively. In the last formulae, Eν and Eλ denote the energy of the solute molecule in the respective (ν-th and λ-th) states, and Eps and E0s - the energy of a solvent molecule in the p-th and in the ground state, respectively. The first term in equation [11.1.18] represents therefore the electrostatic interaction of the unperturbed charge distribution of the two molecules, given as the interaction between the respective permanent point dipoles. The second term in this equation corresponds to the interaction of the permanent dipole of the solute with the dipole induced in the solvent whereas the third term reflects the interaction of the permanent dipole of the solvent with the induced dipole of the solute. The last term represents the second-order interaction of both molecules in excited states and quantifies thus effectively the dispersion interaction in the solute-solvent system. The equation [11.1.18] refers, of course, to a single fixed configuration of the solute and the solvent molecules. In order to find the effective interaction energy in the liquid medium, an appropriate statistical averaging over all configurations has to be carried out. In most practical applications, this procedure is very complicated and thus the semiclassical continuum approaches are employed to describe the solvent. The description of the electrostatic interactions between the solute and the solvent has been based on the Onsager’s reaction field model. According to this model, the energy of electrostatic interaction between an arbitrary charge distribution inside the solute molecule and the surrounding polarizable dielectric medium is given by the following equation16 E el =
1 Es E o (ε − 1)dV 8π V∫
[11.1.25]
where Eo is the electrostatic field of the charges in the molecule in vacuo and Es is the modified field in the presence of dielectric medium. Notably, within the formalism of the last [11.1.25], the dielectric constant ε of the medium is still a function of the space coordinates, i.e., both the interior of the molecule and the surrounding medium are treated by the same equation. However, the integral in the last equation cannot be found analytically and even the numerical integration over the space presents a difficult mathematical task. Therefore, the electrostatic equation is usually simplified by the application of the Gauss divergence theorem. According to this theorem, the volume integral in [11.1.25] is transformed into a surface integral over some boundary
652
Mati Karelson
E el =
r ε −1 ΦS E o ndS ∫ 8π S
[11.1.26]
r where S is the boundary surface, n the outward normal unit vector on S and the reaction potential Φ s is defined as follows: Es = -grad ΦS. Depending on the shape of solute molecular cavity, different approaches have been applied for the calculation of the electrostatic solvation energies of compounds in liquids. Within the classical reaction field theory of Kirkwood and Onsager,17,18 the solute molecule is represented by a set of point charges fixed inside of sphere of a radius a 0 and the electrostatic equation [11.1.26] is solved by applying the appropriate boundary conditions inside and outside the sphere. It is also assumed that the dielectric constant inside the cavity (sphere) is equal to unity (vacuum) and outside the cavity has a constant value, corresponding to the macroscopic dielectric constant of the medium studied. In that case, the energy of the electrostatic interaction between the solute charge distribution and the surrounding dielectric medium is given by the following infinite expansion E el
l l ∞ (l + 1)(1 − ε) r r 1 i j = ∑ ei e j ∑ 2 l+1 Pl cosθ ij 2 i, j l = 0 ε(l + 1) + 1 a 0
(
)
[11.1.27]
where ei and ej are the charges inside the sphere at positions ri and rj, respectively, and θij is the angle at the center of the sphere between the vectors ri and rj. In the last equation, the summation proceeds over all charged particles (nuclei and electrons) of the solute and P1(cosθij) are the Legendre polynomials of l-th order. By expressing the Legendre polynomials as the products of the respective spherical harmonics of order m (-l ≤ m ≤ l), equation [11.1.27] can be rewritten as E el = −
1 ∞ l ∑ ∑ Rlm M lm 2 l = 0 m = −1
[11.1.28]
where Rlm = f l M lm
[11.1.29]
and fl =
(l + 1)(ε − 1) 1 (l + 1)ε + 1 a 02 l+1
[11.1.30]
In these equations, M ml and R ml represent the electrical momentum and the respective reaction field component. The first term (l = 0) in the expansion [11.1.27] gives the interaction of the excess (ionic) charge of the solute with the respective reaction field created in the dielectric medium (Born term) E Born =
1− ε Q 2 2ε a 0
[11.1.31]
11.1 Theoretical treatment of solvent effects
653
where Q is the numerical value of the ionic charge. The next term (l = 0) corresponds to the total dipole interaction with the corresponding reaction field (Onsager dipolar term) E Onsager
r (1 − ε) µ 2 = ( 2ε + 1) a 03
[11.1.32]
r where µ is the dipole moment of the solute. In most applications, only these two terms that are the largest by size are considered in the calculation of the electrostatic interaction energy. However, depending on system studied, the interaction of higher electrical moments with the corresponding reaction field may become also significant and the terms corresponding to higher moments of order 2l (quadruple, octuple, hexadecuple, etc.) should be taken into account. In many cases, the shape of the solute molecule may be very different from the sphere and therefore, it is necessary to develop the methods of calculation of the electrostatic solvation energy for more complex cavities. In the case of the ellipsoidal cavity with main semiaxes a, b, and c, the analytical formulas are still available for the calculation of the charge and dipolar terms of the electrostatic interaction with the reaction field. The charge term is simply ell E Born =
(1 − ε) Q 2 2ε abc
[11.1.33]
whereas in the respective dipolar term19 r ell E Onsager = Rµ
[11.1.34]
the reaction field R is presented using a special tensor as follows 3Aa (1− Aa )( 1− ε)µ a abc[ε + ( 1− ε)Aa ] R = 0 0
0 3Ab (1− Ab )( 1− ε)µ b abc[ε + ( 1− ε)Ab ] 0
r µ 0 3Ac (1− Ac )( 1− ε)µ c abc[ε + ( 1− ε)Ac ] 0
[11.1.35]
where Aa, Ab and Ac are the standard ellipsoidal shape factor integrals, and µa, µb and µc are the dipole moment components along the main semiaxes of the ellipsoid. Several methods have been developed to define the semiaxes of the ellipsoidal cavity. For instance, these have been taken collinear with the axes of the solute dipole polarizability tensor, and their lengths proportional to the respective eigenvalues.20 Another definition proceeds from the inertia tensor of the van-der-Waals solid, i.e., a solid or uniform density composed of interlocking van-der-Waals spheres.21 Also, the ellipsoidal surface has been defined in terms of the best fitting of a given molecular electrostatic isopotential surface.22 The above-discussed theoretical formulation of the electrostatic solute-solvent interaction is applicable for the fixed charge distribution inside the solute molecule. However, the solvent reaction field may cause a redistribution of the charge inside the solute. The
654
Mati Karelson
magnitude of this redistribution depends on both the dielectric constant of the solvent and the polarizability of the solute molecule. Within the approximations of the spherical solute cavity and the point dipole interactions between the solute and solvent, the dynamically changed Onsager reaction field can be expressed by the following formula14 Rl =
µ (aν ) + α (aν ) Rl 2(ε − 1) a 03 (2ε + 1)
[11.1.36]
Notably, the use of the macroscopic dielectric constant ε = ε0 in the last formula is justified only when the lifetime of the solute molecule in a given (ν-th) state is much longer than the rotational-vibrational relaxation time of the solvent at given temperature. This is not a valid assumption in the case of the Franck-Condon states, which have the lifetime much shorter than the rotational-vibrational relaxation time of the solvent. Therefore, the solvent is only partially relaxed for these states and the corresponding reaction field is characterized by the dielectric constant at infinite frequency of external electric field, ε ∞ . By inserting the expression for the reaction field [11.1.36] into the equation [11.1.18] and assuming that the static polarizability of the solute molecule is approximately equal to the one third of the cube of Onsager’s cavity radius α a( ν ) ≈
a 03 3
[11.1.37]
the following semiclassical equation can be obtained for the solvation energy of the ν-th (Franck-Condon) state of the solute molecule14 ε − 1 ε∞ − 1 2µ 0aµ (aν) 2µ 0aµ (aν) + Es = − 0 − 3 a30 ε0 + 2 ε∞ + 2 a 0
( )
( ν) ε0 − 1 2 µ a − a30 ε0 + 2
(
)
2
ε∞ − 1 + ε∞ + 2
2
3 (2ε0 + 1) 2 ( ε0 − ε∞ ) 2 kT 2ε − 2 2 µ (aλν) 0 + + + ∑ (2ε0 − ε∞ )ε0 3a30 2ε0 + 1 λ≠ ν Eλ − Eν 2ε0 + 1 ε − 1 2 3 + ∞ ε∞ + 2 a 0
∑ (µ ) λ≠ ν
( λν ) a
2
E − Eλ 1− ν E −E ps 0s
[11.1.38]
The solvatochromic shift due to the difference in the electrostatic solvation energy of the ground state and the excited state of the solute, respectively, is thus given as follows: ∆E s =
2 ε 0 − 1 ε ∞ − 1 2ε 0 + 1 − a 03 ε 0 + 2 ε ∞ + 2 ε 0 + 2
3 + 2ε0 +
[(µ
0 a
)
2
ε − 1 − µ a0µ a(ν ) + ∞ ε∞ + 2
]
2 2 (2ε0 + 1) (ε0 − ε∞ ) kT (µ (a0 λ ) ) (µ a(νλ) )2 + − ∑ ∑ 1 (2ε0 − ε∞ )ε0 3a03 λ≠0 E 0 − E λ λ≠ν E ν − E λ
[(µ
0 a
) − (µ ) ] 2
(ν ) a
2
11.1 Theoretical treatment of solvent effects
2ε − 2 + 0 ε0 + 2
2
(µ a(ν ) )2 6 a 0
(µ ) ∑E − E (νλ ) a
λ≠ν
ν
2
− λ
655
(µ ) (0 ) a 6 0
a
ε − 1 2 E − E0 2 3 ∑(µ (a0 λ ) ) 1− λ + ∞ + ε 2 a E 0 ≠ λ 0 s − E ps ∞ 0
2
(µ ) ∑E − E (0 λ ) a
λ≠0
0
+ λ 2
E − Eν − ∑(µ (aνλ ) )2 1− λ λ≠ν E ps − E 0 s
[11.1.39]
The last expression represents the solvent effect on the transition energy of the 0-0 band of the solute molecule. McRae15,23 has given a different derivation of the electrostatic solvation energy based on semiclassical reaction field approach. The final result is however similar to the above equation for the solvatochromic shift in the electronic transition:15 ∆E s =
+
1 a03
2 ε 0 − 1 ε ∞ − 1 2ε 0 + 1 − a 03 ε 0 + 2 ε ∞ + 2 ε 0 + 2
[(µ
0 a
)
2
− µ a0µ (aν ) +
]
ε∞ − 1 2 2 1 1 2 2 (µ a(ν ) ) − (µ 0a ) − α a(ν ) E a(ν ) + α 0a E a0 + Da(ν ) − Da0 + ε 2 2 2 ∞
[
]
[11.1.40]
where E0a and E(aν ) are the solvent fields due to the permanent dipole moments of the solvent molecules applying to the ground state and to the excited state of the solute molecule, respectively. The terms D0a and D(aν ) denote the solute-solvent intermolecular dispersion energies in the corresponding states. Abe24 has developed an alternative semiclassical theory of the solvent effects on electronic spectra. This theory is based on the averaging of the intermolecular interaction energy over all solute-solvent configurations within the approximation of pair interactions. The theory involves the dipole moments and polarizabilities of the solute molecule and takes into account the temperature dependence arising from the Boltzmann factor. In all above-listed theoretical approaches, the response of the solute charge distribution to the solvent field is expressed by using the static polarizability of the solute molecule. However, it would be plausible to account for this response directly within the quantum mechanical theoretical framework. The quantum-chemical approaches to the calculation of the solvation effects on the ground and excited states of the molecules in the solution can be classified using two possible ways. First, it can be based on the traditional division of the quantum chemistry into the non-empirical (ab initio) and the semiempirical methods. Within both those classes of methods, the Hartree-Fock method based on the independent particle model and the methods accounting for the static and dynamic electron correlation are usually distinguished. The second way of classification of methods can be based on the differences of the models used for the description of solute-solvent interactions. In general, these interactions can be taken into account in the framework of continuum representation of the solvent or using the resolution of solute-solvent interactions at molecular level. In the following, we first proceed with the review of models used for the solute-solvent interactions, with the subsequent elaboration of the quantum-chemical methodology for the calculation of the solvent effects on spectra. The simplest continuum model is based on the classical Onsager reaction field theory assuming the spherical or ellipsoidal form of cavities for the solute molecules in dielectric
656
Mati Karelson
media. The respective interaction energy is accounted for as a perturbation V$ (a 0 , ε ) of the $ 0. Hamiltonian of the isolated solute molecule, H $ =H $ 0 + V$ (a , ε) H 0
[11.1.41]
Within the approximation of electrostatic interaction between the solute dipole and the respective reaction field, the perturbation term is simply V$ (a 0 , ε) = Γµ$ 2a
[11.1.42]
where Γ=
2(1 − ε)
[11.1.43]
(2ε + 1)a 03
In the case of ellipsoidal cavities, the last coefficient has to be substituted by the tensor given in equation [11.1.35]. A self-consistent reaction field method (SCRF) has been developed at the level of Hartree-Fock theory to solve the respective Schrödinger equation25 $ = EΨ HΨ
[11.1.44]
Proceeding from the classical expression for the electrostatic solvation energy of a solute molecule in a dielectric medium in the dipole-dipole interaction approximation, the total energy of the solute is presented as follows26 E = Eo −
(
r r 1 Γ ψ µ$ ψ ψ µ$ ψ + 2µ nuc ψ µ$ ψ + µ 2nuc 2 ∧
)
[11.1.45]
∧
where Eo = <ψ|H0|ψ>, H0 is the Hamiltonian for the reaction field unperturbed solute molecule and ψ is the molecular electronic wave function. From the last equation, one can construct the variational functional L = Eo −
(
)
r r 1 Γ ψ µ$ ψ ψ µ$ ψ + 2µ nuc ψ µ$ ψ + µ 2nuc − W ( ψ | ψ − 1) [11.1.46] 2
where W is the Lagrange multiplier ensuring the normalization of the variational wave function. The variation of the last equation with respect to the parameters of the wave function yields
(
r δL = δE o − Γ δψ µ$ ψ ψ µ$ ψ + µ nuc ψ µ$ ψ
) −Wδ( ψ | ψ )
$ 0 ψ − Γ δψ µ$ ψ µr − W δψ | ψ + c. c. = 0 = δψ H tot
[11.1.47]
r where µ$ tot = µ$ nuc + <ψ |µ el |ψ > is the total dipole moment of the solute molecule. The latter is calculated during the SCRF procedure simultaneously with the total energy of the system.
11.1 Theoretical treatment of solvent effects
657
From equation [11.1.47], the following Schrödinger equation is obtained for the electronic state |ψ > of the solute molecule $ 0 − Γ ψ µ$ ψ µ$ = W | ψ H tot el
[11.1.48]
W plays the role of the quantum mechanical motif that is directly obtained from the Schrödinger equation as follows $ 0 − Γ ψ µ$ ψ µ$ ψ = ψ H $ 0 ψ − Γ ψ µ$ ψ ψ µ$ ψ W = ψH tot el tot el
[11.1.49]
By adding the part for the interaction of the nuclear component of the solute dipole with the total reaction field and assuming $ 0 ψ ≈ ψ0 H $ 0 ψ0 ψH
[11.1.50]
where ψ 0 is the wavefunction of the solute molecule, unperturbed by the reaction field, one obtains that $0ψ −Γ E el = E rf − E 0 = ψ H
( ψµ$
el
r ψ + µ nuc
)
2
$ 0 ψ 0 ≈ − Γ µr − ψ0 H tot
2
[11.1.51]
The comparison of the last equation with the starting equation [11.1.44] reveals a difference by the factor of two in the final result. Of course, the approximation [11.1.50] brings up a certain error and it has been therefore proposed25,27 to correct the last formula by the addition of the “solvent cost”, i.e., the additional work required to reorganize the solvent due to the electrostatic field of the solute E el = E rf − E 0 +
1 r Γ µ tot 2
2
[11.1.52]
Alternatively, the electrostatic solvation energy can be derived proceeding from the following variational functional27 L = E0 −
(
)
r r2 1 − W ( ψ | ψ − 1) [11.1.53] Γ ψ µ$ ψ ψ µ$ ψ + 2µ nuc ψ µ$ ψ + µ nuc 4
This leads to a Schrödinger equation which eigenvalue is directly related to the total electrostatic (dipolar) solvation energy, Eel, $0 1 (e ) H − Γ ψ µ$ tot ψ µ$ el | ψ = E el | ψ 2 E el = E rf( e ) −
1 r r Γµ nuc µ tot − E 0 2
[11.1.54]
[11.1.55]
658
Mati Karelson
For the solution of equation [11.1.54], the molecular wavefunction can be presented as a proper spin-projected antisymmetrized product of molecular (or atomic) orbitals27 |ψ§ = Os A[φ1,L, φn ]
[11.1.56]
Recalling that the dipole moment operator is a one-electron operator (as are all electric moment operators), the following orbital equations are obtained f (k)φi (k ) = ε i φi (k )
[11.1.57]
f (k) = f 0 (k) − Γ ψ µ$ tot ψ µ$ el (k)
[11.1.58]
with
or f (k) = f 0 (k) −
1 Γ ψ µ$ tot ψ µ$ el (k) 2
[11.1.59]
where f0(k) is the usual Fock operator for the isolated molecule, ε i is the molecular orbital energy for |φi > and µ$ el (k) is the electronic part of the dipole moment operator. Both equations are solved iteratively, using the usual SCF procedure and the expectation value of the total dipole moment from the previous SCF cycle. A scheme for the treatment of the solvent effects on the electronic absorption spectra in solution had been proposed in the framework of the electrostatic SCRF model and quantum chemical configuration interaction (CI) method.27 Within this approach, the absorption of the light by chromophoric molecules was considered as an instantaneous process. Therefore, during the photon absorption no change in the solvent orientational polarization was expected. Only the electronic polarization of solvent would respond to the changed electron density of the solute molecule in its excited (Franck-Condon) state. Consequently, the solvent orientation for the excited state remains the same as it was for the ground state, the solvent electronic polarization, however, must reflect the excited state dipole and other electric moments of the molecule. Considering the SCRF Hamiltonian $ =H $ 0 + Γ φµ$ φ µ$ H tot
[11.1.60]
it is possible to write for the state |ψ I ¦ the following expression $ 0 + Γ ψ µ$ ψ µ$ ψ = ψ H $ 0 ψ − Γ ψ µ$ ψ ψ µ$ ψ HII = ψ I H I I I I I 0 0
[11.1.61]
that is the zeroth order estimate of the energy of the state |ψ I ¦. Then, for a single excitation, I → A, the excitation energy is given as follows
11.1 Theoretical treatment of solvent effects
659
2 2 ∆E ia = εa − εi − J ia + Kia = εa0 − εi0 − J ia + Kia − 0 0 − Γ ψ 0 µ$ ψ 0 ψ ai µ$ ψ ai − ψ 0 µ$ ψ 0
[
[11.1.62]
]
where Jia and Kia are the respective Coulomb’ and exchange matrix elements and Γ is the reaction field tensor at the dipole level. The terms ε 0i are the eigenvalues of the Fock operator for the k-th electron in the isolated solute molecule. The off-diagonal CI matrix elements are given by $ 0 + Γ ψ µ$ ψ µ$ ψ = ψ H $ 0 ψ − Γ ψ µ$ ψ ψ µ$ ψ HIJ = ψ I H J I J I J 0 0
[11.1.63]
Equations [11.1.60] - [11.1.63] demonstrate that some part of the solvent effect is already included in the ordinary CI treatment when proceeding from the SCRF Fock matrix. It $ 0 ¦ should represent the ground-state dipole moment has to be noticed that the terms ¢ψ 0 |µ|ψ after CI, and therefore, an iterative procedure would be required to obtain a proper solution. However, at the CIS (CI single excitations) level, commonly used for the spectroscopic calculations, this is no concern because of Brillouin’s theorem, which implies that the CI does not change the dipole moment of the molecule. Even at higher levels of excitation in CI, this effect should not be large and might be estimated from the respective perturbation operator.27 There are two approaches to address the instantaneous electronic polarization of the solvent during the excitation of the solute molecule. In the first case, the following correction term has to be added to the CI excitation energy ∆E I =
1 Γ(ε ∞ ) ψ 0 µ$ ψ 0 ψ I µ$ ψ I − ψ I µ$ ψ I 2
2
[11.1.64]
where Γ( ε ∞ ) is the reaction field tensor for the optical relative dielectric permittivity of the solvent, ε ∞ . In the last equation, the first term removes the incorrect term arising from the SCRF orbitals and energies in forming the CI matrix whereas the second term adds the response of the electronic polarization of the solvent to the dipole of the excited state. Equation [11.1.64] is first order in electron relaxation. Higher orders can be examined by defining the perturbation X ( I ) = λΓ(ε ∞ )
[ψ
0
µ$ ψ 0 − ψ I µ$ ψ I
]
[11.1.65]
2
which is clearly different for each excited state and would, if pursued, lead to a set of excited states that were nonorthogonal. In principle, these corrections need not to be small. Depending on the Fock operator used (equation [11.1.58] or [11.1.59]), the excitation energy from the ground state |ψ 0 ¦ to the excited state |ψ I ¦ of a solute molecule in a dielectric medium is given as follows $ψ − ψ H $ ψ + 1 Γ ψ µ$ ψ WI A − W0A = ψ I H I 0 0 0 0 2
[ ψ µ$ ψ I
I
− ψ 0 µ$ ψ 0
]−
660
Mati Karelson
1 − Γ(ε ∞ ) ψ I µ$ ψ I 2
[ψ
I
µ$ ψ I − ψ 0 µ$ ψ 0
]
[11.1.66]
in the first case, and $ψ − ψ H $ ψ − 1 Γ (ε ) ψ µ$ ψ W I B − W 0B = ψ I H I 0 0 ∞ I Ι 2
[ ψ µ$ ψ I
I
− ψ 0 µ$ ψ 0
]
[11.1.67]
in the second case. The last two equations are first order in electron polarization of the solvent. The second approach to the calculation of spectra in solutions is based on the assumption that the ground and excited states are intimately coupled in an instantaneous absorption process.28,29 In this model, the solute ground state electron distribution responds to the electron distribution in the excited state through the instantaneous polarization of the solvent. In such a case, the energy of the absorbing (ground) state is shifted by the following amount 1 1 1 Γ (ε∞ ) ψ 0 µ$ ψ 0 − ψ 0 µ$ ψ 0 − Γ (ε∞ ) ψ 0 µ$ ψ 0 ψ I µ$ ψ i − ψ 0 µ$ ψ 0 2 2 2
[11.1.68]
In the last equation, the first term removes the first order in the electron polarization part of the dielectric relaxation included in the SCRF of the ground state, and the second term adds back the appropriate interaction of the ground state with the “mean” reaction field, created by the excited state, |ψ I ¦ . This leads to the following equation for the excitation energy $ψ − ψ H $ ψ + 1 Γ ψ µ$ ψ WI A − W0A = ψ I H I 0 0 0 0 2
1 − Γ(ε ∞ ) ψ I µ$ ψ I 4
[ψ
I
µ$ ψ I − ψ 0 µ$ ψ 0
]
[ ψ µ$ ψ I
I
− ψ 0 µ$ ψ 0
2
]+
[11.1.69]
for the Fock operator [11.1.58] and
[
$ψ − ψ H $ ψ + 1 Γ (ε ) ψ µ$ ψ − ψ µ$ ψ W I B − W 0B = ψ I H I 0 0 ∞ I I 0 0 4
]
2
[11.1.70]
for the Fock operator [11.1.59]. However, it should be noticed that all four equations for spectral transition energies [[11.1.66], [11.1.67], [11.1.69] and [11.1.70]] yield very similar results when π → π * and n → π* transition solvatochromic shifts have been compared between the nonpolar and polar solvents.27 In Table 11.1.3, the results of the INDO/S (ZINDO)30,31 SCRF CIS calculated spectroscopic transition energies are given for some solvatochromic dyes. The relative shifts due to the solvent are reproduced theoretically in most cases. Even the absolute values of the spectroscopic transition energies are in satisfactory agreement with the respective experimental values, which demonstrates the applicability of the spectroscopic INDO/S parameterization for the spectra in solutions.
11.1 Theoretical treatment of solvent effects
661
Table 11.1.3. INDO/S SCRF CI calculated and experimental spectroscopic transition energies of some dyes in different solvents27 Molecule
Solvent
νcalc, cm-1
νexp, cm-1
(Scheme 23)
Gas phase Cyclohexane Water
29,700 26,300 22,500
27,400 23,300
(Scheme 24)
Gas phase n-Hexane Water
36,900 34,800 31,100
30,200 26,100
(Scheme 25)
Gas phase Chloroform Water
20,200 21,800 24,600
19,600 22,100
Scheme 23
Scheme 24
Scheme 25
The INDO/S SCRF CI method has been also successfully applied for the prediction of the solvatochromic shifts in various nitro-substituted porphyrins.32 The SCRF methodology has been employed also for the prediction of the solvatochromic shifts on emission spectra.33,34 A satisfactory agreement was obtained between the calculated and experimental fluorescence energies of p-N,N-dimethylaminobenzonitrile in different solvents. Finally, the solvent-induced shifts in the vibrational spectra of molecules have been also calculated using the SCRF theory.35 The SCRF approach has been also implemented for the treatment of solute-continuum solvent systems at the ab initio Hartree-Fock level of theory.36,37 In addition, a general SCRF (GSCRF) approach has been proposed to account for the interaction of the solvent reaction field with the arbitrary charge distribution of the solute molecule. According to this theory,38,39 the effective Hamiltonian of the solute in the solvent has the following form $ =H $ 0 + drΩ (r) V 0 (r) + dr ′G(r,r ′)Ω (r ′) H s s m s ∫ s ∫
[11.1.71]
where Ω s ( r ) is the solute charge density operator given by Ω s (r) = −∑ δ(r − ri ) + ∑ Z a δ(r − R a ) i
a
[11.1.72]
662
Mati Karelson
where ri stands for the i-th electron position vector operator, Ra is the position vector of the a-th nucleus with the charge Za in the solute and δ(r) is the Dirac’s delta function. The first term in square brackets in Eq. [11.1.71], Vm0 ( r ), represents the electrostatic potential created by the solvent in the absence of the solute and the second, integral term corresponds to the reaction potential response function of the polarizable solvent. Together these terms produce the reaction field potential applying to the solvent molecule in the polarizable dielectric medium. Notably, a principal part of the Hamiltonian [11.1.71] is the solute charge density that can be represented using different approximations of which the multipolar expansion has been mostly applied. By using the distributed multipole model, it is possible to obtain the GSCRF equations for the molecules of a complex shape. However, it has been mentioned that the use of multipole expansions of the solvent electrostatic and reaction potentials in Eq. [11.1.71] may cause this Hamiltonian to become unbound and special damping procedures have been invented to overcome this difficulty. The GSCRF total energy of the solute is given by the following equation $ 0 ψ + ψ drΩ ( r )V 0 ( r ) ψ + EGSCRF = ψ H ∫ s m
1 ψ dr Ω s ( r ) ψ 2 ∫
∫ dr ′G ( r, r ′)Ω ( r ′) [11.1.73] s
This energy expression can be used to build up the respective variational functional to get the molecular orbitals [above]. A crucial step in the general self-consistent reaction field procedure is the estimation of the solvent charge density needed to obtain the response function G(r,r') and the reaction potential. The use of Monte Carlo or molecular dynamics simulations of the system consisting the solute and surrounding solvent molecules has been proposed to find the respective solvent static and polarization densities. Several methods have been developed to account for the solute cavities of arbitrary shape in the solution. The polarizable continuum model (PCM) is based on the numerical integration of the relevant electrostatic equations describing the electrostatic interaction between the molecular charge distribution and the charge created on the boundary surface between the solute molecule and surrounding dielectric continuum.40-45 Within this method, the solute cavity is usually constructed from the overlapping van der Waals spheres of constituent atoms in the solute molecule and the solvent reaction field arising from the solute charge distribution is calculated numerically. Alternatively, the cavity can be defined as constructed from the electron isodensity surface around the solute molecule (IPCM).46 According to the classical electrostatics, the electrostatic potential at any point in the space can be described in terms of the apparent charge distribution, σ, on the cavity surface. It consists of two terms Φs = Φ M + Φ σ
[11.1.74]
the first of which (Φ M ) corresponds to the electrostatic potential created by the charge distribution of the solute and the second (Φ σ ) is due to the reaction potential by the solvent. The latter is directly connected with the apparent charge distribution on the surface of the cavity as follows: Φ σ (r) = ∫
Σ
σ(s) r −s
d 2s
[11.1.75]
11.1 Theoretical treatment of solvent effects
663
where Σ is the cavity surface and s vector defines a point on Σ. The Σ surface is usually divided into appropriate number of triangular small areas (tesserea), each of which has an area ∆Sk and contains the charge qk in some internal point sk. Thus, according to this, so-called boundary element method, the reaction potential is found as the following sum over all tesserea Φ σ (r) = ∑ k
qk r − sk
[11.1.76]
with q k = ∆S k σ(s k )
[11.1.77]
In the application of the boundary element method, it is crucial to select appropriate boundary surface for the solute cavity and to proceed as accurate as possible tessellation (triangulation) of this surface. For instance, it has been proposed that in the case of the cavity formation from overlapping van-der-Waals spheres, the atomic van-der-Waals radii should be multiplied by a coefficient equal to 1.2. Other possibilities of the surface definition include the closed envelope obtained by rolling a spherical probe of adequate diameter on the van-der-Waals surface of the solute molecule and the surface obtained from the positions of the center of such spherical probe around the solute. Within the quantum-mechanical theory, the PCM model proceeds from the following Schrödinger equation for a solute molecule in the dielectric continuum
(H$
0
)
+ V$PCM Ψ = EΨ
[11.1.78]
where the reaction field potential is given by equation [11.1.76] as follows qk V$PCM = ∑ k r − sk
[11.1.79]
The charges on the boundary are found from the electrostatic polarization of the dielectric medium on the surface of the cavity due to the potential derived from the charge distribution of the solute and from other (induced) charges on the surface. The induced surface charge is evaluated iteratively at each step of the SCF procedure to solve the Schrödinger equation [11.1.78]. It has been reported that a simultaneous iteration of the surface charge with the Fock procedure reduces substantially the computation time without the loss in the precision of calculations. Also, it has been shown that the expressions which determine the charges qk may be given as a set of linear equations. In the matrix form42 Dq 0f = Ein
[11.1.80]
where D is a square nonsymmetric and nondiagonal matrix with the dimension equal to the number of surface elements, derived from the curvature of the surface and q0f is a column
664
Mati Karelson
vector, containing the unknown surface charges. Ein is also a column vector collecting the effective components of the solute electric field multiplied by the surface elements r [11.1.81] E in0 , k = −∆S k ∇Φ M , in (s k )n k The D matrix depends only on the shape of the cavity and the dielectric constant of the medium. Therefore, when the system of linear equations [11.1.80] has to be solved several times with different Ein, as in the case of the polarizable solute, it may be convenient to work with the single inversion matrix D-1. The PCM model has been implemented for the calculation of the electronic excitation energies of solvated molecules within the quantum-mechanical configuration interaction method.47,48 The respective final expression for the excitation energy from the ground state (0) to the I-th state has the following form ∆W (0 I ) = ∆ECI(0 I ) − +
1 [J20(PI − P0) + P0 T0(PI − P0)] + 2
[11.1.82]
1 [J2∞(PI − P0) + PI T∞(PI − P0)] 2
where PI and P0 are the electronic density matrices of the solute in the excited state and in the ground state, respectively, ∆ECI( 0 I ) = EI − E0 +
[
]
1 2( PI − P0 ) T0 P0 + ( PI − P0 ) J10 + J20 ( PI − P0 ) 2
[11.1.83]
are the diagonal elements of CI matrix, and T = C t W −1 ∂C$
[11.1.84]
$ J1 = CW −1 ∂MZ
[11.1.85]
J 2 = Z t M t W −1 ∂C$
[11.1.86]
$ B = Z t M t W −1 ∂MZ
[11.1.87]
$ In the last equations, C and ∂Care the matrices representing the electrostatic potential and the electric field generated from the electron distribution in the solute molecule, respec$ are the matrices representing the electrostatic potential and tively. The matrices M and ∂M the electric field generated from the nuclear charges in the solute molecule, respectively. The diagonal elements of the matrix W are defined as the following function of the dielectric constant of the solvent W ii =
ε+1 ε −1
[11.1.88]
The subscripts 0 and ∞ in the J1, J2, and T matrices in equation [11.1.82] correspond to the static and optical dielectric constant of the solvent. Equation [11.1.82] can be considered
11.1 Theoretical treatment of solvent effects
665
as an analog of equation [11.1.66] for the case of arbitrary cavity shape. The PCM-CI method has been applied for the calculation of solvatochromic shifts in the spectrum 4-[(4’-hydroxyphenyl)azo]-N-methylpyridine in a variety of solvents.48 An integral equation formalism (IEF) has been developed as particularly suitable for the description of solvent effects on spectral transition energies within the PCM model.49 The respective theoretical equations have been applied for the calculation of solvatochromic shifts of several carbonyl-group containing molecules at the self-consistent field (SCF), configuration interaction (CI) and multiconfiguration self-consistent (MC SCF) field level of theory. The calculated spectral shifts accompanying the transfer of a solvatochromic compound from the gas phase to water were comparable with the experimental data. In Table 11.1.4, the results of calculations are presented for three carbonyl compounds, formaldehyde, acetaldehyde and acetone. Table 11.1.4. The calculated and experimental solvatochromic shifts (from the gas phase to water) in the spectra of some carbonyl compounds (cm-1)49 Compound
∆SCF
CI(SDT)
CAS SCF
Exp.
Formaldehyde
1889
839
944
1700-1900a
Acetaldehyde
1854
979
1049
1700-1900a
Acetone
2273
1574
1259
1539-1889
a
an estimate from other compounds
The advantage of the PCM method is in that it is applicable to the solute cavity of practically any shape in the solution. However, it is not clear how precisely should the molecular cavity be defined bearing in mind the classical (quasi-macroscopic) representation of the solvent. It is difficult to perceive that the solvent, e.g., the water molecules, can produce the electrical polarization corresponding to the statistically average distribution in the macroscopic liquid at infinitely small regions on the cavity surface. However, it is conceivable that larger chemical groups in the molecules may possess their own reaction field created by their charge distribution and the reaction fields of other groups in the solute molecule. A multi-cavity self-consistent reaction field (MCa SCRF) has been proposed50 for the description of rotationally flexible molecules in condensed dielectric media. It proceeds from the observation that the interaction of the charge and higher electrical moments of a charge distribution in a spherical cavity with the corresponding reaction fields localized in the center of the cavity does not depend on the position of charge or (point) multipole centers in this cavity. Therefore, it is possible to divide a rotationally flexible solute molecule or a hydrogen-bonded molecular complex between two or more spherical cavities that embed the rotationally separated fragments of the solute or solute and solvent molecules, respectively. Assuming the classical Born-Kirkwood-Onsager charge density expansion (Eq. 11.1.27) for each of these fragments, the total energy of the solute in a dielectric medium can be expressed as a sum of terms that correspond to the energies arising from the interaction of the partial charge and the electric moments of a given molecular fragment with the reaction field of its own and the reaction fields of other fragments, as well as from the interaction between the reaction fields of different fragments. The Hartree-Fock-type equations derived from the variational functional for the total energy E can then be solved iteratively using the
666
Mati Karelson
SCRF procedure. The PCM approach has been further refined to account for the curvature of surface elements.4 Also, this approach has been applied within different quantum-chemical frameworks.47 An alternative method for the description of solute-continuum electrostatic interaction has been developed as based on the notion that the electrostatic equations referring to the boundary surface between the solute and dielectric medium can be substantially simplified if to assume that the solvent is a homogeneous ideally conducting medium. Within this method (called the COSMO method), the electrostatic screening energy of a solute is given by the following equation (in matrix form)51 1 ∆E = − QBA −1BQ 2
[11.1.89]
with the following matrix elements b ik ≈ t k − ri
−1
[11.1.90]
for the point charges and b ik ≈
∫
µ, ν∈ i
χ µ (r)χ ν (r) 3 d r tk −r
[11.1.91]
for the continuous charge distribution, and a kl ≈ t k − t l
−1
, k ≠ l, a kk ≈ 3.8S k
−1/ 2
[11.1.92]
In the last equations, tk denotes the position vectors of the centers of small surface elements k on the arbitrary cavity surface; ri are the position vectors of the point charges in the solute molecule; r is the vector for electronic charge position described on the atomic basis {χ µ ( r )} and Sk are the areas of the surface elements. In equation [11.1.82], Q is the matrix of source charges in the solute. The COSMO model has been extended to account for the solvents with any dielectric constant.52-54 Within the respective GCOSMO method,53 the surface charges σ(r) on the boundary between the solute and continuum solvent are first determined for the medium with the infinite dielectric constant under the assumption that the electrostatic potential on the surface S is zero. For a dielectric medium specified by the dielectric constant ε, the actual surface charges are then calculated by scaling the screening conductor surface charge σ(r) by a factor of f(ε) = (ε - 1)/ε. This scaling preserves the validity of the Gauss theorem for the total surface charge. The boundary element method is applied for the calculation of the surface charge density, with the boundary divided into small areas and the surface charge approximated as the point charge in the center of this area. The charges are calculated either using the charge distribution in the molecule or by minimizing variationally of the total electrostatic solvation energy. The total free energy of the system of solute and surface charges is then calculated within the Hartree-Fock theory as
11.1 Theoretical treatment of solvent effects
(
)
(
667
)
1 1 s s 0 0 + + −1 Etot = Pµν Hµν + Hµν + Gµν + Gµν − 2 f ( ε) Z B A BZ + Enn + Enes [11.1.93] 2
where Enn is solute nuclear-nuclear repulsion and Enes is the solvation energy related to the dispersion and repulsion between the solute and solvent, and cavity formation; H0µν and G0µν are the one-electron and two-electron parts of the Fock matrix for the isolated solute, respectively, and Pµν are the density matrix elements. The solvent perturbations to the corresponding operators have been expressed as Hsµν = −f (ε)Z+B+ A −1BL µν
[11.1.94]
G sµν = −f (ε) ∑ PλσL+λσ A −1L µν λ, σ
[11.1.95]
and
where A and B are the N x N square matrices (N - number of atomic nuclei in the solute molecule) with the elements defined by equations [11.1.90] and [11.1.92]. The matrices Lµν consist of the one-electron integrals [11.1.91]. The first and second derivatives needed for the calculation of the molecular potential surfaces and the respective solvent effects on vibrational spectra have been also supplied within the framework of GCOSMO approach.53 A semi-quantitative agreement between the computational and experimental results has been obtained for the vibrational frequencies of acetone in water. Several approaches have been developed to account for the electron correlation effects on the solvation energy of both the ground and the excited states of the molecule in the solution. A multiconfigurational self-consistent reaction field (MC SCRF) theory has been proposed as based on the classical Onsager’s reaction field model.55 Notably, the higher order electrical moments of the solute molecule and the respective reaction field in the solvent were taken into account within this method. Thus, the dielectric solvation energy of a solute in a given state embedded into a linear isotropic medium has been calculated as the product of the expectation values for the reaction field ¢R ml ¦ and the respective multipole charge moments ¢M ml ¦ of the solute55-58 E el = −
+ 1 R ml M ml ∑ 2 l, m
[11.1.96]
where M ml = ∑ Z k S lm (Rk ) − S lm
[11.1.97]
R ml = f M ml
[11.1.98]
k
and
668
Mati Karelson
fl =
(l + 1)(ε − 1) 1 1 + ε(l + 1) a − ( 2 l+1)
4π S lm (r) = 2l + 1
1/ 2
r lY lm (θ, ϕ)
[11.1.99]
[11.1.100]
where Ylm(θ, ϕ) are the Legendre’ polynomials. Interestingly, it had been suggested that the dispersion interaction energy between the solute and the solvent could be accounted for, in principle, in the framework of this approach as related to the full distribution of dielectric relaxation frequencies of the solvent. Thus, the formula for the MC SCRF solvation energy has been expressed as follows E el = −
1 2 f l (Tlm ) ∑ 2 l, m
[11.1.101]
with the terms Tlm obtained from the expectation values of the nuclear and electronic solvent operators Tlm = Tlmn − Tlme
[11.1.102]
Tlmn = ∑ Z a R lm (R a )
[11.1.103]
Tlme = R lm (r)
[11.1.104]
a
where Rlm are the special solvent effect integrals.55 The solvent contributions have been developed also for the multiconfigurational energy gradient and Hessian, necessary for the solution of the MC SCRF equations. Notably, the results of the model calculations on water molecule implied that the higher multipole terms might play a significant role in the total electrostatic solvation energy of the molecule. Thus, the quadrupole term consisted approximately 20% of the dipolar term whereas the 4-th order term was even more significant (~ 30% of the dipolar term). The MC SCRF method has been applied for the calculation of the solvent effects on the spectral transitions of water and formaldehyde in different media.59 The MC SCRF methodology has been further refined in the framework of the response theory approach.60 This development describes the response of the solute or the solute-solvent complex to a time-independent or time-dependent high-frequency perturbation such as the spectral excitation. In order to take into account the electron correlation effects, another combination of the self-consistent reaction field theory with the configuration interaction formalism has been introduced as follows.61 Within this approach, the usual CI wavefunction has been constructed as follows N
ΨCI = ∑ Ca Da a =1
[11.1.105]
11.1 Theoretical treatment of solvent effects
669
where Ca are the CI expansion coefficients and Da are the basis functions (Slater determinants or their linear combinations). In the case of orthogonal basis functions, the normalization condition of the function ΨCI is given as N
∑C a =1
2 a
=1
[11.1.106]
and the coefficients Ca are determined from the following equation
(H$
0
)
+ V$rf Ψ = EΨ
[11.1.107]
$ is $ 0 is the Hamiltonian for the molecule, unperturbed by the reaction field and V where H rf the reaction field perturbation. The latter can be presented, for example, in the framework of the boundary element method as follows (cf. Eq. [11.1.75]) σ(r ′) V$rf = ∫ d 2 r r − r′
[11.1.108]
where σ( r ′ ) is the charge density on the surface of the cavity. For the spectroscopic applications, it would be again instructive to separate the noninertial and inertial components of the electrostatic polarization of the dielectric medium. The first of them corresponds to the electrostatic polarization of the electron charge distribution in the solvent that is supposedly instantaneous as compared to any electronic or conformational transition of the solute. The second component arises from the orientational polarization of the solvent molecules in the electrostatic field of the solute. The noninertial polarization can be described by the optical dielectric permittivity of the solvent that corresponds to the infinite frequency of external electromagnetic field (ε ∞ ≈ n D2 ) whereas the inertial polarization represents the slow, orientational part of the total dielectric constant of the solvent, ε. In order to separate the noninertial polarization, it is helpful to determine the solute charge density as the sum of the respective nuclear and electronic parts ρ = ρ n + ρ e = ∑ δ(r − rA )Z A + ∑ Ca Cb (ab | r, r ′) = ∑ Ca Cb ρ ab [11.1.109] A
a, b
a, b
where δ(r - rA) is the Dirac’s delta function and (ab|r,r') are the elements of the single-determinant matrices of transitions between the configurations. Notably, the values of ρ ab = δ ab ρ n − ρ(ab | r, r ′)
[11.1.110]
do not depend on the coefficients Ca and Cb. The noninertial component of the polarization field, Φ ∞ ( r ), is always in equilibrium and thus it can be represented as follows Φ ∞ (r) = ∑ Ca Cb Φ (ab∞ ) (r)
[11.1.111]
a, b
where Φ (ab∞ ) ( r ) is the solution of equation [11.1.107] for the basic charge distribution ρab. Since the latter does not depend on the coefficients Ca and Cb, the values of Φ (ab∞ ) ( r ) can be
670
Mati Karelson
determined before the calculation of the wavefunction ΨCI [11.1.105]. The coefficients Ca can be calculated, as usual, from the matrix equation H|C¦ = E|C¦
[11.1.112]
where |C¦ denotes the column-vector of the coefficients and the elements of the matrix H are given as follows61 $ 0 D + C C d 3 rρ Φ ( ∞ ) + d 3 rρ Φ Hab = Da H ∑ a b∫ 0 b ab ab ab ∫
[11.1.113]
a, b
where Φ0 is inertial (nuclear) part of the polarization field. For a given Φ0, the set of equations [11.1.98) can be solved iteratively. Implicitly, the last equations describe both the electron subsystems of the solute and solvent, the latter being taken into account as the field of noninertial (electron) polarization of the solvent, Φ ( ∞ ) . The electron correlation effects on the solvation energy of a solute have been also accounted for within the framework of the perturbation theory.62,63 By starting from the Hamiltonian of the solute molecule as follows $ =H $ 0 − M mf m ψ M m′ ψ H l l l′
[11.1.114]
where ψ denotes the exact (correlated) wavefunction, the Hartree-Fock operator may be written as F = F 0 − M lm f l m ψ 0 M lm′ ′ ψ 0
[11.1.115]
where ψ 0 denotes the electronic wavefunction at the Hartree-Fock level. The Hamiltonian may be then written as a perturbed expression of the Hartree-Fock operator
(
$ =F + H $ 0 − F0 H
)+M f ( ψ m m l l
0
M lm′ ′ ψ 0 − ψ M lm′ ′ ψ
)
[11.1.116]
The perturbation has two contributions, the standard Møller-Plesset perturbation and the non-linear perturbation due to the solute-solvent interaction. If C(i)j denotes the coefficient of the eigenstate |J¦ in the corrections of ψ to the i-th order, then the perturbation operators H(i) of the i-th order are given by the following formulae $ (1) = H $ 0 − F0 H
[11.1.117]
$ ( 2 ) = −2 C (1) M m f m 0M m ′ I H ∑ I l l l′
[11.1.118]
I≠0
$ ( 3 ) = −2 C ( 2 ) M m f m 0M m ′ J − H ∑ J l l ∑ ∑ CJ(1)CP(1)M lmf l m I M lm′ ′ P l′ J ≠0
J ≠0 P ≠0
[11.1.119]
11.1 Theoretical treatment of solvent effects
671
The calculation is performed, as usual, by comparing the coefficients of the Schrödinger equation to successive orders. The first order energy is the same as given by the usual Møller-Plesset treatment, $ 0 |0 > and the first order electrostatic contribution to the free energy of solvation is < 0| H identical to the result obtained at the Hartree-Fock level of theory. The second order correction to the free energy is given as ∆G s(2 ) = 2∑CS(2 ) 0M lm′ ′ S fl m 0M lm′ ′ 0 + S≠0
∑ ∑C
D≠0 D ′≠0
CD(1)′ D M lm′ ′ D ′ fl m 0M lm′ ′ 0
(1) D
[11.1.120]
where |S¦ stands for the singly excited states, and |D¦ and |D'¦ for a pair of doubly excited states different by just one orbital. Without excessive difficulty, it is possible to derive the correction terms to the electrostatic free energy of solvation of higher orders. A many-body perturbation theory (MBPT) approach has been combined with the polarizable continuum model (PCM) of the electrostatic solvation.64-66 The first approximation called by authors the perturbation theory at energy level (PTE) consists of the solution of the PCM problem at the Hartree-Fock level to find the solvent reaction potential and the wavefunction for the calculation of the MBPT correction to the energy. In the second approximation, called the perturbation theory at the density matrix level only (PTD), the calculation of the reaction potential and electrostatic free energy is based on the MBPT corrected wavefunction for the isolated molecule. At the next approximation (perturbation theory at the energy and density matrix level, PTED), both the energy and the wave function are solvent reaction field and MBPT corrected. The self-consistent reaction field model has been also applied within the complete active space self-consistent field (CAS SCF) theory12,67 and the complete active space second-order perturbation theory.12,67,68 Several groups69-73 have also proposed the quantum mechanical density functional theory (DFT) based methods for the calculation of the electrostatic solvation energy in dielectric media. However, the application of this theory for excited states is not straightforward.74,75 11.1.4 THEORETICAL TREATMENT OF SOLVENT DISPERSION EFFECTS ON ELECTRONIC-VIBRATIONAL SPECTRA OF MOLECULES The dispersion interaction between two atomic or molecular systems can be theoretically presented at different levels of theory.76-78 The modelling of the dispersion interactions in condensed media is more complicated and proceeds either from the discrete molecular description of the liquid or from the continuum model. According to a contemporary classification,4 the theoretical approaches to the dispersion effect in solutions can be divided into following classes: • pair-potential approaches • reaction field based approaches • cavity surface-dispersion energy relationship approaches The pair-potential approach is based on the discrete representation of the pairs of solvent and solute molecules or some fragments of them. The respective dispersion potentials are expressed as truncated asymptotic expansions in powers of 1/r, the reciprocal of the distance between the interacting entities4 Ums (disp) =
∑d
k = 6, 8,10
k −k ms ms
r
[11.1.121]
672
Mati Karelson
where the indexes m and s denote the structural entities (atoms, bonds, chemical groups) belonging to the solute and solvent molecules, respectively. The powers in expansion [11.1.121] are based on the formal theory of two-body interactions. In the practical calculations, only the first term of the expansion (k = 6) is frequently applied. The expansion coefficient d (6) ms can be calculated using the London formula I I 3 (6 ) d ms = − α m αs m s 2 I m + Is
[11.1.122]
where αm and αs are the isotropic polarizabilities for interacting systems and Im and Is are the mean excitation energies of these systems. This approximate formula is, in principle, valid only for interacting atoms. In the case of molecular systems, the atomic or group polarizabilities and local excitation energies are, as a rule, not isotropic and require the use of the respective tensor quantities. The absence of information about of accurate solute-solvent atom-atom distribution functions in dense media complicates further the accurate treatment of the dispersion interaction. These distribution functions can be calculated either using the molecular dynamics or Monte Carlo computer simulations or from the experimental scattering data on the respective systems. However, almost all these methods give only the averaged distribution functions and lack, therefore, the information about the local anisotropy of the atom-atom distributions. Similarly to the treatment of electrostatic effects, the dispersion potential can be limited to the dipole-dipole term and the mean excitation energies are approximated by the respective ionization potentials for the solute and solvent molecules. Thus, when a small cluster of solvent molecules surrounds the solute molecule, the first approximation of the dispersion energy can be presented by the following formula:79 UMS (disp) = −
x I M IS 4 I M + IS
BM
BS
∑ ∑ {r u =1 v =1
}
Tr [Tuv A u Tuv Av ]
−6 uv
[11.1.123]
where BM and BS are the number of bonds in the solute and in the solvent molecules, respectively, Im and Is are the corresponding mean excitation energies, Tuv is the tensor Tuv = 3
ruv r ⊗ uv − 1 ruv ruv
[11.1.124]
where ruv and ruv are the distance and the radius-vector between the bonds u and v, respectively, and Au is the polarizability tensor for bond u. The factor x in equation [11.1.123] is introduced to achieve the agreement between the molecule-molecule pair dispersion potential and a simpler expression derived on the basis of assumption that the dispersion energy between two molecules may be reduced to the sum of independent atom-atom contributions80 ( 6 ) −6 UMS (disp) = ∑ ∑ d ms rms m
s
[11.1.125]
11.1 Theoretical treatment of solvent effects
673
A scheme has been developed that reduces the spatial representation of the dispersion interaction into a surface representation of this interaction.4 According to this approach, the average dispersion-repulsion energy of a solute-solvent system has been written as follows: E disp − rep = ∫ L ∫ U(Ω) g (Ω)dΩ
[11.1.126]
where Ω stands for the set of all coordinates of the molecules involved, g(Ω) is the solute-solvent pair distribution function and U(Ω) is expressed as a sum of two-body dispersion-repulsion potentials. In the case of the fixed geometry of the solute molecule E disp − rep = nS ∑ NS s∈S
∑ ∑d ∫ r (k ) ms
m∈ M k
(k ) ms
3 g ms (rms )drms
[11.1.127]
The integrals in the last formula can be limited only to a certain minimum distance defined, for instance, by the van-der-Waals envelopes of interacting molecules. By introducing the auxiliary vector functions A (k) ms ( r ms ) such that r (k ) (k ) − k [11.1.128] rms g ms (rms ) ∇Ams (rms ) = d ms the average dispersion-repulsion energy between the solute and solvent molecules in solution may be written as follows E disp − rep = nS ∑ NS s∈S
∑∑ ∫A
m∈ M k
(k ) ms
n σdσ
[11.1.129]
Σs
where nσ is the outer normal to the surface Σs at the position σ. The integral in the last equation may be calculated numerically using an appropriate partitioning (tessellation) of the surface. A quantum-mechanical method of calculation of the dispersion energy has been developed on the basis of the above-cited semiclassical Abe’s theory.81 According to this method, the dispersion energy, Edisp , for a solute molecule in a spherical cavity is given as follows E disp
2 1 =− 3 aS3 a M3
∑ ∑E J ≠ I K ≠O
(µ ) (µ ) M IJ
S K
2
S KO
2
− E OS + E JM − E IM
[11.1.130]
where the superscript S refers to the solvent molecule and the superscript M to the solute S are the transition dipoles between the respective states of the molecule. Thus, µ IJM and µ KO solute (I and J) and the solvent (K and O) molecules. In equation [11.1.130], EKS , E SO and M EM J , EI denote the energies of the K-th and O-th state of the solvent and of the J-th and I-th state of the solute molecule, respectively. The cavity radii for the solvent and solute molecules are denoted as aS and aM, respectively.
674
Mati Karelson
Table 11.1.5. The INDO/CI calculated solvatochromic shifts ∆ν (from the gas phase to cyclohexane) of some aromatic compounds and the respective experimental data in low polarity solvents (cm-1)81 Compound
∆ν (calc)
∆ν (exp)
1
B2u
-316
-209a
Lb(x) Lb(y)
-332 -879
-300b; -275a -950b; -902a
Lb(1Bu) La(1Bu) 1 Bb(1Bu)
-243 -733 -1666
-252a -1030a -1620a
+162 -288 -446 -1475
+164c -333c -285c -1650c
Transition
Benzene 1
Naphthalene
1 1
Chrysene
1
1
Lb(y) La(x) 1 Kb(y) 1 Bb(x) 1
Azulene
a
in n-pentane, bin cyclohexane, cin 2-chloropropane
The equation [11.1.130] has been used within the semiempirical quantum-chemical INDO/CI formalism to calculate the solvent shifts of some aromatic compounds in cyclohexane.81 The results compare favorably with the experimental data for some nonpolar solvents (cf. Table 11.1.5). 11.1.5 SUPERMOLECULE APPROACH TO THE INTERMOLECULAR INTERACTIONS IN CONDENSED MEDIA The supermolecule approach to the calculation of solute-solvent interaction energies is based on the discrete molecular representation of the solvent. The supermolecule can be treated quantum-mechanically as a complex consisting of the central solute molecule and the surrounding closest solvent molecules. This supermolecule complex can be treated individually or as submerged into the dielectric continuum.82 In the last case, some continuum theory (SCRF, PCM) is applied to the supermolecule complex consisting of the solute molecule and the solvent molecules in its first coordination sphere.83-86 Therefore, the short-range solute-solvent electron correlation, dispersion and exchange-repulsion interactions are taken into account explicitly at the quantum level of theory as the electrons and nuclei both from the solute and solvent are included explicitly in the respective Schrödinger equation. The long-range electrostatic polarization of the solvent outside the first coordination sphere is, however, treated according to the dielectric continuum theory. Thus, the energy of solvation of a solute molecule can be expressed as follows: $ (S ) Ψ $ (0 ) Ψ $ (0 ) Ψ E sol = ΨSM H − ΨSM H − n ΨSM H SM SM SM SM M S
[11.1.131]
$ (0) and H $ (0) are $ (S) is the Hamiltonian for the supermolecule in the solution, and H where H SM M S the Hamiltonians for the isolated solute and the solvent molecules, respectively. In the last equation, n denotes the number of the solvent molecules applied in the supermolecule, ΨSM is the total wavefunction of the supermolecule immersed into dielectric medium, and ΨM and ΨS are the wavefunctions for isolated solute and solvent molecules, respectively.
11.1 Theoretical treatment of solvent effects
675
For instance, the n → π * electronic transition (11A1 →11A2) of formaldehyde solvated by varying number of water molecules has been investigated using multi-reference CI calculations.87 This simple supermolecule approach has given already satisfactory results as compared to experimental shifts in liquid water. However, it has been shown that in general, both the short-range quantum mechanical effects and the long-range solvent polarization play important role in determining the spectral shifts in liquid media. Thus, the INDO/S SCRF CIS theory alone has explained the solvatochromic shifts of azoles in different solvents, except those observed in water (Table 11.1.6).85 In both the water and acetonitrile, the compounds are predicted to have practically the same shift that is not the case in experiment. The explicit bonding of two water molecules to the nitrogen lone pairs leads in the cases of pyrimidine and pyridazine to the calculated large red shift instead of the experimentally observed solvatochromic blue shift. However, by treating the complex of an azole and two water molecules quantum-mechanically in the surrounding reaction field leads to quantitatively correct blue shifts. Table 11.1.6. The INDO/S SCRF/CI calculated and experimental spectral transition energies in different solvents for azoles (cm-1)85 νcalc, cm-1
νexp, cm-1
Pyrimidine
gas phase isooctane diethyl ether acetonitrile water 2H2O water + 2H2O
32966 33559 34127 34697 34743 30982 36572
34200 34400 34800 36900 36900 36900
Pyridazine
gas phase isooctane diethyl ether acetonitrile water 2H2O water + 2H2O
28329 29460 30382 31296 31368 26490 33927
29740 30150 31080 33570 33570 33570
Pyrazine
gas phase isooctane diethyl ether acetonitrile water 2H2O water + 2H2O
30387 30387 30387 30387 30387 32900 33301
31610 31610 31740 33160 33160 33160
Molecule
Solvent
As a general remark, in the calculations of the intermolecular interactions using the supermolecule approach, the “size-extensivity”88 of the methods applied is of crucial importance. Furthermore, the interaction energies calculated in the supermolecule approach usually suffer from what is called the basis set superposition error (BSSE),89 a spurious energy improvement resulting from the use of truncated basis sets. This error seems to be unavoidable in most practical calculations except for very small systems.90
676
Mati Karelson
The intermolecular interactions that correspond to the fixed geometry of the solute-solvent complex can be also studied by a perturbation approach.91,92 It has been suggested that the perturbation theory has some advantages over the supermolecule approach and may therefore be considered conceptually more appropriate for the calculation of intermolecular interaction energies. In this case, the interaction energy is calculated directly and it may be separated into components of well-defined physical meaning. Within the direct reaction field (DRF) method,93-96 the classical part of the solute-solvent system (solvent) is treated as a distribution of the polarizable point dipoles, interacting with each other. The DRF Hamiltonian of the solute-solvent system is thus given by the following formula: $ $0 1 H ∑ ∑ Fip+ α pq Fjq DRF = H − 2 i, j p, q
[11.1.132]
where indices i an j correspond to the solute particles (electrons and nuclei) and p and q run over the external polarizable points. Fip is the field of the particle i at the position p, and αpq gives the induced dipole at point q by a field applied at point p. The respective Schrödinger equation can be solved directly, without the iterative adjustment of the solvent charge distribution and the respective reaction field potential. The DRF method proceeds from the direct reaction field obtained as the linear solute-solvent interaction operator, proportional to the square of the electric field operator while the GSCRF approach uses the average reaction field model. It has been suggested that the additional energy contributions can be interpreted as due to the dispersion interaction between the solute and solvent molecules.97 More recently, the DRF approach has been combined with the continuum approach by dividing the space around the solute into a closer surrounding treated by direct reaction field method and to more distant space represented by the macroscopic dielectric properties of the solvent.98,99 A good quantitative agreement has been obtained between the experimental and DRF calculated solvatochromic shifts of the n → π * transition of acetone in different solvents.100,101 Nevertheless, even at the highest theoretical level accessible for practical calculations, the static approach is strictly valid only for the description of the molecular clusters of fixed geometry. However, in the cases of strong and weak intermolecular interactions, the energy of interaction in the molecular cluster in the gas phase or on the inert-gas matrix is substantially different from the total solvation energy in the condensed phase.102 A direct solution of a time-dependent Schrödinger equation for the condensed low-order bulk matter, needed to overcome this problem, is premature. Therefore, the molecular dynamics method (MD)103-105 based on the computer modelling of a system of molecules which interact by the known model potential to each other and undergoes the rotational and translational movement in the field caused by this interaction according to the classical (Newtonian) mechanics is widely applied for this purpose. By applying various boundary conditions and performing the calculation of the potential energies and forces for hundreds of thousands configurations obtained by step-by-step time evolution the time - averages such as internal energy (or enthalpy) can be obtained. An alternative is the stochastic Monte Carlo method that is based on the ergodic theorem and provides the ensemble averages calculated from randomly generated and weighted configurations. These methods suffer from several shortcomings, of which the problem of the applicability of the ergodic theorem (i.e., the identity
11.1 Theoretical treatment of solvent effects
677
of the time-averaged and ensemble-averaged thermodynamic and dynamic observables), the path sampling and difficulties to obtain precise intermolecular interaction potentials are most serious. Also, the real dense systems, i.e., liquids and solutions are intrinsically quantified systems and therefore a quantum molecular dynamics should be developed which accounts for the quantum effects in the microscopic system from the first principles. The combined quantum-mechanical/molecular dynamics (QM/MD) or quantum-mechanical/molecular mechanics (QM/MM) approaches have been used for the calculation of solvatochromic shifts in different media.106-108 Numerous computational schemes have been developed to calculate the total molecular solvation energy or free energy using the combination of different theoretical solute-solvent interaction models. From these, one of the most popular is the SMx methodology.109,110 This methodology proceeds from the division of the total molecular solvation energy into the solute-solvent electrostatic and inductive polarization terms, standard-state free energy of cavity creation in the solvent plus the solute-solvent dispersion interaction, and an empirical part of the nuclear motion free energy. The solvent polarization term is presented using the generalized Born formula: 1 1 N N G p = − 1 − ∑ ∑ q k q k ′ γ kk ′ 2 ε k =1 k ′=1
[11.1.133]
where the double summation is performed over all atomic partial charges qk in the solute molecule, ε is the relative dielectric permittivity of the solvent and γkk’ - the Coulomb’ integral between two centers k and k’, parameterized for the interactions with the solvent. The cavity creation plus dispersion term is calculated as N
0 = ∑ σ k Ak GCD
[11.1.134]
k =1
where N is the number of atoms in the solute, Ak is the solvent accessible surface area of a given atom and σk is the parameter for this atom that is called the accessible surface tension. The latter is obtained from the fit with the experimental data and is, thus, essentially an empirical parameter for a given type of atom. Thus, in essence the SMx methodology represents a semiempirical approach to the calculation of solvent effects. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14
P. Suppan, N. Ghoneim, Solvatochromism, The Royal Society of Chemistry, London, 1997. N.G. Bakhshiev (Ed.), Solvatochromism - Problems and Methods (in Russian), Leningrad Univ. Publ. House, Leningrad, 1989. M. Karelson, Adv. Quant. Chem., 28, 142 (1997). J. Tomasi, M. Persico, Chem. Rev., 94, 2027 (1994). T. Simonson, A.T. Brünger, J. Phys. Chem., 98, 4683 (1994). O. Sinanoglu, Chem. Phys. Lett., 1, 283 (1967). V. Gogonea, E. Osawa, J. Mol. Struct. (THEOCHEM), 311, 305 (1994). H. Reiss, Adv. Chem. Phys., 9, 1 (1966). R.A. Pierotti, Chem. Rev., 76, 717 (1976). R.M. Gibbons, Mol. Phys., 17, 81 (1969). M. Irisa, K. Nagayama, F. Hirata, Chem. Phys. Lett., 207, 430 (1993). A. Bernhardsson, R. Lindh, G. Karlström, B.O. Roos, Chem. Phys. Lett., 251, 141 (1996). W. Jaskólski, Phys. Rep., 271, 1 (1996). S. Basu, Adv. Quant. Chem., 1, 145 (1964).
678
15 16 17 18 19 20 21 22 23 24 25 26
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66
Mati Karelson A.T. Amos, B.L. Burrows, Adv,. Quant. Chem., 7, 289 (1973). R.J. Abraham, B.D. Hudson, M.W. Kermode, J.R. Mines, J. Chem. Soc. Farad. Trans., 84, 1911 (1988). J.G. Kirkwood, J. Chem. Phys., 7, 911 (1939). L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). C.J.F. Böttcher and P. Bordewijk, Theory of Electric Polarization, 2nd ed., vol. 2, Elsevier Co, Amsterdam, 1978. D. Rinaldi, M.F. Ruiz-Lopez, J.-L. Rivail, J. Chem. Phys., 78, 834 (1983). D. Rinaldi, J.-L. Rivail, N. Rguini, J. Comput. Chem., 13, 675 (1992). J.-L. Rivail, B. Terryn, D. Rinaldi, M.F. Ruiz-Lopez, J. Mol. Struct. (THEOCHEM), 166, 319 (1988). E.G. McRae, J. Phys. Chem., 61, 562 (1957). T. Abe, Bull Chem. Soc. Japan, 38, 1314 (1965). O. Tapia, O. Goscinski, Mol. Phys., 29, 1653 (1975). M. Karelson, G.H.F. Diercksen, “Models for Simulating Molecular Properties in Condensed Systems, in: Problem Solving in Computational Molecular Science. Molecules in Different Environments, S. Wilson and G.H.F. Diercksen (Eds.), Kluwer Academic Publishers, Dordrecht, 1997, pp. 215- 248. M. Karelson, M.C. Zerner, J. Phys. Chem., 92, 6949 (1992). R. Marcus, J. Chem. Phys., 24, 979 (1956). J. Jortner, Mol. Phys., 5, 257 (1962). J. Ridley, M.C. Zerner, Theor. Chim. Acta, 42, 223 (1976). A.D. Bacon, M.C. Zerner, Theor. Chim. Acta, 53, 21 (1979). M. Karelson, K. Pihlaja, T. Tamm, A.Uri, and M.C. Zerner, J. Photochem. and Photobiol. A Chemistry, 85, 119 (1995). A. Broo, Chem. Phys., 183, 85 (1994). A. Broo, M.C. Zerner, Chem. Phys. Lett., 227, 551 (1994). J.-L. Rivail, D. Rinaldi, V. Dillet, Mol. Phys., 89, 1521 (1996). M.M. Karelson, Org. React., 17, 357 (1980). M.W. Wong, K.B. Wiberg, M.J. Frisch, J. Am. Chem. Soc., 113, 4776 (1991). O. Tapia, F. Colonna, J. Ángyán, J. Chim. Phys., 87, 875 (1990). G. Jansen, F. Colonna, J. Ángyán, Int. J. Quant. Chem., 58, 251 (1996). S. Miertuš, E. Scrocco, J. Tomasi, Chem. Phys., 55, 117 (1981). H. Hoshi, M. Sakurai, Y. Inoue, R. Chûjô, J. Chem. Phys., 87, 1107 (1987). M.L. Drummond, J. Chem. Phys., 88, 5014 (1988). B. Wang, G.P. Ford, J. Chem. Phys., 97, 4162 (1992). T. Fox, N, Rösch, R.J. Zauhar, J. Comput. Chem., 14, 253 (1993). F.J. Luque, M. Orozco, P.K. Bhadane, S.R. Gadre, J. Chem. Phys., 100, 6718 (1994). C. Gonzalez, A. Restrepo-Cossio, M. Márquez, K.B. Wiberg, M. De Rosa, J. Phys. Chem. A, 102, 2732 (1998). M.A. Aguilar, J. Olivares del Valle, J. Tomasi, J. Chem. Phys., 98, 7375 (1993). H. Houjou, M. Sakurai, Y. Inoue, J. Chem. Phys., 107, 5652 (1997). B. Mennucci, R. Cammi, J. Tomasi, J. Chem. Phys., 109, 2798 (1998). M. Karelson, T. Tamm, M.C. Zerner, J. Phys Chem., 97, 11901 (1993). A. Klamt, G. Schüürmann, J. Chem. Soc. Perkin Trans. 2, 799 (1993). A. Klamt, J. Phys. Chem., 99, 2224 (1995). E.V. Stefanovich, T.N. Truong, J. Chem. Phys., 105, 2961 (1996). A. Klamt, J. Phys. Chem., 100, 3349 (1996). K.V. Mikkelsen, H. Ågren, H.J. Aa Jensen, T. Helgaker, J. Chem. Phys., 89, 3086 (1988). J.L. Rivail, D. Rinaldi, Chem. Phys., 18, 233 (1976). M.D. Newton, J. Phys. Chem., 79, 2795 (1975) K.V. Mikkelsen, E. Dalgaard, P. Swanstrom, J. Phys. Chem., 91, 3081 (1987). K.V. Mikkelsen, A. Cesar, H. Ågren, H.J. Aa. Jensen, J. Chem. Phys., 103, 9010 (1995). K.V. Mikkelsen, P. Jørgensen, H.J. Aa. Jensen, J. Chem. Phys., 100, 6597 (1994). M.V. Basilevsky, G.E. Chudinov, Chem. Phys., 157, 245 (1991). J.-L. Rivail, C. R. Acad. Sci. Paris, 311 II, 307 (1990). J.-L . Rivail, D. Rinaldi, M.F. Ruiz-Lopez, in: Theoretical and Computational Models for Organic Chemistry; S.J. Formosinho, I.G. Csizmadia, and L.G. Arnaut, (Eds.), Kluwer, Dordrecht. 1991, pp. 77-92. F.J. Olivares del Valle, J. Tomasi, Chem. Phys., 150, 139 (1991). F.J. Olivares del Valle, M.A. Aguilar, J. Comput. Chem., 13, 115 (1992). F.J. Olivares del Valle, M.A. Aguilar, J. Mol. Struct. (THEOCHEM), 280, 25 (1993).
11.1 Theoretical treatment of solvent effects
67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
106 107 108 109 110
679
N.A. Besley, J.D. Hirst, J. Phys. Chem. A, 102, 10791 (1998). L. Serrano-Andrés, M.P. Fülscher, G. Karlström, Int. J. Quant. Chem., 65, 167 (1997). R.R. Contreras, A.J. Aizman, Int. J. Quant. Chem. Quantum Chem. Symp., 25, 281 (1991). R.J.Hall, M.M. Davidson, N.A. Burton, I.H. Hillier, J. Phys. Chem., 99, 921 (1995). R.R. Contreras, P. Pérez, A.J. Aizman, Int. J. Quant. Chem., 56, 433 (1995). G.J. Tawa, R.L. Martin, L.R. Pratt, T.V. Russo, J. Phys. Chem., 100, 1515 (1996). T.N. Truong, E.V. Stefanovich, Chem. Phys. Lett., 240, 253 (1995). E.K.U. Gross, W. Kohn, Phys. Rev. Lett., 55, 2850 (1985). C.A. Ullrich, E.K.U. Gross, Phys. Rev. Lett., 74, 872 (1995). H. Margenau, N.R. Kestner, Theory of Intermolecular Forces, Pergamon Press, Oxford, 1971. J. Mahanty, W.B. Ninham, Dispersion Forces, Academic Press, New York, 1976. P. Hobza, P. Zahradnik, Intermolecular Complexes, Elsevier, Amsterdam, 1988. M.J. Huron, P. Claverie, Chem. Phys. Lett., 4, 429 (1969). A.J. Pertsin, A.I. Kitaigorodsky, The Atom-Atom Pair Potential Method, Springer, Berlin, 1986. N. Rösch, M.C. Zerner, J. Phys. Chem., 98, 5817 (1994). G.H.F. Diercksen, M. Karelson, T. Tamm and M.C. Zerner, Int. J. Quant. Chem., S28, 339 (1994). M.M. Karelson, Org. React., 20, 127 (1983). O. Tapia, J.M. Lluch, R. Cardenas, J. Andres, J. Am. Chem. Soc., 111, 829 (1989). M.M. Karelson, M.C. Zerner, J. Am. Chem. Soc., 112, 9405 (1990). L.C.G. Freitas, R.L. Longo, A.M. Simas, J. Chem. Soc. Faraday Trans., 88, 189 (1992). I. Frank, S. Grimme, M. von Arnim, S.D. Peyerimhoff, Chem. Phys., 199, 145 (1995). E. Clementi, G. Corongiu, (Eds.) Computational Chemistry, METECC-95, STEF, Cagliari, 1995. S.F. Boys, F. Bernardi, F. Mol. Phys., 19, 553 (1970). Z. Latajka, S. Scheiner, J. Comput. Chem., 8, 674 (1987). P. Arrighini, P. Intermolecular Forces and Their Evaluation by Perturbation Theory, Lecture Notes in Chemistry, vol. 25, Springer-Verlag, Berlin, 1981. B.J. Jeziorski, R. Moszynski, A. Ratkiewicz, S. Rybak, K. Szalewicz, H.L. Williams, Chapter 3 in: Methods and Techniques in Computational Chemistry, METECC-94, vol. B. STEF, Cagliari, 1994. B.T. Thole, P.Th. van Duijnen, Chem. Phys., 71, 211 (1982). B.T. Thole, P.Th. van Duijnen, Theor. Chim. Acta, 63, 209 (1983). B.T. Thole, P.Th. van Duijnen, Biophys. Chem., 18, 53 (1983). P.Th. van Duijnen, A.H. Juffer, H. Dijkman, H. J. Mol. Struct. (THEOCHEM), 260, 195 (1992). A.H. deVries, P.Th. van Duijnen, Int. J. Quant. Chem., 57, 1067 (1996). H. Dijkman, P. Th. van Duijnen, Int. J. Quant. Chem. Quantum Biol. Symp., 18, 49 (1991). P.Th. van Duijnen, A.H. de Vries, Int. J. Quant. Chem. Quantum Chem. Symp., 29, 531 (1995). A.H. de Vries, P.Th. van Duijnen, Int. J. Quant. Chem., 57, 1067 (1996). F.C. Grozema, P.Th. van Duijnen, J. Phys. Chem. A, 102, 7984 (1998). F.H Stillinger, Adv. Chem. Phys., 31, 1 (1975). S. Engström, B. Jönsson, R.W. Impey, J. Chem. Phys., 80, 5481 (1984). W.F. van Gunsteren, Current Opinion in Structural Biology, 3, 277 (1993). C. Corongiu, V. Martorana, Chapter 3 in: Methods and Techniques in Computational Chemistry, METECC-94; vol. C. STEF, Cagliari, 1994.; W.R.P. Scott, W.F. van Gunsteren, Chapter 9 in: Methods and Techniques in Computational Chemistry, METECC-95; Ed. by E. Clementi and G. Corongiu, STEF, Cagliari, 1995. J.T. Blair, K. Krogh-Jespersen, R.M. Levy, J. Am. Chem. Soc., 111, 6948 (1989). V. Luzhkov, A. Warshel, J. Am. Chem. Soc., 113, 4491 (1989). J.S. Bader, C.M. Cortis, B.J. Berne, J. Chem. Phys., 106, 2372 (1997). C.J. Cramer, D.G. Truhlar, J. Am. Chem. Soc., 113, 8305 (1991). C.J. Cramer, D.G. Truhlar, J. Am. Chem. Soc., 113, 8552 (1991).
680
Tai-ichi Shibuya
11.2 DIELECTRIC SOLVENT EFFECTS ON THE INTENSITY OF LIGHT ABSORPTION AND THE RADIATIVE RATE CONSTANT Tai-ichi Shibuya Faculty of Textile Science and Technology Shinshu University, Ueda, Japan
11.2.1 THE CHAKO FORMULA OR THE LORENTZ-LORENZ CORRECTION The intensity of light absorption by a molecule is generally altered when the molecule is immersed in a solvent or transferred from one solvent to another. The change may be small if the solvents are inert and non-polar, but often a significant increase or decrease is observed. The first attempt to correlate such effects with the nature of the solvent was made by Chako1 in 1934. Chako’s formula reads as
(
)
n2 + 2 f ′′ = 9n f
2
[11.2.1]
where: f f′ n
oscillator strength of an absorption band of a molecule apparent oscillator strength of the molecule in solution refractive index of the solution at the absorbing frequency
The apparent oscillator strength is proportional to the integrated intensity under the molar absorption curve. To derive the formula, Chako followed the classical dispersion theory with the Lorentz-Lorenz relation (also known as the Clausius-Mosotti relation), assuming that the solute molecule is located at the center of the spherical cavity in the continuous dielectric medium of the solvent. Hence, the factor derived by Chako is also called the Lorentz-Lorenz correction. Similar derivation was also presented by Kortòm.2 The same formula was also derived by Polo and Wilson3 from a viewpoint different from Chako. Chako’s formula always predicts an increase of the absorption intensity with the refractive index. This does not hold, for instance, for the allowed π → π * electronic transitions of cyclohexadiene and cyclopentadiene,4 and monomethyl substituted butadienes.5 11.2.2 THE GENERALIZED LOCAL-FIELD FACTOR FOR THE ELLIPSOIDAL CAVITY A natural generalization of the Chako formula was made by generalizing the spherical cavity to an ellipsoidal cavity. Such a generalization was shown by Shibuya6 in 1983. The generalized formula derived by him reads as
[(
) ]
s n2 −1 + 1 f ′′ = f n
2
[11.2.2]
where: s
shape parameter which takes a value between 0 and 1
This parameter s is more generally known as the depolarization factor, whose values are listed for special cases in general textbooks.7 For the spherical cavity, s = 1/3 in any axis; for
11.2 Dielectric solvent effects
681
a thin slab cavity, s = 1 in the normal direction and s = 0 in plane; and for a long cylindrical cavity, s = 0 in the longitudinal axis and s = 1/2 in the transverse direction. The shape of the ellipsoidal cavity is supposed to be primarily determined by the shape of the solute molecule. Typical cases are long polyenes and large planar aromatic hydrocarbons. One can assume s = 0 for the strong π → π * absorption bands of these molecules. For smaller molecules, however, one should assume s ≈ 1/3 regardless of the shape of the solute molecule, as the cavity shape then may be primarily determined by the solvent molecules rather than Figure 11.2.1. Dependence of f ′ / f on the refractive in- the solute molecule. Note that Eq. [11.2.2] dex n for different values of s. [After reference 6] gives the Chako formula for s = 1/3, i.e., for the spherical cavity. For transitions whose moments are in the longitudinal axis of a long cylindrical cavity or in the plain of a thin slab cavity, Eq. [11.2.2] with s = 0 leads to f ′′ / f = 1/ n, so that the absorption intensity always decreases with the refractive index. If the transition moment is normal to a thin slab cavity, Eq. [11.2.2] with s = 1 leads to f ′′ / f = n 3 . The dependence of the ratio f ′′/f on the refractive index n according to Eq. [11.2.2] is illustrated for different values of s in Figure 11.2.1. The slope of the ratio is always positive for s > 1/4. For 0 < s < 1/4, it is negative in the region 1 ≤ n ≤ (1− s) / 3s and positive in the other region. Eq. [11.2.2] can be also written as the following form:
(nf ′′) =
(
)
fs n 2 − 1 + f
[11.2.3]
This equation shows a linear relationship between (nf ′′) and (n2 - 1). If a set of measured values of f ′′ vs. n are provided for a solute, the least-squares fitting to Eq. [11.2.3] of (nf ′′) against (n2 - 1) gives the values of f and s for the solute molecule. Note that f ′′ and f in Eq. [11.2.3] can be replaced by any quantities proportional to the oscillator strengths. Thus, they can be replaced by the integrated intensities or by their relative quantities. Figure 11.2.2 shows such plots for the π → π * absorption bands of β-carotene and the n → π * absorption bands of pyrazine measured8 in various organic solvents. Here, the relative intensities f ′′ / f c′′, where f c′′ is the absorption intensity measured in cyclohexane as the reference solvent, are considered, and y = nf ′′ / f c′′ is plotted against x = n2 - 1. The least-squares fittings give s = 0 for the allowed π → π * transition of β-carotene and s = 0.29 for the vibronic n → π * transition of pyrazine. Note that in this case the least-squares fitted line gives f / f c′′ as its intercept and s f / f c′′as its slope so that s is given as the ratio of the slope divided by the intercept. A similar study was made9 on the n → π * absorption bands of acetone and cyclopentanone, giving the results s = 0.88 and f = 1.8×10-4 for acetone and s=0.72 and f = 2.2×10-4 for cyclopentanone. A similar generalization was also made by Buckingham.10 He followed Kirkwood’s 11 idea in deriving the electric moment of a dielectric specimen produced by a fixed mole-
682
Tai-ichi Shibuya
cule in its interior and Scholte’s extension12 of the cavity field and the reaction field in the Onsager-BØttcher theory13,14 to an ellipsoidal cavity. Buckingham’s formula involves the polarizability of the solute molecule and appears quite different from Eq. [11.2.2]. It was shown6 that the Buckingham formula reduces to Eq. [11.2.2]. 11.2.3 DIELECTRIC SOLVENT Figure 11.2.2. Plots of y = nf ′ / f c′ vs. x = n2 - 1 for the π → π * EFFECT ON THE absorption bands of β-carotene (crosses) and the n → π * abRADIATIVE RATE sorption bands of pyrazine (solid circles). [After reference 6] CONSTANT The radiative rate constant is related to the absorption intensity of the transition from the ground state to the excited state under consideration. The application of Eq. [11.2.2] leads15 to
[(
) ]
k ′′r / k r = n s n 2 − 1 + 1
2
[11.2.4]
where: k r′ kr
apparent radiative rate constant of the solute molecule measured in a solvent of the refractive index n radiative rate constant of the molecule in its isolated state
Note that the local-field correction factor n[s(n2 - 1) + 1]2 varies from n to n5 as s varies from 0 to 1. For 9,10-diphenylanthracene (DPA), the correction factor was given15 as n[(0.128)(n2 - 1) +1]2, which lies between n and n2. This agrees with the observed data16 of fluorescence lifetimes of DPA in various solvents. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
N. Q. Chako, J. Chem. Phys., 2, 644 (1934). G. Kortòm, Z. Phys. Chem., B33, 243 (1936). (a) V. Henri and L. W. Pickett, J. Chem. Phys., 7, 439 (1939); (b) L. W. Pickett, E. Paddock, and E. Sackter, J. Am. Chem. Soc., 63, 1073 (1941). L. E. Jacobs and J. R. Platt, J. Chem. Phys., 16, 1137 (1948). S. R. Polo and M. K. Wilson, J. Chem. Phys., 23, 2376 (1955). T. Shibuya, J. Chem. Phys., 78, 5176 (1983). C. Kittel, Introduction to Solid State Physics, 4th Ed., Wiley, New York, 1971, Chap. 13. A. B. Myers and R. R. Birge, J. Chem. Phys., 73, 5314 (1980). T. Shibuya, Bull. Chem. Soc. Jpn. , 57, 2991 (1984). A. D. Buckingham, Proc. Roy. Soc. (London), A248, 169 (1958); A255, 32 (1960). J. G. Kirkwood, J. Chem. Phys., 7, 911 (1939). T. G. Scholte, Physica (Utrecht), 15, 437 (1949). L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). C. J. F. BØttcher, (a) Physica (Utrecht), 9, 937, 945 (1942); (b) Theory of Electric Polarization, Elsevier, New York, 1952; 2nd Ed., 1973, Vol. I. T. Shibuya, Chem. Phys. Lett., 103, 46 (1983). R. A. Lampert, S. R. Meech, J. Metcalfe, D. Phillips, A. P. Schaap, Chem. Phys. Lett., 94, 137 (1983).
12
Other Properties of Solvents, Solutions, and Products Obtained from Solutions 12.1 RHEOLOGICAL PROPERTIES, AGGREGATION, PERMEABILITY, MOLECULAR STRUCTURE, CRYSTALLINITY, AND OTHER PROPERTIES AFFECTED BY SOLVENTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
12.1.1 RHEOLOGICAL PROPERTIES The modification of rheological properties is one of the main reasons for adding solvents to various formulations. Rheology is also a separate complex subject which requires an in-depth understanding that can only be accomplished by consulting specialized sources such as monographic books on rheology fundamentals.1-3 Rheology is such a vast subject that the following discussion will only outline some of the important effects of solvents. When considering the viscosity of solvent mixtures, solvents can be divided into two groups: interacting and non-interacting solvents. The viscosity of a mixture of non-interacting solvents can be predicted with good approximation by a simple additive rule rule: i =n
log η = ∑ φi log η i
[12.1.1]
i =1
where: η i φ ηi
viscosity of solvent mixture iteration subscript for mixture components (i = 1, 2, 3, ..., n) fraction of component i viscosity of component i.
Interacting solvents contain either strong polar solvents or solvents which have the ability to form hydrogen bonds or influence each other on the basis of acid-base interaction. Solvent mixtures are complicated because of the changes in interaction that occurs with changes in the concentration of the components. Some general relationships describe vis-
684
George Wypych
cosity of such mixtures but none is sufficiently universal to replace measurement. Further details on solvent mixtures are included in Chapter 9. The addition of solute(s) further complicates rheology because in such mixtures solvents may not only interact among themselves but also with the solute(s). There are also interactions between solutes and the effect of ionized species with and without solvent participation. Only very dilute solutions of low molecular weight substances exhibit Newtonian viscosity. In these solutions, viscosity is a constant, proportionality factor of shear rate and shear stress. The viscosity of these solutions is usually well described by the classical, Einstein’s equation: η = ηs (1 + 2.5φ)
[12.1.2]
where: ηs φ
solvent viscosity volume fraction of spheres (e.g. suspended filler) or polymer fraction
If φ is expressed in solute mass concentration, the following relationship is used: φ=
NVc M
[12.1.3]
where: N V c M
Avogadro’s number molecular volume of solute ((4/3)πR3) with R - radius solute mass concentration molecular weight
Combination of equations [12.1.2] and [12.1.3] gives: η − ηs 2.5NV = ηs c M
[12.1.4]
The results of studies of polymer solutions are most frequently expressed in terms of intrinsic, specific, and relative viscosities and radius of gyration; the mathematical meaning of these and the relationships between them are given below: η − ηs [η] = lim c→ 0 ηs c ηsp c
= [η] + k1 [η] 2 c + ⋅ ⋅ ⋅
ln η r = [η] − k1′ [η] 2 c + ⋅ ⋅ ⋅ c η r = ηsp + 1 =
η ηs
[12.1.5]
[12.1.6]
[12.1.7]
[12.1.8]
12.1 Rheological properties, aggregation, permeability
1000
685
1.2
Log viscosity, cP
Viscosity, Pa s
1 100
10
0.8 0.6 0.4
1 0.001 0.01
0.1
1
Shear rate, s-1
10
100
0.2
-1
-0.8 -0.6 -0.4 -0.2
0
0.2
-1
Log concentration, g dl
Figure 12.1.1. Viscosity vs. shear rate for 10% solution Figure 12.1.2. Viscosity of polyphenylene solution in pyrilidinone. [Data from F. Motamedi, M Isomaki, of polyisobutylene in pristane. [Data from C R Schultheisz, G B McKenna, Antec ‘99, SPE, New M S Trimmer, Antec ‘98, SPE, Atlanta, 1998, p. 1772.] York, 1999, p 1125.] where: [η] ηsp ηr k1 k1′
intrinsic viscosity specific viscosity relative viscosity coefficient of direct interactions between pairs of molecules coefficient of indirect (hydrodynamic) interactions between pairs of molecules
In Θ solvents, the radius of gyration of unperturbed Gaussian chain enters the following relationship: [η] 0 =
Φ 0 Rg3, 0 M
[12.1.9]
where: Φ0 R g,0
coefficient of intramolecular hydrodynamic interactions = 3.16± 0.5×1024 radius of gyration of unperturbed Gaussian chain
In good solvents, the expansion of chains causes an increase of viscosity as described by the following equation: [η] =
Φ 0 α η3 Rg3, 0 M
[12.1.10]
where: αη
= [ η]1/3 / [ η]10/3 is and effective chain expansion factor.
Existing theories are far from being universal and precise in prediction of experimental data. A more complex treatment of measurement data is needed to obtain characteristics of these “rheological” liquids. Figure 12.1.1 shows that the viscosity of a solution depends on shear rate. These data comes from the development of a standard for instrument calibration by NIST to improve
686
George Wypych
109
500 400
7 -1
10
Shear rate, s
Viscosity, Poise
108
106 10
5
10
4
300 200 100
1000 100 1000
0 4
5
6
10 10 10 Molecular weight, Daltons
10
7
0
5
10
15
20
-6
Molecular weight x 10
Figure 12.1.3. Viscosity of 40% polystyrene in Figure 12.1.4. Shear rate of polystyrene in DOP vs. modi-2-ethyl hexyl phthalate. [Data from G D J Phillies, lecular weight. [Data from M Ponitsch, T Hollfelder, J Springer, Polym. Bull., 40, No.2-3, 345-52 (1998).] Macromolecules, 28, No.24, 8198-208 (1995).]
1.2
1.4
DXN
1.1
1.2 CCl
1
4
rel
0.9 0.8
log η
Viscosity, mPa s
1
DCE
0.7
0.6
associated unassociated
0.4
0.6
CHCl3
0.5 THF 0.4
0.8
TOL
0.2
DCM
0
0
10
20
30
40
50
PMMA concentration, % Figure 12.1.5. Viscosity of PMMA solutions in different solvents vs. PMMA concentration. Basic solvents: tetrahydrofuran, THF, and dioxane, DXN; neutral: toluene, TOL and CCl4; acidic: 1,2-dichloroethane, DCE, CHCl3, and dichloromethane, DCM. [Adapted, by permission, from M L Abel, M M Chehimi, Synthetic Metals, 66, No.3, 225-33 (1994).]
0
0.5
1
1.5
2
2.5
3
Concentration, g dl-1 Figure 12.1.6. Relative viscosity of block copolymers with and and without segments capable of forming complexes vs. concentration. [Data from I C De Witte, B G Bogdanov, E J Goethals, Macromol. Symp., 118, 237-46 (1997).]
the accuracy of measurements by application of nonlinear liquid standards.4 Figure 12.1.2 shows the effect of polymer concentration on the viscosity of a solution of polyphenylene in N-methyl pyrilidinone.5 Two regimes are clearly visible. The regimes are divided by a critical concentration above which viscosity increases more rapidly due to the interaction of chains leading to aggregate formation. These two sets of data show that there
12.1 Rheological properties, aggregation, permeability
400
80 MEK toluene
PEO/PAA PEO PAA
350 300
60
Viscosity, Poise
3
Intrinsic viscosity, cm g
-1
70
50 40 30 20 10
687
250 200 150 100 50
0
20
40
60
80
100
PVAc, wt% Figure 12.1.7. Intrinsic viscosity of PS/PVAc mixtures in methyl-ethyl-ketone, MEK, and toluene. [Data from H Raval, S Devi, Angew. Makromol. Chem., 227, 27-34 (1995).]
0
0
2
4
6
8
10
12
14
pH Figure 12.1.8. Viscosity of poly(ethylene oxide), PEO, poly(acrylic acid), PAA, and their 1:1 mixture in aqueous solution vs. pH. [Adapted, by permission, from I C De Witte, B G Bogdanov, E J Goethals, Macromol. Symp., 118, 237-46 (1997).
are considerable departures from the simple predictions of the above equations because, based on them, viscosity should be a simple function of molecular weight. Figure 12.1.3 shows, in addition, that the relationship between viscosity of the solution and molecular weight is nonlinear.6 Also, the critical shear rate, at which aggregates are formed, is a nonlinear function of molecular weight (Figure 12.1.4).7 These departures from simple relationships are representative of simple solutions. The relationships for viscosities of solution become even more complex if stronger interactions are included, such as the presence of different solvents, the presence of interacting groups within polymer, combinations of polymers, or the presence of electrostatic interactions between ionized structures within the same or different chains. Figure 12.1.5 gives one example of complex behavior of a polymer in solution. The viscosity of PMMA dissolved in different solvents depends on concentration but there is not one consistent relationship (Figure 12.1.5). Instead, three separate relationships exist each for basic, neutral, and acid solvents, respectively. This shows that solvent acid-base properties have a very strong influence on viscosity. Figure 12.1.6 shows two different behaviors for unassociated and associated block copolymers. The first type has a linear relationship between viscosity and concentration whereas with the second there is a rapid increase in viscosity as concentration increases. This is the best described as a power law function.8 Two polymers in combination have different reactions when dissolved in different solvents (Figure 12.1.7). In MEK, intrinsic viscosity increases as polymer concentration increases. In toluene, intrinsic viscosity decreases as polymer concentration increases.9 The polymer-solvent interaction term for MEK is very small (0.13) indicating a stable compatible system. The interaction term for toluene is much larger (0.58) which indicates a decreased compatibility of polymers in toluene and lowers viscosity of the mixture. Figure 12.1.8 explicitly shows that the behavior of
688
George Wypych
106
104 4 wt% CO
2
no CO2 Viscosity, Pa s
Viscosity, Poise
20% solvent no solvent
105
104
1
10
100 -1
Shear rate, s
1000
1000
100 10
100
1000 -1
Shear rate, s
Figure 12.1.9. Apparent melt viscosity of original PET Figure 12.1.10. Viscosity behavior of PS with and and PET containing 20% 1-methyl naphthalene vs. without CO2. [Data from M Lee, C Tzoganikis, C B Park, Antec ‘99, SPE, New York, 1999, p 2806.] shear rate. [Adapted, by permission, from S Tate, S Chiba, K Tani, Polymer, 37, No.19, 4421-4 (1996).]
individual polymers does not necessarily have a bearing on the viscosity of their solutions. Both poly(ethylene oxide) and poly(propylene oxide) are not affected by solution pH but, when used in combination, they become sensitive to solution pH. A rapid increase of viscosity at a lower pH is ascribed to intermolecular complex formation. This behavior can be used for thickening of formulations.8 Figures 12.1.9 and 12.1.10 show one potential application in which a small quantity of solvents can be used to lower melt viscosity during polymer processing. Figure 12.1.9 shows that not only can melt viscosity be reduced but also that the viscosity is almost independent of shear rate.10 In environmentally friendly process supercritical fluids can be used to reduce melt viscosity. The above data illustrate that the real behavior of solutions is much more complex than it is intuitively predicted based on simple models and relationships. The proper selection of solvent can be used to tailor the properties of formulation to the processing and application needs. Solution viscosity can be either increased or decreased to meet process technology requirements or to give the desired material properties.
REFERENCES 1 2 3 4 5 6 7 8 9
A Ya Malkin, Rheology Fundamentals, ChemTec Publishing, Toronto, 1994. Ch W Macosko, Rheology. Principles, Measurements, and Applications, VCH Publishers, New York, 1994. R I Tanner, K. Walters, Rheology: an Historical Perspective, Elsevier, Amsterdam, 1998. C R Schultheisz, G B McKenna, Antec ‘99, SPE, New York, 1999, p 1125. F Motamedi, M Isomaki, M S Trimmer, Antec ‘98, SPE, Atlanta, 1998, p. 1772. G D J Phillies, Macromolecules, 28, No.24, 8198-208 (1995). M Ponitsch, T Hollfelder, J Springer, Polym. Bull., 40, No.2-3, 345-52 (1998). I C De Witte, B G Bogdanov, E J Goethals, Macromol. Symp., 118, 237-46 (1997). H Raval, S Devi, Angew. Makromol. Chem., 227, 27-34 (1995).
12.1 Rheological properties, aggregation, permeability
10 11 12
689
S Tate, S Chiba, K Tani, Polymer, 37, No.19, 4421-4 (1996). M L Abel, M M Chehimi, Synthetic Metals, 66, No.3, 225-33 (1994). M Lee, C Tzoganikis, C B Park, Antec ‘99, SPE, New York, 1999, p 2806.
12.1.2 AGGREGATION The development of materials with an engineered morphological structure, such as selective membranes and nanostructures, employs principles of aggregation in these interesting technical solutions. Here, we consider some basic principles of aggregation, methods of studies, and outcomes. The discipline is relatively new therefore for the most part, only exploratory findings are available now. The theoretical understanding is still to be developed and this development is essential for the control of industrial processes and development of new materials. Methods of study and data interpretation still require further work and refinement. Several experimental techniques are used, including: microscopy (TEM, SEM);1,2 dynamic light scattering3-6 using laser sources, goniometers, and digital correlators; spectroscopic methods (UV, CD, fluorescence);7,8 fractionation; solubility and viscosity measurements;9 and acid-base interaction.10 Dynamic light scattering is the most popular method. Results are usually expressed by the radius of gyration, Rg, the second viral coefficient, A2, the association number, p, and the number of arms, f, for starlike micelles. Zimm’s plot and equation permits Rg and A2 to be estimated: 2 1 16 2 Rg KC = π 2 sin 2 θ + ⋅ ⋅ ⋅ + 2A2 C + ⋅ ⋅ ⋅ 1 + 3 Rθ Mw λ0
[12.1.11]
where: K C Rθ Mw λ0 θ
optical constant polymer concentration Rayleigh ratio for the solution weight average molecular weight wavelength of light scattering angle.
The following equations are used to calculate p and f: M agg
p=
M1
Rg Rgarm
; M agg = KI q→0 / C
= f (1−v )/ 2
where: Magg M1 I Rgarm v
mass of aggregates mass of free copolymers scattered intensity radius of gyration of linear polymer excluded volume exponent.
[12.1.12]
[12.1.13]
690
George Wypych
Figure 12.1.11. Morphological features: starlike and crew-cut. The bottom drawing illustrates polymer brush. [Adapted, by permission from G Liu, Macromol. Symp., 113, 233 (1997).]
Figure 12.1.12. Morphological features of PS-PAA copolymer having different ratios of PS and PAA block lengths: A - 8.3, B- 12, C - 20.5, and D - 50. [Adapted, by permission, from L Zhang, A Eisenberg, Macromol. Symp., 113, 221-32 (1997).]
Figure 12.1.13. PS-PAA aggregates of different morphologies depending on its concentration in DMF: A - 2, B - 2.6, C - 3, D - 4 wt%. [Adapted, by permission, from L Zhang, A Eisenberg, Macromol. Symp., 113, 221-32 (1997).]
Figure 12.1.14. TEM micrograph of nanospheres. [Adapted, by permission, from G Liu, Macromol. Symp., 113, 233 (1997).]
Three major morphological features are under investigation: starlike, crew-cut, and polymer brushes (Figure 12.1.11). Morphological features have been given nick-names characterizing the observed shapes, such as “animals” or “flowers” to distinguish between various observed images.6 Figure 12.1.12 shows four morphologies of aggregates formed by polystyrene, PS,-poly(acrylic acid), PAA, diblock copolymers.1 The morphology produced was a direct result of the ratio between lengths of blocks of PS and PAA (see Figure 12.1.12). When this ratio is low (8.3), spherical micelles are formed. With a slightly higher ratio (12), rod-like micelles result which have narrow distribution of diameter but variable length. Increasing ratio even further (20.5) causes vesicular aggregates to form. With the highest ratio (50), large spherical micelles are formed. One of the reasons for the differences in these formations is that the surface tension between the core and the solvent varies widely. In order to decrease interfacial tension between the core and the solvent, the aggregate increases the core size.1 Polymer concentration in solution also affects aggregate for-
12.1 Rheological properties, aggregation, permeability
Figure 12.1.15. TEM micrograph of assemblies of cylindrical aggregates. [Adapted, by permission, from G Liu, Macromol. Symp., 113, 233 (1997).]
691
Figure 12.1.16. TEM micrograph of nanofibers. [Adapted, by permission, from G Liu, Macromol. Symp., 113, 233 (1997).]
350
PVC/BrBz PVC/DOA
300
Rg, nm
250 200 150 100 50 20
30
40
50
60
70
o
Temperature, C Figure 12.1.17. TEM micrograph of knots and strands formed in carbohydrate amphiphile. [Adapted, by permission, from U Beginn, S Keinath, M Moller, Macromol. Chem. Phys., 199, No.11, 2379-84 (1998).]
Figure 12.1.18. Radius of gyration vs. temperature for PVC solutions in bromobenzene, BrBz, and dioxane, DOA. [Adapted, by permission, from Hong Po-Da, Chen Jean-Hong Chen, Polymer, 40, 4077-4085, (1999).]
mation as Figure 12.1.13 shows. As polymer concentration increases, spherical micelles are replaced by a mixture of rod-like structures and vesicles. With a further increase in concentration, the rod-like shapes disappear and only vesicles remain. Figures 12.1.14-12.1.16 show a variety of shapes which have been observed. These include nanospheres, assemblies of cylindrical aggregates, and nanofibers.11 Nanospheres were obtained by the gradual removal of solvent by dialysis, fibers were produced by a series of processes involving dissolution, crosslinking, and annealing. Figure 12.1.17 sheds some light on the mechanism of aggregate formation. Two elements are clearly visible from micrographs: knots and strands. Based on studies of carbohydrate amphiphiles, it is concluded that knots are formed early in the process by spinodal decomposition. Formation of
692
George Wypych
100
Association number, p
1000
Specific viscosity
100
10
1
0.1
1
10 -1
Concentration, g l
Figure 12.1.19. Specific viscosity of PVC in bromobenzene solution vs. concentration. [Adapted, by permission, from Hong Po-Da, Chen Jean-Hong Chen, Polymer, 40, 4077-4085, (1999).]
1
10
1 0.001
0.01
0.1
1 -3
Concentration, mg cm
Figure 12.1.20. Diblock copolymer association number vs. its concentration. [Adapted, by permission, from D Lairez, M Adam, J-P Carton, E Raspaud, Macromolecules, 30, No.22, 6798-809 (1997).]
I560/I524
strands between knots occurs much later.2 A similar course of events occurs with the molecular aggregation of PVC solutions.5 Figure 12.1.18 gives more details on the mechanism of aggregate formation.5 PVC dissolved in bromobenzene, BrBz, 0.1 undergoes a coil to globule transition which does not occur in dioxane, DOA, solution. Bromobenzene has a larger molar volume than dioxane and the polymer chain must readjust to the solvent molar volume in order to interact. Figure 12.1.19 0.01 0.01 0.1 1 10 100 shows that two regimes of aggregation are involved which are divided by a certain Concentration, g dl-1 critical value. Below the critical value Figure 12.1.21. Fluorescence intensity ratio vs. concenchains are far apart and do not interact due tration of polyimide in chloroform. [Data from H Luo, L Dong, H Tang, F Teng, Z Feng, Macromol. Chem. to contact self-avoidance. At the critical Phys., 200, No.3, 629-34 (1999).] point, knots, similar to shown in Figure 12.1.17, begin to form and the resultant aggregation increases viscosity. As the polymer concentration increases, the association number, p, also increases, but then levels off (Figure 12.1.20). Fluorescence studies (Figure 12.1.21) show that there are two characteristic points relative to concentration. Below the first point, at 0.13 g/dl, in very dilute solution, molecules are highly expanded and distant from one another and fluorescence does not change. Between the two points, individual chain coils begin to sense each other and become affected
12.1 Rheological properties, aggregation, permeability
693
by the presence of neighbors forming intermolecular associations. Above the second point, coils begin to overlap leading to dense packing.8 These data are essentially similar to those presented in Figure 12.1.20 but cover a broader concentration range. The above studies show that there are many means of regulating aggregate size and shape which is likely to become an essential method of modifying materials by morphological engineering.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11
L Zhang, A Eisenberg, Macromol. Symp., 113, 221-32 (1997). U Beginn, S Keinath, M Moller, Macromol. Chem. Phys., 199, No.11, 2379-84 (1998). P A Cirkel, T Okada, S Kinugasa, Macromolecules, 32, No.2, 531-3 (1999). K Chakrabarty, R A Weiss, A Sehgal, T A P Seery, Macromolecules, 31, No.21, 7390-7 (1998). Hong Po-Da, Chen Jean-Hong Chen, Polymer, 40, 4077-4085, (1999). D Lairez, M Adam, J-P Carton, E Raspaud, Macromolecules, 30, No.22, 6798-809 (1997). R Fiesel, C E Halkyard, M E Rampey, L Kloppenburg, S L Studer-Martinez, U Scherf, U H F Bunz, Macromol. Rapid Commun., 20, No.3, 107-11 (1999). H Luo, L Dong, H Tang, F Teng, Z Feng, Macromol. Chem. Phys., 200, No.3, 629-34 (1999). A Leiva, L Gargallo, D Radic, J. Macromol. Sci. B, 37, No.1, 45-57 (1998). S Bistac, J Schultz, Macromol. Chem. Phys., 198, No. 2, 531-5 (1997). G Liu, Macromol. Symp., 113, 233 (1997).
12.1.3 PERMEABILITY The phenomenon of solvent transport through solid barriers has three aspects which discussed under the heading of permeability. These are the permeation of solvent through materials (films, containers, etc.); the use of pervaporation membranes to separate organic solvents from water or water from solvents; the manufacture of permeate selective membranes. The permeability of different polymers and plastics to various solvents can be found in an extensive, specialized database published as a book and as a CD-ROM.1 The intrinsic properties of polymers can be modified in several ways to increase their resistance and reduce their permeability to solvents. The development of plastic gas tanks was a major driving force behind these developments and now various plastic containers are manufactured using similar processes. Fluorination of plastics, usually polypropylene or polyethylene, is by far the most common modification. Containers are typically manufactured by a blow molding process where they are protected by a fluorination process applied on line or, more frequently, off line. The first patent2 for this process was issued in 1975 and numerous other patents, some of them issued quite recently have made further improvements to the process.3 The latest processes use a reactive gas containing 0.1 to 1% fluorine to treat parison within the mold after it was expanded.3 Containers treated by this process are barrier to polar liquids, hydrocarbon fuels, and carbon fuels containing polar liquids such as alcohols, ethers, amines, carboxylic acids, ketones, etc. Superior performance is achieved when a hydrogen purge precedes the exposure of the container to oxygen. In another recent development,4 containers are being produced from a blend of polyethylene and poly(vinylidene fluoride) with aluminum stearate as compatibilizer. This process eliminates the use of the toxic gas fluorine and by-products which reduces environmental pollution and disposal problems. It is argued that fluorinated containers may not perform effectively when the protective layer becomes damaged due to stress cracking.
694
George Wypych
Such a blend is intended as a replacement for previous blends of polyamide and polyethylene used in gas tanks and other containers. These older blends were exposed to a mixture of hydrocarbons but now oxygenated solvents (e.g., methanol) have been added which cause an unacceptable reduction of barrier properties. A replacement technology is needed. Improved barrier properties in this design4 comes from increased crystallinity of the material brought about by controlling laminar thickness of polyethylene crystals and processing material under conditions which favor the crystallization of poly(vinylidene fluoride). Another approach involves the use of difunctional telechelic polyolefins with ester, hydroxyl, and amine terminal groups.5 Telechelic polyolefins can be used to make polyesters, polyamides, and polyurethanes with low permeability to solvents and gases. Separation processes such as ultrafiltration and microfiltration use porous membranes which allow the passage of molecules smaller than the membrane pore size. Ultrafiltration membranes have pore sizes from 0.001 to 0.1 µm while microfiltration membranes have pore sizes in the range of 0.02 to 10 µm. The production of these membranes is almost exclusively based on non-solvent inversion method which has two essential steps: the polymer is dissolved in a solvent, cast to form a film then the film is exposed to a non-solvent. Two factors determine the quality of the membrane: pore size and selectivity. Selectivity is determined by how narrow the distribution of pore size is.6,7 In order to obtain membranes with good selectivity, one must control the non-solvent inversion process so that it inverts slowly. If it occurs too fast, it causes the formation of pores of different sizes which will be non-uniformly distributed. This can be prevented either by an introduction of a large number of nuclei, which are uniformly distributed in the polymer membrane or by the use of a solvent combination which regulates the rate of solvent replacement. Typical solvents used in membrane production include: N-methylpyrrolidinone, N,N-dimethylacetamide, N,N-dimethylformamide, dimethylsulfoxide, tetrahydrofuran, dioxane, dichloromethane, methyl acetate, ethyl acetate, and chloroform. They are used alone or in mixtures.6 These are used most frequently as non-solvents: methanol, ethanol, 1-propanol, 2-propanol, 1-butanol, 2-butanol, and t-butanol. Polymers involved include: polysulfone, polyethersulfone, polyamide, polyimide, polyetherimide, polyolefins, polycarbonate, polyphenyleneoxide, poly(vinylidene fluoride), polyacrylonitrile, and cellulose and its derivatives. A second method of membrane preparation is based on the thermally-induced phase separation process.8 The goal of this process is to produce a membrane which has an ultrathin separation layer (to improve permeation flux) and uniform pores.9 In this process, polymer is usually dissolved in a mixture of solvents which allow the mixture to be processed either by spinning Figure 12.1.23. PEI fiber from or by coating. The desired Figure 12.1.22. PEI fiber from DMF. [Adapted, by permission membrane morphology is obNMP. [Adapted, by permission from D Wang, K Li, W K Teo, from D Wang, K Li, W K Teo, J. Appl. Polym. Sci., 71, No.11, tained through cooling to inJ. Appl. Polym. Sci., 71, No.11, duce phase separation of 1789-96 (1999).] 1789-96 (1999).]
695
40
1.6
35
1.4
30 25 20 15
propionic acid 2-propanol water
10 5 15 20 25 30 35 40 45 50 55 o
Temperature, C Figure 12.1.24. Precipitation value for PEI/NMP system with different non-solvents vs. temperature. [Data from D Wang, Li K, W K Teo, J. Appl. Polym. Sci., 71, No.11, 1789-96 (1999).]
Skin layer thickness,µm
Precipitation value, g
12.1 Rheological properties, aggregation, permeability
1.2 1 0.8 0.6 0.4 0.2
0
2
4
6
8
10
Evaporation time, min Figure 12.1.25. Skin thickness vs. evaporation time of formation of asymmetric membrane from polysulfone. [Data from A Yamasaki, R K Tyagi, A E Fouda, T Matsura, K Jonasson, J. Appl. Polym. Sci., 71, No.9, 1367-74 (1999).]
polymer/solvent/non-solvent system. One practices in use is the addition of non-solvents to the casting solution.10 The mixture should be designed such that homogeneous casting is still possible but the thermodynamic condition approaches phase separation. Because of cooling and/or evaporation of one of the solvents, demixing occurs before material enters coagulation bath.10 Figures 12.1.22 and 12.1.23 explain technical principles behind formation of efficient and selective membrane. Figure 12.1.22 shows a micrograph of hollow PEI fiber produced from N-methyl-2-pyrrolidone, NMP, which has thin surface layer and uniform pores and Figure 12.1.23 shows the same fiber obtained from a solution in dimethylformamide, DMF, which has a thick surface layer and less uniform pores.9 The effect depends on the interaction of polar and non-polar components. The compatibility of components was estimated based on their Hansen’s solubility parameter difference. The compatibility increases as the solubility parameter difference decreases.9 Adjusting temperature is another method of control because the Hansen’s solubility parameter decreases as the temperature increases. A procedure was developed to determine precipitation values by titration with non-solvent to a cloud point.9 Use of this procedure aids in selecting a suitable non-solvent for a given polymer/solvent system. Figure 12.1.24 shows the results from this method.9 Successful in membrane production by either non-solvent inversion or thermally-induced phase separation requires careful analysis of the compatibilities between polymer and solvent, polymer and non-solvent, and solvent and non-solvent. Also the processing regime, which includes temperature control, removal of volatile components, uniformity of solvent replacement must be carefully controlled. Efforts must be made to select solvents of low toxicity and to minimize solvent consumption. An older method of producing polyetherimide membranes involved the use of a mixture of two solvents: dichloromethane (very volatile) and 1,1,2-trichloroethane together
George Wypych
1000
35 30 25
-1 -2
20 15 10
-1
Separation factor
A/l B/l
-1
-2
A/l, mol s m Pa or B/l, mol s m Pa
-2
696
800
600
400
5 0
0
50
100
150
200
200
250
0
Evaporation time, min
0.8
1
100 Hexane in permeate, wt%
12
-1
10 -2
0.6
Figure 12.1.27. Separation factor vs. solubility of solvents in water. [Adapted, by permission, from M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).]
14
Flux, g m h
0.4
Solubility of solvent in water, wt%
Figure 12.1.26. Pervaporation transport parameters: A/l (liquid transport) and B/l (vapor transport) vs. evaporation time during membrane preparation from aromatic polyamide. [Data from A Yamasaki, R K Tyagi, A Fouda, T Matsuura, J. Appl. Polym. Sci., 57, No.12, 1473-81 (1995).]
8 6 4 2 0
0.2
0
0.02
0.04
0.06
0.08
0.1
Triclene in feed, wt% Figure 12.1.28. Effect of triclene concentration in feed solution on its pervaporation flux through acrylic membrane. [Data from M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).]
80 60 40 20 0
0
20
40
60
80
100
Hexane in feed, wt% Figure 12.1.29. Effect of hexane concentration in feed solution on its concentration in permeate through acrylic membrane. [Data from M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).]
with a mixture of non-solvents: xylene and acetic acid. In addition, large amounts of acetone were used in the coagulation bath. A more recent process avoided environmental problems by using tetrahydrofuran and γ-butyrolactone in a process in which water only is used in the coagulating bath (membrane quality remained the same).10
12.1 Rheological properties, aggregation, permeability
697
Thermally-induced phase separation has been applied to the production of polysilane foams.11 Variation of polymer concentration, solvent type, and cooling rate have been used to refine the macrostructure. Both these membrane production methods rely on the formation of aggregates of controlled size and shape as discussed in the previous section. Figure 12.1.25 shows the effect of of solvent evaporation time on the skin layer thickness. It is only in the beginning of the process that skin thickness does not increase. Thereafter there is rapid skin thickness growth. Polymer concentration is the other important determinant of skin thickness. Figure 12.1.26 shows that during very short evaporation times (5 min), even though skin is thin the pores diameter is so small that the pervaporation parameters A/l and B/l, which characterize liquid and vapor transport, respectively, do not increase. Longer evaporation times bring about a gradual decrease in transport properties.12,13 Figure 12.1.27 shows the relationship between the separation factor and the solubility of solvents in water. The separation of solvent by a pervaporation membrane occurs less efficiently as solvent solubility increases.The more concentrated the solution of solvent, the faster is the separation (Figure 12.1.28). Separation of hexane from a mixture with heptane is similar (Figure 12.1.29). The acrylic membrane shows good selectivity. These examples demonstrate the usefulness of pervaporation membranes in solvent recovery processes.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Permeability and Other Film Properties, Plastics Design Library, Norwich, 1996. US Patent 3,862,284, 1975. J P Hobbs, J F DeiTos, M Anand, US Patent 5,770,135, Air Products and Chemicals, Inc., 1998 and US Patent 5,244,615, 1992. R T Robichaud, US Patent 5,702,786, Greif Bros. Corporation, 1997. P O Nubel, H B Yokelson, US Patent 5,731,383, Amoco Corporation, 1998 and US Patent 5,589,548, 1996. J M Hong, S R Ha, H C Park, Y S Kang, K H Ahn, US Patent 5,708,040, Korea Institute of Science and Technology, 1998. K-H Lee, J-G Jegal, Y-I Park, US Patent 5,868,975, Korea Institute of Science and Technology, 1999. J M Radovich, M Rothberg, G Washington, US Patent 5,645,778, Althin Medical, Inc., 1997. D Wang, Li K, W K Teo, J. Appl. Polym. Sci., 71, No.11, 1789-96 (1999). K-V Peinemann, J F Maggioni, S P Nunes, Polymer, 39, No.15, 3411-6 (1998). L L Whinnery, W R Even, J V Beach, D A Loy, J. Polym. Sci.: Polym. Chem. Ed., 34, No.8, 1623-7 (1996). A Yamasaki, R K Tyagi, A Fouda, T Matsuura, J. Appl. Polym. Sci., 57, No.12, 1473-81 (1995). A Yamasaki, R K Tyagi, A E Fouda, T Matsura, K Jonasson, J. Appl. Polym. Sci., 71, No.9, 1367-74 (1999). M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).
12.1.4 MOLECULAR STRUCTURE AND CRYSTALLINITY The gelation of polymer-solvent systems was initially thought of as a process which occurs only in poor solvents which promote chain-chain aggregation. Further studies have revealed that many polymers also form gels in good solvents. This has prompted research which attempts to understand the mechanisms of gelation. These efforts have contributed the current understanding of the association between molecules of solvents and polymers.1 Studies on isotactic, iPS, and syndiotactic, sPS, polystyrenes2-6 have confirmed that, although both polymers have the same monomeric units, they are significantly different in terms of their solubility, gelation, and crystallization. In systems where sPS has been dissolved in benzene and carbon tetrachloride, gelation is accomplished in a few minutes
698
George Wypych
whereas in chloroform it takes dichloromethane tens of hours. no solvent 15 Adding a small amount of benzene to the chlo10 roform accelerates the gelation process. There is 5 also evidence from IR which shows that the 0 orientation of the 0 200 400 600 ring plane is perStrain, % pendicular to the Figure 12.1.30. Molecular Figure 12.1.31. Stress-strain behavior at 130oC of chain axis. This model of polystyrene and glassy sPS with and without immersion in that benzene rings alignment. dichloromethane. [Data from C Daniel, L Guadagno, suggests there is a strong [Adapted by permission, from V Vittoria, Macromol. Symp., 114, 217 (1997).] J M Guenet, Macromol. Symp., interaction be114, 97 (1997).] tween benzene and sPS molecules. The other interesting observation came 1715 from studies on decalin sPS and iPS systems. iPS forms a transparent gel and then becomes 1710 turbid, gradually forming trigonal crystallites of iPS. sPS does not convert to a gel but the 1705 fine crystalline precipitate particles instead. 1700 This shows that the molecular arrangement cyclohexanone depends on both the solvent type and on the N-methyl-2-pyrrolidone 1695 molecular structure of the polymer.2 Figure 12.1.30 shows how benzene mol1690 ecules align themselves parallel to the phenyl rings of polystyrene and how they are housed 1685 0 0.2 0.4 0.6 0.8 1 within the helical form of the polymer structure which is stabilized by the presence of solSolvent molar fraction vent.1 Work on poly(ethylene oxide) gels7 indiFigure 12.1.32. Carbonyl frequency vs. molar fraction of solvent. [Adapted, by permission, from cates that the presence of solvents such as J M Gomez-Elvira, P Tiemblo, G Martinez, J Milan, chloroform and carbon disulfide contributes to Macromol. Symp., 114, 151 (1997).] the formation of a uniform helical conformation. The gelation behavior and the gel structure depend on the solvent type which, in turn, is determined by solvent-polymer interaction. In a good solvent, polythiophene molecules exist in coiled conformation. In a poor solvent, the molecules form aggregates through the short substituents.8 Polyvinylchloride, PVC, which has a low crystallinity, gives strong gels. Neutron diffraction and scattering studies show that these strong gels result from the formation of Carbonyl frequency shift, cm
-1
Stress, MPa
20
12.1 Rheological properties, aggregation, permeability
699
Figure 12.1.33. Crystal structure of poly(ethylene oxide) molecular complex with: (a) p-dichlorobenzene, (b) resorcinol, (c) p-nitrophenol. [Adapted, by permission, from M Dosiere, Macromol. Symp., 114, 51 (1997).]
PVC-solvent complexes.9 This indicates that the presence of solvent may affect the mechanical properties of such a system. Figure 12.1.31 shows that the presence of dichloromethane in sPS changes the mechanical characteristic of the material. A solvent-free polymer has a high elongation and yield value. An oriented polymer containing dichloromethane has lower elongation, no yield value, and approximately four times greater tensile strength.10 The interaction of polymer-solvent affects rheological properties. Studies on a divinyl ether-maleic anhydride copolymer show that molecular structure of the copolymer can be altered by the solvent selected for synthesis.11 Studies Figure 12.1.34. Optical micrograph stability of PVC is afof PEO-resorcinol complex. have shown that the thermal and UV 12-14 fected by the presence of solvents. [Adapted, by permission, from M Dosiere, Macromol. Symp., 114, Figure 12.1.32 shows that solvent type and its molar 51 (1997).] fraction affect the value of the carbonyl frequency shift. This frequency shift occurs at a very low solvent concentration, The slope angles for each solvent are noticeably different.15 Similar frequency shifts were reported for various solvents in polyetheretherketone, PEEK, solutions.16 These associations are precursors of further ordering by crystallization. Figure 12.1.33 shows different crystalline structures of poly(ethylene oxide) with various solvents.17 Depending on the solvent type, a unit cell contains a variable number of monomers and molecules of solvent stacked along the crystallographic axis. In the case of dichlorobenzene, the molecular complex is orthorhombic with 10 monomers and 3 solvent molecules. Figure 12.1.34 shows optical micrograph of spherulite formed from unit cells having 8 monomers and 4 molecules of solvent. The unit cell of the triclinic crystal, formed with nitrophenol, is composed of 6 monomers and 4 molecules of solvent. This underlines the importance of solvent selection in achieving the desired structure in the formed material. The technique is also advantageous in polymer synthesis, in improving polymer stability, etc.
700
George Wypych
Temperature is an essential parameter in the crystallization process.18 Rapid cooling of a polycarbonate, PC, solution in benzene resulted in extremely high crystallinity (46.4%) as compared to the typical PC crystallinity of about 30%. Polymer-solvent interaction combined with the application of an external force leads to the surface crazing of materials. The process is based on similar principles as discussed in this section formation of fibrilar crystalline structures. Although research on molecular structure and crystallization is yet to formulate a theoretical background which might predict the effect of different solvents on the fine structure of different polymers, studies have uncovered numerous issues which cause concern but which also point to new applications. A major concern is the effect of solvents on craze formation, and on thermal and UV degradation. Potential applications include engineering of polymer morphology by synthesis in the presence of selected solvent under a controlled thermal regime, polymer reinforcement by interaction with smaller molecules, better retention of additives, and modification of surface properties to change adhesion or to improve surface uniformity, etc.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
J M Guenet, Macromol. Symp., 114, 97 (1997). M Kobayashi, Macromol. Symp., 114, 1-12 (1997). T Nakaoki, M Kobayashi, J. Mol. Struct., 242, 315 (1991). M Kobayashi, T Nakoaki, Macromolecules, 23, 78 (1990). M Kobayashi, T. Kozasa, Appl. Spectrosc., 47, 1417 (1993). M. Kobayashi, T Yoshika, M Imai, Y Itoh, Macromolecules, 28, 7376 (1995). M Kobayashi, K Kitagawa, Macromol. Symp., 114, 291 (1997). P V Shibaev, K Schaumburg, T Bjornholm, K Norgaard, Synthetic Metals, 97, No.2, 97-104 (1998). H Reinecke, J M Guenet, C. Mijangos, Macromol. Symp., 114, 309 (1997). C Daniel, L Guadagno, V Vittoria, Macromol. Symp., 114, 217 (1997). M Y Gorshkova, T L Lebedeva, L L Stotskaya, I Y Slonim, Polym. Sci. Ser. A, 38, No.10, 1094-6 (1996). G Wypych, Handbook of Material Weathering, ChemTec Publishing, Toronto, 1995. G Wypych, Poly(vinyl chloride) degradation, Elsevier, Amsterdam, 1985. G Wypych, Poly(vinyl chloride) stabilization, Elsevier, Amsterdam, 1986. J M Gomez-Elvira, P Tiemblo, G Martinez, J Milan, Macromol. Symp., 114, 151 (1997). B H Stuart, D R Williams, Polymer, 36, No.22, 4209-13 (1995). M Dosiere, Macromol. Symp., 114, 51 (1997). Gending Ji, Fengting Li, Wei Zhu, Qingping Dai, Gi Xue, Xinhong Gu, J. Macromol. Sci. A, A34, No.2, 369-76 (1997). I V Bykova, E A Sinevich, S N Chvalun, N F Bakeev, Polym. Sci. Ser. A, 39, No.1, 105-12 (1997).
12.1.5 OTHER PROPERTIES AFFECTED BY SOLVENTS Many other properties of solutes and solutions are affected by solvents. Here, we will discuss material stability and stabilization, some aspects of reactivity (more information on this subject appears in Chapter 13), physical properties, some aspects of electrical and electrochemical properties (more information on this subject appears in Chapter 11), surface properties, and polarity and donor properties of solvents. Degradative processes have long been known to be promoted by the products of solvent degradation. Tetrahydrofuran is oxidized to form a peroxide which then dissociates to form two radicals initiating a chain of photo-oxidation reactions. Figure 12.1.35 shows the kinetics of hydroperoxide formation.1 Similar observations, but in polymer system, were made in xylene by direct determination of the radicals formed using ESR.2 An increased concentration in trace quantities of xylene contributed to the formation of n-octane radicals
0.5 0.4 0.3 0.2 0.1 0
0
0.5
1
1.5
2
2.5
Time, h Figure 12.1.35. Peroxide formation from tetrahydrofuran during irradiation at 254 nm. [Data from J F Rabek, T A Skowronski, B Ranby, Polymer, 21, 226 (1980).]
3
Relative signal intensity, arbitrary units
Peroxide concentration, arbitrary units
12.1 Rheological properties, aggregation, permeability
701
5 4 3 2 1 0 -5 10
0.0001 0.001 0.01 Xylene concentration, mol l-1
0.1
Figure 12.1.36. Spectral intensity of n-octane radical formed during irradiation of PE at -196oC for 30 min vs. concentration of p-xylene. [Adapted, by permission, from H Kubota, M Kimura, Polym. Deg. Stab., 38, 1 (1992).]
by abstracting hydrogen from polyethylene chain in an α-position to the double bond (Figure 12.1.36). 3-hydroperoxyhexane cleaves with formation of carboxylic acid and hydrocarbon radical (ethylene or propylene).3 These known examples show that the presence of even traces of solvents may change the chemistry and the rate of photo-oxidative processes because of formation of radicals. In another aspect of photo-oxidative processes, solvents influence the reactivity of small molecules and chain segments by facilitating the mobility of molecules and changing the absorption of light, the wavelength of emitted fluorescent radiation, and the lifetime of radicals. Work on anthraquinone derivatives, which are common photosensitizers, has shown that when photosensitizer is dissolved in isopropanol (hydrogen-donating solvent) the half-life of radicals is increased by a factor of seven compared to acetonitrile (a non-hydrogen-donating solvent).4 This shows that the ability of a photosensitizer to act in this manner depends on the presence of a hydrogen donor (frequently the solvent). In studies on another group of photosensitizers − 1,2-diketones, the solvent cyclohexane increased the absorption wavelength of the sensitizer in the UV range by more than 10 nm compared with solutions in ethanol and chloroform.5 Such a change in absorption may benefit some systems because the energy of absorbed radiation will become lower than the energy required to disrupt existing chemical bonds. But it may also increase the potential for degradation by shifting radiation wavelength to the range absorbed by a particular material. Presence of water may have a plasticizing action which increases the mobility of chains and their potential for interactions and reactions.6 Studies on photoresists, based on methacryloyethyl- phenylglyoxylate, show that, in aprotic solvents, the main reaction mechanism is a Norrish type II photolysis leading to chain scission.7 In aprotic solvents, the polymer is photoreduced and crosslinks are formed.
702
George Wypych 105
80 70
10
4
60 Yield, %
log η0
1000 100
50 40 30 20
10
10 1 4 10
5
10 log M
6
10
w
Figure 12.1.37. Effect of viscosity on molecular weight of polyethylene wax obtained by thermolysis in the presence of phenylether at 370oC. [Adapted, by permission, from L Guy, B Fixari, Polymer, 40, No.10, 2845-57 (1999).]
0
methanol
DMF
DMSO
Solvent Figure 12.1.38. Yield of condensation products depending on solvent. [Adapted, by permission, from J Jeczalik, J. Polym. Sci.: Polym. Chem. Ed., 34, No.6, 1083-5 (1996).]
Polarity of the solvent determines quantum yields in polyimides.8 The most efficient photocleavage was in medium-polar solvents. The selection of solvent may change the chemical mechanism of degradation and the associated products of such reactions. Singlet oxygen is known to affect material stability by its ability to react directly with hydrocarbon chains to form peroxides. The photosensitizers discussed above are capable of generating singlet oxygen. Solvents, in addition to their ability to promote photosensitizer action, affect the lifetime of singlet oxygen (the time it has to react with molecules and form peroxides). Singlet oxygen has very short lifetime in water (2 µs) but much longer times in various solvents (e.g., 24 µs in benzene, 200 µs in carbon disulfide, and 700 µs in carbon tetrachloride).9 Solvents also increase the oxygen diffusion coefficient in a polymer. It was calculated10 that in solid polystyrene only less than 2% of singlet oxygen is quenched compared with more than 50% in solution. Also, the quantum yield of singlet oxygen was only 0.56 in polystyrene and 0.83 in its benzene solution.10 The photostabilizer must be durable. It was found that salicylic stabilizers are efficiently degraded by singlet oxygen in polar alkaline media but in a less polar, non-alkaline solvents these stabilizers are durable.11 In hydrogen-bonding solvents, the absorption spectrum of UV absorbers is changed.12 Thermal decomposition yield and composition of polystyrene wastes is affected by addition of solvents.13 Solvents play two roles: First, they stabilize radicals and, second, by lowering solution viscosity they contribute to homogeneity, increase the mobility of components and increase reactivity. 85% benzyl and phenoxy radicals were stabilized by hydrogens from tetralin.13 In polyethylene thermolysis, the use of solvent to reduce viscosity resulted in obtaining lower molecular weight, narrow molecular weight distribution and high crystallinity in resultant polyethylene wax.14 Figure 12.1.37 illustrates the relationship
12.1 Rheological properties, aggregation, permeability
100
Surface DMS, molar %
Glass transition temperature, K
165
160
155
150
145
703
0
5
10
15
20
25
MIBK concentration, % Figure 12.1.39. Glass transition temperature of perfluoro polyoxyalkylene oligomers vs. MIBK concentration. [Data from S Turri, M Scicchitano, G Gianotti, C Tonelli, Eur. Polym. J., 31, No.12, 1227-33 (1995).]
98 96 2% DMS 1% DMS
94 92 90
0
5
10
15
20
25
30
Toluene content, v% Figure 12.11.40. Surface DMS vs. toluene concentration for two bulk contents of DMS in polystyrene. [Data from Jiaxing Chen, J A Gardella, Macromolecules, 31, No.26, 9328-36 (1998).]
between molecular weight and viscosity of solutions of polyethylene in phenylether as a solvent.14 Figure 12.1.38 shows that polymerization yield depends on the solvent selected for the polymerization reaction. The polymerization yield increases as solvent polarity increases.15 Work on copolymerization of methyl methacrylate and N-vinylpyrrolidone shows that the solvent selected regulates the composition of copolymer.16 The smaller the polarity of the solvent and the lower the difference between the resonance factors of the two monomers the more readily they can copolymerize. Work on electropolymerization17 has shown an extreme case of solvent effect in the electrografting of polymer on metal. If the donocity of the monomer is too high compared with that of the solvent, no electrografting occurs. If the donocities are low, a high dielectric constant of solvent decreases grafting efficiency. This work, which may be very important in corrosion protection, illustrates that a variety of influences may affect the polymerization reaction. The addition of solvent to polymer has a plasticizing effect. The increase in free volume has a further influence on the glass transition temperature of the polymer. Figure 12.1.39 shows the effect of methyl-isobutyl-ketone, MIBK, at various concentrations on glass transition temperature.18 Several solvents were used in this study18 to determine if the additivity rule can be useful to predict the glass transition temperature of a polymer-solvent system. The results, as a rule, depart from the linear relationship between glass transition temperature and solvent concentration. Generally, the better the solvent is for a particular polymer the higher is the departure. Hydrogen bond formation as a result of the interaction of polymer with solvent was found to contribute to changes in the electric properties of polyaniline.19 Hydrogen bonding causes changes in conformal structure of polymer chains. This increases the electrical conductivity of polyaniline. Water is especially effective in causing such changes but other hy-
704
George Wypych
drogen bonding solvents also affect conductivity.19 This phenomenon is applied 99 in antistatic compounds which require a 98 certain concentration of water to perform their function. 97 Surface properties such as smooth96 ness and gloss are affected by the rheological properties of coatings and solvents play 95 an essential role in these formulations. Sur94 face composition may also be affected by 2% DMS 1% DMS solvent. In some technological processes, 93 the surface composition is modified by 92 small additions of polydimethylsiloxane, 0 2 4 6 8 10 PDMS, or its copolymers. The most well Cyclohexanone content, v% known application of such technology has Figure 12.1.41. Surface DMS vs. cyclohexanone been the protection of external surface of concentration for two bulk contents of DMS in the space shuttle against degradation using polystyrene. [Data from Jiaxing Chen, J A Gardella, PDMS as the durable polymer. Other appliMacromolecules, 31, No.26, 9328-36 (1998).] cations use PDMS to lower the coefficient of friction . In both cases, it is important that PDMS forms a very high concentrations on the surface (possibly 100%). Because of its low surface energy, it has the intrinsic tendency to migrate to the surface consequently, surface concentrations as high as 90% can be obtained without special effort. It is more difficult to further increase this surface concentration. Figure 12.1.40 shows the effect of a chloroform/toluene mixture on the surface segregation of PDMS. When the poor solvent for PDMS (toluene) is mixed with chloroform, the decrease in the mobility of the PDMS segments makes the cohesive migration of the polystyrene, PS, segments more efficient and the concentration of PDMS on the surface increases as the concentration of toluene increases. Increase in PDMS concentration further improves its concentration on the surface to almost 100%. This shows the influence of polymer-solvent interaction on surface segregation and demonstrates a method of increasing the additive polymer concentration on the surface. Solvents with higher boiling points also increase surface concentration of PDMS because they extend the time of the segregation process.20 Figure 12.1.41 shows the effect of addition of cyclohexanone to chloroform.20 This exemplifies yet another phenomenon which can help to increase the surface concentration of PDMS. Cyclohexanone is more polar solvent than chloroform (Hansen parameter, δp, is 3.1 for chloroform and 6.3 for cyclohexanone). The solvation of cyclohexanone molecules will selectively occur around PDMS segments which helps in the segregation of PDMS and in the cohesive migration of PS. But the figure shows that PDMS concentration initially increases then decreases with further additions of cyclohexanone. The explanation for this behavior is in the amount of cyclohexanone required to better solvate PDMS segments. When cyclohexanone is in excess, it also increases the solvation of PS and the effect which produces an increased surface segregation is gradually lost. The surface smoothness of materials is important in solvent welding. It was determined that a smooth surface on both mating components substantially increases the strength of the weld.21 This finding may not be surprising since weld strength depends on close surSurface DMS, molar %
100
12.1 Rheological properties, aggregation, permeability
705
face contact but it is interesting to note that the submicrostructure may have an influence on phenomena normally considered to be influenced by features which have a larger dimensional scale. Non-blistering primer is another example of a modification which tailors surface properties.22 The combination of resins and solvents used in this invention allows solvent to escape through the surface of the cured primer before the material undergoes transitions that occur at high temperatures. Additional information on this subject can be found in Chapter 7. A review paper23 examines the nucleophilic properties of solvents. It is based on accumulated data derived from calorimetric measurements, equilibrium constants, Gibbs free energy, nuclear magnetic resonance, and vibrational and electronic spectra. Parameters characterizing Lewis-donor properties are critically evaluated and tabulated for a large number of solvents. The explanation of the physical meaning of polarity and discussion of solvatochromic dyes as the empirical indicators of solvent polarity are discussed (see more on this subject in Chapter 10).23
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
J F Rabek, T A Skowronski, B Ranby, Polymer, 21, 226 (1980). H Kubota, M Kimura, Polym. Deg. Stab., 38, 1 (1992). G Teissedre, J F Pilichowski, J Lacoste, Polym. Degrad. Stability, 45, No.1, 145-53 (1994). M Shah, N S Allen, M Edge, S Navaratnam, F Catalina, J. Appl. Polym. Sci., 62, No.2, 319-40 (1996). P Hrdlovic, I Lukac, Polym. Degrad. Stability, 43, No.2, 195-201 (1994). M L Jackson, B J Love, S R Hebner, J. Mater. Sci. Materials in Electronics, 10, No.1, 71-9 (1999). Hu Shengkui, A Mejiritski, D C Neckers, Chem. Mater., 9, No.12, 3171-5 (1997). H Ohkita, A Tsuchida, M Yamamoto, J A Moore, D R Gamble, Macromol. Chem. Phys., 197, No.8, 2493-9 (1996). G Wypych, Handbook of Material Weathering, ChemTec Publishing, Toronto, 1995. R D Scurlock, D O Martire, P R Ogilby, V L Taylor, R L Clough, Macromolecules, 27, No.17, 4787-94 (1994). A T Soltermann, D de la Pena, S Nonell, F Amat-Guerri, N A Garcia, Polym. Degrad. Stability, 49, No.3, 371-8 (1995). K P Ghiggino, J. Macromol. Sci. A, 33, No.10, 1541-53 (1996). M Swistek, G Nguyen, D Nicole, J. Appl. Polym. Sci., 60, No.10, 1637-44 (1996). L Guy, B Fixari, Polymer, 40, No.10, 2845-57 (1999). J Jeczalik, J. Polym. Sci.: Polym. Chem. Ed., 34, No.6, 1083-5 (1996). W K Czerwinski, Macromolecules, 28, No.16, 5411-8 (1995). N Baute, C Calberg, P Dubois, C Jerome, R Jerome, L Martinot, M Mertens, P Teyssie, Macromol. Symp., 134, 157-66 (1998). S Turri, M Scicchitano, G Gianotti, C Tonelli, Eur. Polym. J., 31, No.12, 1227-33 (1995). E S Matveeva, Synthetic Metals, 79, No.2, 127-39 (1996). Jiaxing Chen, J A Gardella, Macromolecules, 31, No.26, 9328-36 (1998). F Beaume, N Brown, J. Adhesion, 47, No.4, 217-30 (1994). M T Keck, R J Lewarchik, J C Allman, US Patent 5,688,598, Morton International, Inc., 1996. Ch Reichardt, Chimia, 45(10), 322-4 (1991).
706
Ranieri Urbani and Attilio Cesàro
12.2 CHAIN CONFORMATIONS OF POLYSACCHARIDES IN DIFFERENT SOLVENTS Ranieri Urbani and Attilio Cesàro Department of Biochemistry, Biophysics and Macromolecular Chemistry, University of Trieste, Italy
12.2.1 INTRODUCTION Carbohydrate monomers and polymers are present in all living organisms and are widely used in industrial applications. In life forms they show and express very diverse biological functions: as structural, storage and energy materials, as specific molecules in the immunochemistry of blood, as important polymers of cell walls determining cell-cell recognition,1 antigenicity and viral infection, etc.. The variability of primary structure and conformation makes the carbohydrate molecule extremely versatile, for example, for specific recognition signals on the cell surface.2 In any biological system whatsoever the shape and size adopted by carbohydrates in different solvent environments have been widely demonstrated to be responsible for the biological function of these molecules. In recent years, significant progress has been made in the improvement of both the experimental and theoretical research tools needed to study the conformational complexity of carbohydrates in solution, such as X-ray and neutron scattering techniques (SAXS and SANS), atomic force microscopy (AFM), high-resolution NMR spectroscopy and relaxation techniques, and computational methods. All the experimental and computational methods unequivocally indicate the relevance of the environment (e.g., solvent composition, pH and salt conditions, temperature) on the topological shape and the properties of the carbohydrate solutes. The general problem of solvent effect on the conformational states and the preferential solvation of oligo- and polysaccharides has been tackled mainly to validate detailed molecular models which were developed for relating the structural characteristics of these macromolecules to their chemical, physical, and biological properties in solution.3-5 Semi-empirical methods (unrefined in the sense that molecular parameters are not adjusted for the specific case studied) have been proven useful and generally applicable to different chain linkages and monomer composition. However, it must be clear that all these methods, unless specifically stated otherwise, refer to calculations of the unperturbed chain dimension and therefore do not take into account the excluded volume effects (which arise typically from long-range interactions). In most cases, solvation effects (which are short range) are not explicitly taken into consideration, e.g., molecular parameters of the solvent do not enter in the calculations, although some exceptions are found in literature. The rationale for the correct setting of current knowledge about the shape of polysaccharides in solution is based on three factors: the correlation between primary structure (i.e., the chemical identity of the carbohydrates polymerized in the chain), intrinsic conformational features dictated by the rotational equilibria (often the major contributions are due to the rotation about the glycosidic linkages) and the interaction with the other mo-
12.2 Chain conformations of polysaccharides
707
lecular species in the system (mainly the solvent which determines, therefore, the solubility of the chains). In this chapter the description of the solvent effect is given within the framework of some specific experimental results and computational methods for studying and predicting oligo- and polysaccharide conformations in solution. It is not the authors’ intention to make an in depth investigation into the general methodologies which have been widely reported over recent years (and in this book) but rather to provide a step-wise presentation of some conformational features which have upheld theoretical predictions with experimental observations. The number of examples and approaches is necessarily limited and the choice undoubtedly reflects the authors’ preferences. Nonetheless, the aim is to be as informative as possible about the conceptual difficulties and conceivable results. 12.2.2 STRUCTURE AND CONFORMATION OF POLYSACCHARIDES IN SOLUTION 12.2.2.1 Chemical structure The primary structure of polysaccharides (glycans) is often complicated by different kinds of linkages in homopolymers and different kinds of monomeric units, which give rise to a huge number of different polymers. Glucans (see a general formula in Figure 12.2.1) are those composed exclusively of glucose, while glucuronans are polymers of glucuronic acid. Similarly mannans and galactans as well as mannuronans and galacturonans, are homopolymers of mannose, galactose, mannuronic acid and galacturonic acid, respectively. Although all the polysaccharides discussed here show a structural regularity, they may not be simple homopolymers. Their chemical structure can sometimes be fairly complicated.
Figure 12.2.1. Example of the chemical structure of a polysaccharidic chain as a sequence of β-(1-4)- linked D-glucose units having a side chain of β-glucose linked (1-6) to the backbone. The glycosidic dihedral angles are also indicated.
12.2.2.2 Solution chain conformation A regularity of primary structure could imply that the chains may assume ordered helical conformations, either of single or multiple strand type, both in the solid state and in solution. A knowledge of both polysaccharidic chain structure, up to the three-dimensional molecular shape, and the interaction of the polymers with other molecular components, is essential, in order to understand their capability to form supramolecular structures, including physical gels, with specific rheological properties, which have important implications for controlling and upgrading properties in industrial applications. The rationale is that the physico-chemi-
708
Ranieri Urbani and Attilio Cesàro
Figure 12.2.2. (a) Snapshot of random chains of homoglucans obtained from Monte Carlo calculation and (b) ordered helical structures of the same polymers as measured by X-ray diffraction method.
cal properties, which form the basis of the industrial applications of polysaccharides are directly related to the structure/conformation of the chain. Many biological macromolecules in the solid state assume regular helices, which can be represented by means of a few geometrical parameters and symmetry relations. These helical structures originate from the stereo regularity of backbone monomers and are easily described, conceptually simple, and therefore used (all too often) as an idealized model for all the actual shapes. Elements of helical regularity are essential in the description of the structure of nucleic acids and polypeptides. Some biopolymers, e.g., globular proteins, almost completely preserve their structural regularity in solution. Such globular structures are not known in polysaccharides. Nonetheless, helical conformations have been proposed to represent the structure of many polysaccharides, microbial glycans in particular. An excellent overview of the stereo-regular helical conformations of polysaccharides deduced from X-ray fiber diffraction studies has recently been published.6
12.2 Chain conformations of polysaccharides
709
However, thermodynamic arguments suggest that a partially disordered state is an essential prerequisite for the stability of polymeric systems in solution. Under these circumstances, realistic chain pictures of polysaccharides will not necessarily be generated from the repetition of a single conformational state, which is usually identified by the minimum energy state found in the internal conformational energy calculations. Statistical approaches to these conformational energy surfaces7,8 suggest a more disordered solution conformation (Figure 12.2.2, snapshots marked with a) than the chain structures proposed so as to fit the helical regularity deduced from x-ray fiber diffraction studies (Figure 12.2.2, structures marked with b). Thermal fluctuations are in general sufficient to generate delocalized disorder, unless diffuse interactions, cooperative in nature, ensure long-range order. From the experimental point of view, several polysaccharides with different chain linkage and anomeric configuration have been studied to determine to what extent the polymeric linkage structure and the nature of the monomeric unit are responsible for the preferred solvation and for the chain topology and dimensions.9 Conversely, since it is generally understood that the structure and topology of many macromolecules are affected by solvation, theoretical models must include these solvent effects in addition to the internal flexibility, in order to estimate changes in the accessible conformations as a result of the presence of the solvent molecules. The concept of chain conformational disordering and dynamics in solution is associated with the existence of a multiplicity of different conformations with accessible energy and, moreover, exhibiting their topological differentiation. The above conformational variability of polymeric chains is implicitly recognizable by the great difficulty in crystallization and by the typical phenomenon of polymorphism. This discussion is relevant in particular to ionic polysaccharides (see below) because, from the polyelectrolytic point of view, the transition from a more compact conformation (also an “ordered chain”) to an extended coil conformation, is usually associated with a net variation in ionic charge density along the chain. Polysaccharides generally dissolve only in strongly solvating media. Water displays a complicated behavior: it is a good solvent for monomers and oligosaccharides inasmuch as it is able to compete with the specific inter- and intra-molecular hydrogen bond network (Figure 12.2.3). In many cases it is the thermodynamic stability of the solid-state form which protects the solute molecules from being solubilized.10 Nevertheless, some other strong solvents, such as, dimethylsulfoxide, DMSO, and 1,4-dioxane are known to be good solvents for carbohydrate polymers. Many commercial applications of polysaccharides require compatibility with different solvents and solutes (organic solvents, salts, emulsifiers, plasticizers, enzymes, etc.), for example, in pharmaceutical matrices, paints and foods. In this field, solvent compatibility of some glycans has been improved and controlled by functionalization and derivatization in order to obtain a proper degree of substitution, which determines a wide range of compatibility properties. At the molecular level, various specific and non-specific solvent-solute interactions may occur in polysaccharide solutions that may result in a change in the conformational shape, solubility, viscosity and other hydrodynamic and thermodynamic properties. Hydrophilic interactions such as hydrogen bonding and electrostatic interactions are believed to be factors that influence the conformation of polysaccharides in solution, although the question is being raised (more and more) as to the implication of patches of hydrophobic intermolecular interactions, especially for chain aggregations. One important feature is the
710
Ranieri Urbani and Attilio Cesàro
Figure 12.2.3. Role of structural, conformational and substituent features on the solution properties of oligo- and polysaccharides.
surface that saccharide segments address to solvent molecules that allows a high degree of favorable interactions. The water structuring in the solvation of polysaccharides also contributes to the stability of saccharides in solution,11 which may be altered by competition of other co-solutes or co-solvents which are able to modify the extent of hydrogen-bonded interactions among components. Urea, for example, is considered a breaker of ordered polysaccharide conformation in gel networks, such as those obtained with agar and carrageenan as well as with microbial polysaccharides. Over the last twenty years or so, several approaches have been made to determine the three-dimensional structure of oligosaccharides and theoretical calculations are becoming
12.2 Chain conformations of polysaccharides
711
an increasingly important tool for understanding the structure and solution behavior of saccharides.11-15 Although computer facilities and calculation speeds have grown exponentially over the years some of these techniques, like quantum mechanic methods (ab initio methods), have been shown not to be useful in dealing with the complexity of systems which involve many atoms such as in macromolecular systems. On the other hand, these techniques have been successfully applied to small molecules, e.g., mono- and disaccharides, in predicting charge distribution on the atoms, conformations and transitions among accessible conformations,16 thus providing a background knowledge for the more complicated systems. 12.2.3 EXPERIMENTAL EVIDENCE OF SOLVENT EFFECT ON OLIGOSACCHARIDE CONFORMATIONAL EQUILIBRIA The problem of sugar conformation and dynamics in solution is related to the question of to what extent are oligo- and polysaccharides intrinsically flexible under the different experimental conditions. The answer to this question, which requires a complete knowledge of the time-space dependence of the chain topology,17 is often “rounded-off” by the use of empirical terms like “flexibility”. A further problem is to what extent does the solvent contribute to stabilizing some conformational states rather than others. Solution properties are functions of the distribution of conformations of the molecules in the solvated states, in the sense that the experimental data are statistical thermodynamic averages of the properties over all the accessible conformational states of the molecule, taking each state with a proper statistical weight. This aspect can be better illustrated by taking into consideration the accessible conformational states of a simple sugar unit and the conformational perturbation arising from the changes in the interactions with the surrounding solvent medium. For an α-pyranose ring, for example, the rotation about carbon-carbon bonds and the fluctuations of all ring torsional angles give rise to a great number of possible conformers with different energies (or probabilities). Some of these are identified as the preferred rotational isomeric states in various environments. The boat and boat-skew conformers of a pyranose ring, which are higher in energy than the preferred chair form, correspond to a major departure from the lowest energy chair conformation as illustrated by the globular conformational surface of Figure 12.2.4. Additional conformational mobility in a monosaccharide is due to the rotations of exocyclic groups, namely, OH and especially CH2OH. Let us point out that in a polymer these ring deformations do not normally occur, but, when they do, they may determine to a great extent the equilibrium mean properties and the overall mean chain dimensions.18,19 However, one very recent theory is that the elastic properties of amorphous polysaccharides are related to the glycosidic ring deformation.20 D-ribose is probably the best example of sugar that reaches a complex conformational equilibrium, giving rise to the mixture composition shown in Figure 12.2.5. The percentage of each form is taken from Angyal’s data21,22 with integration of the sub-splitting between the 4C1 and the 1C4 forms. The prime purpose of this analysis is to point out that the stability of each conformer is not determined solely by the intrinsic internal energy, which can be evaluated by means of e.g., ab-initio quantum mechanics calculations, but is strongly influenced by all the solvation contributions. Therefore, the conformer population may be shifted by changing temperature, solvent composition, or any other external variable such as, for example, adding divalent cations (see Table 12.2.1). The evaluation of the actual concentration of the several conformers involved in the equilibrium immediately leads to
712
Ranieri Urbani and Attilio Cesàro
Figure 12.2.4. Representation of all possible conformations of the pyranosidic ring and respective inter-conversion paths.
Figure 12.2.5. D-ribose conformational equilibria.
12.2 Chain conformations of polysaccharides
713
the partition function of the system (i.e., the free-energy change or the equilibrium constant for a transition between conformers). Table 12.2.1. Percentage composition of furanose and pyranose forms of D-ribose and glucose in different solvent systems Sugar
Solvent H2O, 31oC o
D-ribose
D2O, 30 C o
DMSO, 30 C 2+
Ca , 1.27 M
α-F
β-F
α-P
(4C1) β-P
(1C4) β-P
13.5
6.5
21.5
44.0
14.5
8.0
14.0
23.0
41.0
14.0
6.0
22.0
16.2
31.0
24.8
13.0
5.0
40.0
14.0
28.0
45.0
55.0
45.0
53.0
o
Glucose
DMSO, 17 C o
pyridine, 25 C
0.6
1.0
Passing from monomer units to oligomers (disaccharides and higher oligosaccharides), the dominant features of molecular flexibility become those due to rotations about the glycosidic linkages. Although other conformational fluctuations may contribute to the local dynamics of the atoms or group of atoms, only the glycosidic linkage rotations are able to dramatically change the conformational topology of oligomers at ambient temperature. The aim of a conformational analysis of oligosaccharides is thus to evaluate the probability (that is the energy) of all mutual orientations of the two monosaccharidic units, as a function of rotations about the glycosidic linkages, defined by the dihedral angles: ϕ = [H1-C1-O1-Cm] ψ = [C1-O1-Cm-Hm] where m is the aglycon carbon number in the reducing ring. The important region of ϕ and ψ rotations is the one with energy variations in the order of kT, the thermal motion energy, because this may produce a large ensemble of accessible conformational states for the oligosaccharide. Even when the rotational motion is restricted to only a few angles, the fluctuations of many such glycosidic bonds is amplified along the chain backbone, as the molecular weight increases. The accumulation of even limited local rotations may produce very large topological variations in the case of polymeric chains and consequently relevant changes in thermodynamic, hydrodynamic and rheological properties of these systems. Other internal motions often make only small contributions to the observable properties on the macromolecular scale.23 Experimentally, NMR techniques are among the most valuable tools for studying conformations and dynamics of oligosaccharides in solution by determining chemical shifts, coupling constants, NOE’s and relaxation time.24-30 In the study of saccharide conformations the potential of coupling constants evaluation, specially the (hetero-nuclear) carbon-proton spin-spin coupling constants, nJC,H, and their dependence on solvent medium is well known. Several empirical correlations between 1JC,H and structural parameters like dihedral angles have been reported,31 although the low natural abundance of the 13C isotope often made the measurements technically difficult and 13C enrichment was required for the generation of data of sufficient quality for the quantitative analysis. Nowadays, due to the progress of FT-NMR spectrometers, measurements of the three-bond (vicinal) carbon-pro-
714
Ranieri Urbani and Attilio Cesàro
ton coupling constant, 3JC,H have indeed been made possible on the natural-abundance 13C spectra. Much endeavor has been invested in establishing a general Karplus-like relation for the angular dependence of nJC,H, especially on the dihedral angles in the glycosidic region because the overall three-dimensional structure of oligo- and polysaccharides are related to the glycosidic features in terms of dihedral (rotational) angles. From comparison of experimental data and theoretical results (using quantum-mechanical and semi-empirical methods) on model compounds,32-34 the angular dependence of the coupling constants, nJC,H, on the anomeric and aglycon torsional angles ϕ and ψ,31,34 and on the dielectric constant of the solvent35 can be written in a general form such as: 1
JC , H = A1 cos 2χ + A2 cos χ + A3 sin 2χ + A4 sin χ + A5 + A6 ε
3
JC , H = B1 cos 2 χ + B 2 cos χ + B 3 sin χ + B 4 sin 2χ + B 5
[12.2.1]
where: Ai, Bi χ ε
constants different for α and β anomers ϕ or ψ solvent dielectric constant
Experimental nJC,H values for conformationally rigid carbohydrate derivatives allow to calculate the constant values of Ai and Bi in equations [12.2.1]. The major practical use of these equations is their ability to estimate the glycosidic dihedral angles from experimental n JC,H data, in combination with other complementary results, for example, from NOESY,36 X-ray and chiro-optical experiments. Since the experimental values are averaged over all the accessible conformational states in solution, they do not necessary reflect the property of only the most probable conformer23 but they nevertheless include contributions of all the conformers, each one taken with its proper statistical weight. Thus, the quantitative interpretation of experimental data in terms of accessible conformational states of flexible molecules requires the additional theoretical evaluation of the energy of the molecule as a function of internal coordinates. Since the dependence of the observed coupling constants on the conformation is non-linear, it derives that: n
JC , H (χ) ≠ n JC , H ( χ
)
Only for a linear dependence of nJC,H on χ, the equation could hold the equals sign. Because of the simultaneous dependence of the property on χ and on the complexity of potential energy function, E(χ), the ensemble average of
12.2 Chain conformations of polysaccharides
715
extended by Rees and co-workers,37,41 who derived expressions for the contribution to the optical activity of changes in conformational states of oligo- and polysaccharide chains. More recently, Stevens and co-workers42,43 developed a computational model for optical activity and circular dichroism based on the Kirkwood theory in which the calculations of the lowest energy component of a molecular σ − σ * transition, which derives from the mutual interaction of all σ − σ * transitions on C-C, C-H and C-O bonds, are carried out. The calculated optical properties are therefore geometry-dependent and the optical properties have been calculated as statistically averaged properties over an ensemble of all possible conformations, theoretically obtained as a function of ϕ and ψ. This model also provides a useful tool for testing the quality of force field parametrizations and the possibility of some refinement of force constants for a non-explicit inclusion of solvent environment. 12.2.4 THEORETICAL EVALUATION OF SOLVENT EFFECT ON CONFORMATIONAL EQUILIBRIA OF SUGARS 12.2.4.1 Classical molecular mechanics methods Theoretical approaches to the conformational analysis of oligosaccharides in solution become inherently more complicated than those for monomeric sugars, at least for the computer time required to optimize the structures. Depending on the degrees of freedom taken into consideration and on the level of sophistication of the method used, the conformational analysis may give different results. It has been the custom in the past to compute the conformational energy as a function of the dihedral angles ϕ and ψ only, by assuming the sugar ring to be rigid (rigid-residue method). More recently, all the internal coordinates are allowed to adjust at each increment of ϕ and ψ, relaxing the structure toward a local minimum (relaxed-residue method). The rigid-residue approach is still considered suitable as a starting point, although some warning must be alerted to the use of the same set of structural coordinates for all sugars, as was initially done. The approach may however be useful as a starting point in the conformational analysis of polysaccharides, provided that the coordinates of all the atoms in the monomeric unit have been calculated by a suitable independent method.13, 44 Figure 12.2.6 reports, as a general example, the rigid-residue energy surfaces of two representative disaccharides, namely α-(1-4)-D-glucopyranosil-D-glucopyranose (D-maltose, Figure 12.2.6a) and the β-(1-4)-D-glucopyranosil-D-glucopyranose (D-cellobiose, Figure 12.26b). Both maps present multiple minima separated by barriers that are only a few kilocalories high. For many disaccharides the barriers between minima can be very high and they are often overestimated in the rigid-residue approximation. In general, the conformational energy map shows that only a limited portion of the total ϕ, ψ conformation space is actually accessible to the dissaccharide at room temperature. The dissaccharide is not frozen into its lowest energy conformation, nevertheless the steep walls of the allowed region of conformational energy may dramatically limit the multiplicity of thermally accessible conformational states. One useful method, for taking into account the effect of solvent media upon the conformational properties of glycosidic structures, is the continuum reaction field method.45, 46 This method is based on the Scaled Particle Theory (SPT) equations47 and on Onsager’s theory of the reaction field, as applied by Abraham48 by considering the solvent as a dielectric continuum. In this approach the total conformational energy, Gtot, is given by: G tot = G conf + Gsolv
716
Ranieri Urbani and Attilio Cesàro
Figure 12.2.6. Conformational energy maps calculated for (a) D-maltose and (b) D-cellobiose. Contours are in kcal·mol-1 with increments above the minimum (x) of 0.5 kcal·mol-1.
where the total conformational free energy, Gtot, is given by the sum of the contribution due to the in vacuo conformational state energy, Gconf, and the solvation contribution, Gsolv. The latter term is a sum of contributions due to the energy required to create, in a given solvent, a cavity of suitable size to accommodate the solute molecule in a given conformational state and the interaction energy between the solute and the surrounding solvent molecules, as schematically shown in Figure 12.2.7:49 Gsolv = G cav + G el + G disp [12.2.2] The expression for Gcav, that is the free energy required for the formation of a cavity (first step in Figure 12.2.7), at a temperature T and a pressure P, is taken from the Scaled Particle Theory (SPT) which has been successfully applied in the study of thermodynamic properties of aqueous and non-aqueous solutions:47,50
Figure 12.2.7. General scheme of the theoretical approach for the solvation energy calculation.
12.2 Chain conformations of polysaccharides
717
2 3y $ 3y G cav yP $ 3 9 y $ 2 R + R + + = − ln (1 − y) + R 1− y 2 1− y RT ρkT 1− y
[12.2.3]
where: y av, au $ R ρ R k
= 4 πρa 3v / 3, the reduced number density radii of hard sphere equivalent solute (u) and solvent (v) molecules = au / av solvent density gas constant Boltzmann constant
It is noteworthy that Gcav is a function solely of solvent density and solvent and solute dimensions and represents a measure of the cohesive forces among solvent molecules. In a given solvent, this contribution is a function of the radius of solute molecule which may differ remarkably as a function of dihedral angles ϕ and ψ only. The electrostatic interaction term Gel between solute and solvent is based on the continuum reaction field48 which takes into account a reaction potential induced by the solute dipole and quadrupole (third step in Figure 12.2.7) in a continuum medium of dielectric constant ε: G el =
3HX KX bF + bF 1 − exp − 1 − lX 5 − X 16RT
[12.2.4]
with: K=
µ 2u a
3
, H=
Q u2 a
9
, l=
(
n +2 2 u
),
X=
ε −1 2ε + 1
ε≤2
0 for
F=
5
2 n u2 − 1
(ε − 2)(ε + 1) ε 1/ 2
T b = 4.35 300
(
1/ 2
a 3 / 2 ruv3
for
)
3
ε>2
K + Ha 2 r2 uv
1/ 2
where: µu Qu nu a ruv
solute dipole moment solute quadrupole moment solute refractive index = av +au = a/21/2
ruv is the average distance between the solvent and the solute molecule defined in terms of the radius of cavity a.
718
Ranieri Urbani and Attilio Cesàro
The free energy of dispersion, Gdisp, in equation [12.2.2] takes into account both attractive and repulsive non-bonding interactions and is expressed as a combination of the London dispersion equation and Born-type repulsion:51 G disp = −0.327Nv α u α v
I u Iv −6 ruv I u + Iv
[12.2.5]
where: α I Nv
molecular polarizability ionization potential nearest-neighbor solvent molecules
Nv is the number of molecules surrounding the solute molecule in a given conformation and is calculated from the following equation:52
(
N = a + 2a v u v
)
3
4πNA − a 3V v 3 u
3
where: Vv NA
solvent molar volume Avogadro number
In order to elucidate the effect of different solvents on the conformation, the energy differences between conformers in a given solvent are more relevant than the absolute solvation energies in each solvent. In particular, it is important that the perturbation effect on the detailed shape in the low energy regions of the conformational map. Figure 12.2.8 shows the free-energy of solvation and energy contributions for the maltose dimer as a function of ψ calculated at ϕ=-30° and refers to the energy of the (-30°, 180°) conformer for two solvents, water and DMSO (Figure 12.2.8 a and b, respectively). In this section, the cavity term is a complex function of the size of the maltose molecule, as determined by the spatial orientation of the two glucose residues. The cavity energy is, by definition, always unfavorable; the more expanded the conformers, which are usually located in the low energy region of the map (around ψ=0° for maltose), the larger the Gcav due to the size of larger cavities to be created in the solvent. The electrostatic free-energy, Gel, increases with the dipole and quadrupole moments of solute molecule and decreases with the radius of the cavity. On the other hand, the dipole moment µ is a function of dihedrals ϕ and ψ and for maltose has two large maxima around (ϕ, ψ)=(-40°,0°) and (180°, 140°) which is in agreement with the major contributions of Gel shown in Figure 12.2.8. The electrostatic free-energy (eq. [12.2.4]) is also dependent on the solvent dielectric constant and increases in passing from DMSO (ε=46.68 debye) to water (ε=78.30 debye) at 25°C. The dispersion term (eq. [12.2.5]) makes a significant contribution (-20 to -70 kJ·mol-1) to the absolute value of solvation free-energy but the angular dependence is very small giving an almost equal contribution to the energy of conformers. In an early paper53 it was shown that, in the comparative cases of cellobiose and maltose, the probability distribution of conformers (Figure 12.2.9) is affected by the presence of the solvent, changing the shape of the function from one solvent to another. The minimum of the maltose map goes from (-20°,-30°) to (-10°,-20°) in water and DMSO, while that of cellobiose map goes from (0°, 50°) to (-30°, -20°) in both solvents.53 These solvent perturbations (apparently small) on the conformational energies have a great effect on the proba-
12.2 Chain conformations of polysaccharides
719
Figure 12.2.8. Section of the maltose energy map at ϕ=-30° showing the dependence of contributions to the Gsolv as a function of angle ψ in water (a) and DMSO (b).
Figure 12.2.9. Histograms of probabilities of conformational states for cellobiose in vacuo (a), water (b) and DMSO (c), and for maltose (a’, b’, and c’).
720
Ranieri Urbani and Attilio Cesàro
bilities associated with the conformers and therefore produce a rearrangement of statistical weights among conformers and, thus, a variation of the average properties of the system, as for example illustrated by the changes of the chain dimension parameters (see below Figure 12.2.11). Calculated average data like dipole moment, µ, linkage rotation, Λ, and proton-carbon coupling constants across glycosidic linkage, 3JC1-H4 and 3JC4’-H1, for β-D-maltose14 and β-D-mannobiose45 in different solvents are reported in Table 12.2.2. These results show that the equilibrium composition of dimer conformers depends strongly on the solvent and that the departure from the in vacuo conformation increases with increasing solvent dielectric constant. For mannobiose the determining factor of the solvent effect on the conformation is the intra-residue electrostatic interaction, which depends on ϕ and ψ in the same manner as the dipole moment. Table 12.2.2. Calculated average values of dipole moment, four-bonds proton-carbon coupling constant and linkage rotation of β-D-maltose and β-D-mannobiose in different solvents at 25°C Solvent
ε
ϕ, deg
ψ, deg
µ, debye
3
JC1-H4, Hz
JC4'-H1, Hz
Λ, deg
3
solute: β-D-maltose vacuum
-
-21
-28
3.80
4.2
4.5
-31
1,4-dioxane
2.21
-22
-22
3.89
4.1
4.5
-30
pyridine
12.40
-24
-39
4.02
4.0
4.5
-27
ethanol
24.55
-24
-41
4.09
3.9
4.5
-27
methanol
32.70
-25
-46
4.17
3.9
4.5
-25
DMSO
46.68
-24
-41
4.10
3.9
4.5
-28
water
78.30
-28
-65
4.47
3.7
4.7
-19
solute: β-D-mannobiose vacuum
-
81
-18
6.00
2.29
4.14
77
1,4-dioxane
2.21
82
-19
5.92
2.23
4.15
76
pyridine
12.40
91
-22
6.01
1.95
4.22
69
methanol
32.70
97
-24
6.08
1.72
4.25
54
DMSO
46.68
94
23
6.05
1.84
4.23
67
water
78.30
100
-18
6.27
1.53
4.32
57
It should be pointed out that even the small variations in the dihedral angles due to the presence of solvent molecules may produce a large change on a macromolecular scale, as discussed above. 12.2.4.2 Molecular dynamic methods One of the most powerful theoretical tools for modeling carbohydrate solution systems on a microscopic scale and evaluating the degree of flexibility of these molecules is the molecular dynamics technique (MD) which has become popular over the last two decades. The first reported works of MD carbohydrate simulation appeared in 198654,55 and since then an in-
12.2 Chain conformations of polysaccharides
721
creasing number of MD simulations on sugars has been carried out.11,13,56,57 The explicit representation of solvent molecules is required especially in biological systems, where solvent structuring plays an important role. Starting with an appropriate potential energy function for sugar-water interactions,13,58 the common solution simulations are carried out by placing the solute molecule in the center of a cubic box of finite dimensions containing a given number of solute molecules which usually corresponds to at least three solvation shells.11 The macroscopic system is then simulated by using the approximation known as “periodic boundary conditions”,59 in which the entire box is replicated in every direction, leaving solute and solvent molecules to interact with each other both in central and in replica boxes and setting the long-range interactions to smoothly decrease to zero by using the appropriate switching functions. A pair distribution function, g(r), defined as: g(r ) =
1 dN(r ) 4πρr 2 dr
where: r ρ N(r)
interatomic distance the bulk number density number of atoms of given type at distance r
has been used to evaluate the normalized probability of finding a water oxygen atom at a distance r from a given atom on the carbohydrate molecule.11 In this way the anisotropic distribution of solvent molecules around carbohydrate solutes was reported11,60,61 showing an exceptional structuring of water molecules which extends to greater distances around the solute as compared to the pure solvent. One of the most interesting results of these simulations is the identification of the spatial distribution of water molecules on the van der Waals’ surface of the carbohydrate molecules. Figure 12.2.10 shows a probability density excess of the water molecules in a channel, which is effectively a bisector of the two closest sites for hydrogen bonding. When the energy maps have been obtained by molecular mechanics calculations, dynamics simulations of disaccharides in various conformations are carried out to analyze the
Figure 12.2.10. Contours of solvent anisotropic density around the α , α-trehalose disaccharide obtained from MD simulation [unpublished figure kindly provided by J.W. Brady and Q. Liu].
722
Ranieri Urbani and Attilio Cesàro
typical motions of the molecules along the conformational space at moderate temperatures. In general, initial conformations in the simulations are taken from one of those of minimized geometry as identified in the calculation of the energy map. Trajectories are then computed by assigning to the atoms velocity components randomly selected from a thermal distribution at a given temperature. By superimposing the trajectories of fluctuations of ( ϕ, ψ) on the energy map, it is possible to observe a variety of motions, involving both the structures of the molecular rings and in some cases rotations about the glycosidic bonds and exocyclic torsions.62 In the case of multiple minima, conformational transitions between low energy regions may also be revealed. 12.2.5 SOLVENT EFFECT ON CHAIN DIMENSIONS AND CONFORMATIONS OF POLYSACCHARIDES In addition to the change of entropy, minimization of free energy in a binary dilute solution containing solvent and polymer occurs as a result of the favorable interactions between the chain segments and the solvent, which replace the homotactic interactions between solvent molecules and those between chain segments. In a good solvent and a very dilute solution, it is likely that only solute-solvent and solvent-solvent interactions prevail. However, in a bad solvent (and, in general, in a more concentrated solution) persistence of segmental interactions among chains is the major contribution in the macroscopic properties of the system. The theoretical evaluation of the entropic contributions arising from the configurational nature of the chain molecule is possible on the basis of thermodynamic-statistical models.63 More troublesome is the contribution of enthalpy change of mixing, ∆ mix H, of non-ionic polysaccharides and water which cannot be predicted, not even in sign. Furthermore, the scarcity of literature data does not allow any empirical rationalization although, in most cases, contributions significantly smaller than the related monomers are reported. Notwithstanding the above limitations, a general picture can be drawn showing that the average dimension of a chain, experimentally obtained with light scattering or viscometric measurements, depends upon solvent interactions and behavior, in addition to the intrinsic features of the polysaccharide (chemical nature and linkage of monomers, conformational equilibria, etc.). The basic axiom is that solution properties are strictly related to the conformation of the molecules in the solvated state and this state, in turn, is only statistically defined from the primary (chemical) structure. In first instance, one can generalize the statement that non-ionic crystalline molecules and especially polymers, barely preserve their ordered conformation upon dissolution (Figure 12.2.2). Polymers and oligomers (including most carbohydrate molecules) generally assume a statistically disordered conformation in solution since, in the absence of specific favorable enthalpy contributions, polymer-solvent interactions provide a small increment in the entropy of mixing. Under these very common circumstances, the dissolution of a crystalline carbohydrate molecule, which is stabilized in the solid state by a great number of interactions, becomes a thermodynamically unfavorable event. Structurally speaking, the dissolution is made possible only by a sufficient increment in conformational entropy, which is described by the increment in conformational states accessible to the molecule. In general, some conformational features, resembling those observed in the solid state, can be preserved also in solution (Figure 12.2.2), but the degree of order is strictly related to the presence of the solvent, in addition to the temperature and to the entropy of mixing. Figure 12.2.2 shows glucan chains with different types of linkages, and different pictorial trajectories which give different and sometimes surprising values of configurational entropy
12.2 Chain conformations of polysaccharides
723
are also observed. For example, those of (1-4)-linked α-D-glucan and of (1-3)-linked β-D-glucan seem fairly restricted to some pseudo-helical character compared to the more disordered set of trajectories that would be obtained if rotations about the glycosidic bonds were completely unrestricted. Possible interactions between residues of the polysaccharide chain that are not nearest-neighbors in the primary sequence of the polymer can sometimes be ignored. In this case, a computer-based polysaccharide chain can be constructed from the conformational energy map of the dimeric units. The Monte Carlo method64 and the Flory matrix methods65 are commonly used in the so-called “nearest-neighbors approximation” to mimic the polymer chains in the pure amorphous state or in dilute solution. The Monte Carlo sample of chains reflects the range of conformations experienced by any single chain as a function of time or, equivalently, the range of conformations in a large sample of chemically identical polymer molecules at any instant in time. In either sense, the sample can be analyzed to deduce both the characteristics of individual chain conformations and the mean properties of the sample as a whole, which correspond to those in the equilibrium state of the chain. Results refer, however, to an “unperturbed” chain model that ignores the consequences of the long range excluded volume effect, because only nearest-neighbor interactions are accounted for. Given a sufficient Monte Carlo sample of chains in equally probable representative conformations, it is possible to assess many mean properties of the polymer in question simply by computing numerical (unweighted arithmetic) averages over the number of chains in the sample. For example, the mean square end-to-end distance, the mean square radius of gyration, or the angular dependence of scattered light (particle scattering factor) are all mean geometric properties readily computed from a knowledge of the coordinates of the atoms or atomic groups which are generated in the Monte Carlo sample. The average topological properties are described through the chain-length dependence of quantities such as the characteristic ratio, Cn, the persistence length, Pn, and the correlation function, Fn, defined as: r L1 n r Cn = , Pn = ∑ L i L1 i =1 nL2 r2
rr , Fn = uu n
0
0
[12.2.6]
0
where: n L
number of saccharide units average virtual bond length mean square end-to-end distance
The virtual bond vector is often defined for each monosaccharidic unit as connecting oxygen atoms involved in glycosidic linkages. Figure 12.2.11 shows the calculated properties of equations [12.2.6] for two homoglucan chains which differ only in the configuration of the anomeric carbon, i.e., (a) the [α-D-(1-3)-glc]n (pseudonigeran) and (b) the [β-D-(1-3)-glc]n (curdlan). Whatever the chemical features, provided that the molecular weight is very large (that is for a degree of polymerization n which approaches infinity), the distribution of unperturbed polymer end-to-end length is Gaussian and Cn equals an asymptotic characteristic ratio C∞, i.e.:
724
Ranieri Urbani and Attilio Cesàro
C∞ = lim
n→∞
r2 nL2
=
6 Rg2 nL2
where Rg is the average square radius of gyration experimentally accessible. This relation is considered extremely important in the sense that any conformational perturbation is amplified by a factor n in the final value of C∞ (or of
12.2 Chain conformations of polysaccharides
725
Figure 12.2.12. Characteristic ratio, Cn, as a function of degree of polymerization, n, for amylose (a) and cellulose (b) in vacuo, water and DMSO.
By calculating of the conformational energy surface of a dimer as a starting point for the prediction of mean chain properties, the effect of the solvation on the chain conformation is taken into account by evaluation of the perturbations of such surface due to the presence of the solvent. The characteristic ratio (eq. [12.2.6]) has been computed for (a) [β-D-(1-4)-glc]n (cellulose) and (b) [α-D-(1-4)-glc]n (amylose) in different solvents, namely, water and DMSO.53 The results are reported in Figure 12.2.12 where, for comparison, the in vacuo Cn for both polymers is shown. Figure 12.2.12a shows that the chain extension of cellulose decreases significantly in DMSO and much more so in water with a reduction of about 50% when compared to that in vacuo data (C∞=96). The solvation contribution of water seems to favor those conformations, which have smaller cavity volumes and thus force the chain to minor extensions. Tanner and Berry66 reported, for cellulose derivatives in solution, a value for the limiting characteristic ratio of between 30 and 60, which is in good agreement with the theoretical data in Figure 12.2.12a. Figure 12.2.12b shows the characteristic ratio of amylose chains computed in the same solvents as in Figure 12.2.12a. The amylose chain extension is considerably lower than that of cellulose and, as in the case of [β-D-(1-3)-glc]n shown above, presents at low n a remark-
726
Ranieri Urbani and Attilio Cesàro
able pseudo-helical pattern. With respect to cellulose, the amylose chain shows a more coiled and apparently disordered backbone topology, but from a statistical point of view possesses a lower configurational entropy,67 i.e., a more limited set of allowed conformational states. The C∞ values are higher in water (5.6) than in DMSO (4.9) and both are higher than for the unperturbed chain which is in good agreement with the earlier work of Jordan and Brant67 which observed a decrease of about 20% in chain dimensions of amylose DMSO/water mixture with respect to the water alone. More recently, Nakanishi and co-workers68 have demonstrated from light scattering, sedimentation equilibrium and viscosity measurements on narrow distribution samples, that the amylose chain conformation in DMSO is a random coil, which results expanded (C∞ = 5) by excluded-volume effect at high molecular weight. Norisuye69 elaborated a set of published viscosity data reporting a C∞ = 4.2-4.5 for unperturbed amylose and C∞ = 5.3 for aqueous KCl solutions, while Ring and co-workers70 estimated a C∞ ≅ 4.5 for amylose/water solutions by means of the Orofino-Flory theory of the second virial coefficient. Order-disorder conformational transitions very often occur on changing physical and/or chemical conditions of polysaccharide solutions. DMSO, for example, is the solvent, which is commonly used as co-solvent for stabilizing or destabilizing ordered solution conformations. Schizophyllan, a triple helical polysaccharide with a [β-D-(1-3)-glc]n backbone exhibits a highly cooperative order-disorder transition in aqueous solution.71 When small quantities of DMSO are added to aqueous solutions the ordered state is remarkably stabilized, as has been observed in the heat capacity curves by means of the DSC technique.71 Several efforts have been made with MD simulations in order to explicitly take into account the solvent molecule effect on the saccharide conformation, although only oligomeric segments have been considered, given the complexity in terms of computational time required for such a multi-atoms system. One interesting example is that of Brady and co-workers56 on the stability and the behavior of double-helix carrageenan oligomer in aqueous solution compared with the results of the in vacuo calculations. They observed a higher relative stability of the double helix in vacuo, a fact, which is consistent with experimental results under anhydrous conditions, as in the fiber diffraction studies. However, in aqueous solution, the interchain hydrogen bonds that stabilize the double-helix structure appear much less stable, as the glycosidic hydroxyl groups make more favorable interactions with water molecules. They concluded that in the solvation step the double-helix would seem to be unstable and an unwinding process is theoretically predicted, at least for the oligomers. 12.2.6 SOLVENT EFFECT ON CHARGED POLYSACCHARIDES AND THE POLYELECTROLYTE MODEL 12.2.6.1 Experimental behavior of polysaccharides polyelectrolytes Based on the experimental evidence of polyelectrolyte solutions, whenever the degree of polymerization is sufficiently high, all ionic macromolecules are characterized by a peculiar behavior, which sets them apart from all other ionic low molecular weight molecules as well as from non-ionic macromolecules. A general consequence of the presence of charged groups in a chain is a favorable contribution to the solubility of polymer in water. A strongly attractive potential is generated between the charge density on the polymer and the opposite charges in solution. For example, the value of the activity coefficient of the counterions is strongly reduced with respect to that of the same ions in the presence of the univalent opposite charged species. If the charge density of the polyelectrolyte is sufficiently high, such a
12.2 Chain conformations of polysaccharides
727
phenomenon is justified through a ‘condensation process’ of counterions and it has also been interpreted theoretically. On the polymer side, among the dramatic changes that the presence of charged groups imparts to solution properties, there are the enhanced chain dimensions, the increased hydrodynamic volume (i.e., viscosity), and, in general, a strong influence on all conformational properties. Subject to the constraints imposed by the chemical structure of the chain, the distribution of charged groups and their degree of ionization contribute to determining the equilibrium chain conformation; both the Coulombic interaction among the charged groups and the distribution and concentration of the screening counterions are important. Most of the physico-chemical properties of the system result from a non-linear combination of these parameters. However, one has not to forget that the variability of conformation alters the distances between charged groups on the polymeric chain and that the equilibrium is statistically defined by the Gibbs energy minimum of the system. As an important consequence of this energy balance, changes in temperature, ionic strength, pH, etc., can provoke changes in polyelectrolyte conformation, often cooperatively in the case of biopolymers, between states with different values of the charge density. These states may be characterized by different structural orders (e.g. helix →extended chain transition), by different degrees of flexibility of the chain (globular coil → expanded chain) or by different extent of aggregation (monomeric → dimeric or multimeric chains). Theoretical calculations based on molecular grounds are still extremely complicated and incomplete72 and other routes must be more empirically used in order to interpret the experimental data and to understand the correlation between conformational properties and structure. The central problem is to quantify the interactions among charges on the polymer and among these same charges and their respective counterions. As far as it concerns the short-range interactions, the introduction of charged groups modifies the equilibrium geometry of the monomeric units and the contribution of the electrostatic nature on the nearest-neighbor conformational energy. These conclusions also derive from the already demonstrated effect of the solvent interactions on the unperturbed dimensions of amylose and cellulose,53 and from the evidence of the perturbation on the conformational energy surface of several charged saccharidic units.73 There at least two approaches that may be relevant for this review; one is that described by Haug and Smidsrød74 for the rationalization of the dimensional properties of polyelectrolytes as a function of salt concentration, the other is the formulation of a statistical thermodynamic theory for the “physical” framing of the ion-polyelectrolyte interactions. Both these theoretical formulations deal with the conformation of the polymer and predict that the conformational features must be function of ionic strength (see for example refs. 75 and 76). 12.2.6.2 The Haug and Smidsrød parameter: description of the salt effect on the chain dimension A peculiarity of the correlation between the viscometric parameters and the dimensions of the macromolecular chain has long been recognized and theoretical approaches have been developed for several chain models.77,78 The behavior of polyelectrolytes adds some complications especially in the low ionic strength regime. It has however been understood that the intrinsic viscosity, [η], of a polyion (i.e., its hydrodynamic volume) decreases with increasing ionic strength, I, as a consequence of the screening of the fixed charges on the
728
Ranieri Urbani and Attilio Cesàro
Figure 12.2.13. Dependence of the intrinsic viscosity [η] of hyaluronic acid (ξ=0.72) and its benzyl derivatives with decreasing linear charge density, ξ, on the inverse square root of ionic strength, I, at 25°C.
polyion. At infinite ionic strength, the chain dimensions may eventually correspond to the completely uncharged macromolecule. This is a sort of “ideal state” of the polyelectrolyte; “ideal” with respect to the long-range electrostatic repulsive interactions only, without relation to the Θ-conditions. In the absence of a cooperative conformational transition, for many polysaccharide polyelectrolytes a linear dependence of [η] upon I1/2 is observed, with the slope diminishing with the charge density associated to the polysaccharide chain (Figure 12.2.13). A theory has been presented for an estimation of the relative stiffness of the molecular chains by Smidsrød and Haug,74 which is based on the Fixman’s theory and Mark-Houwink equation. The chain stiffness parameter is estimated from the normalized slope B of [η] vs. the inverse square root of the ionic strength: ∂[η] ∂I −1/ 2
= slope = B([η]0.1 )
γ
[12.2.7]
where: γ
has a value between 1.2 and 1.4
The dependence of viscosity on the ionic strength, as given in equation above, has been increasingly popular in the field of polysaccharides with the purpose of comparing the chain stiffness of different macromolecules. The derivation is based on the Fixman theory, which defines the dependence of [η] on the molecular weight through an expansion coefficient which effectively takes into account the electrostatic interactions in the Debye-Hückel approximation. The semi-empirical treatment of the hydrodynamic properties of statistical polyelectrolytes (at sufficiently high values of the ionic strength) is built upon a straightforward extension of the theory of intermolecular interactions for uncharged polymers, for which a linear relation can be written between the expansion coefficient, α 3η , and the square root of the molecular weight, M. It should also be mentioned78 that the various theoretical treatments of the salt dependence of the excluded volume and of the expansion coefficients
12.2 Chain conformations of polysaccharides
729
led to a linearity of α 3η on (M/Cs)1/2 only over a limited range of salt concentration. On varying the salt concentration, one effectively deals with a set of binary solvents with a variety of interaction parameters. The approach proposed by Smidsrød overcomes the indetermination of some parameters by using the slope of [η] as a function of the inverse square-root of the ionic strength of the medium. The ultimate relationship is obtained between the constant B and the effective bond length b θ (B = const·b θ -2), which, in the absence of any reasonable knowledge of the constant, can only be used in an empirical way. There is a compelling although intuitive limit to the use of the Smidsrød-Haug approach for those macroions that do not counterbalance the effective electrostatic field exerted by the ionic strength through the conformational elasticity. Among these polymers, those characterized by a low value of fixed charges, of flexibility, and/or of molecular weight fall behind the limits of the correct applicability of the Smidsrød-Haug approach, which should maintain its validity only for gaussian chains with prominent electrostatic interactions. Most likely, the Smidsrød-Haug parameter has been abused in the field of polysaccharides without the authors’ intention, in the sense that the original treatment was aware about the intrinsic limitations of the approach,74 while the extensions have thereafter been considered as permitted. We wish to point out that the above comment does not imply the failure of the linearity of [η] with I-1/2, but only a meaningless result for the values of b θ obtained for low-charged polysaccharide polyelectrolytes. Reference can be made to two series of polysaccharides (chitosans and hyaluronans), which have been investigated in some detail for the specific application of the Smidsrød-Haug approach.79,80 12.2.6.3 The statistical thermodynamic counterion-condensation theory of Manning Linear polyelectrolytes bear a charge distribution along the chain, properly neutralized by small ions of opposite sign. In the absence of added salt, it is reasonably assumed that the charge density (when sufficiently high) will increase the local stiffness of the chain because of the electrostatic repulsion. For this reason the linear polyelectrolyte is often regarded as a charged rod. Osawa has introduced the concept of a critical charge density on the polymer and described the ion- pairing of polyion and counterions as condensation. The model described here has been extensively used in previous papers of the authors81 and described in the original papers by Manning.82 Let here simply give a few comments on the physical basis of this model relevant to the present case. The counterion condensation (CC) theory, largely developed by Manning, gives analytical solutions to evaluate the electrostatic potential around a linear polyelectrolyte, provided its conformation is regular and fixed. The rigidity of the polymer seems therefore to be both a prerequisite and a result of the molecular polyelectrolyte theory, and it has been thought to be not too far from reality in many cases.82 The application of the above theory to experimental results has been carried out with the assumption that the charge distribution is structurally defined by the monomer repeat as derived from the solid state fiber diffractograms, although sometimes the fully stretched chain conformation has been taken (Figure 12.2.14). The original work of Manning’s counterion condensation theory has provided an elegant tool for describing several properties of polyelectrolytes in terms of the structural parameter, ξ, which is unequivocally defined as e2/εkTb, where e is the value of the elementary charge, ε is the dielectric constant of the medium, k is the Boltzmann constant, T is the Kelvin temperature, and b is the distance between the projections of the fixed charges of the polyelectrolyte on its contour axis. In the case of monovalent ions, for all
730
Ranieri Urbani and Attilio Cesàro
Figure 12.2.14. The rigid chain (a), the corresponding polyelectrolyte model (b) and the realistic flexible model of a polysaccharidic chain (c).
structural values of ξ> 1, a defined amount of counterions will “condense” from the solution into the domain of the polymer chain so as to reduce the “effective” value of ξ to unity. For water, ξ = 0.714/b (with b expressed in nm), and is practically independent of T, being the electrostatic-excess Gibbs free energy of the solution given by: G el = −ξln[1 − exp(−Kb )] where K is the Debye-Hückel screening parameter. Application of this theory has been made to many experimental cases, and in particular an extensive correlation has been made between the theoretical predictions and the thermodynamic data on the processes of protonation, dilution and mixing with ions.76 Besides these nice applications of the theory to problems with a strong “academic” character, there is another very striking example of prediction of solvent-induced conformational changes for the effects of salts on the conformational stability of ordered polyelectrolytes. In fact, in addition to the condensation phenomena predicted by the polyelectrolitic theory, other physical responses may also occur (also simultaneously), which may mask this central statement of the Manning theory “that the onset of the critical value ξ (for univalent ions ξ > 1) constitutes a thermodynamic instability which must be compensated by counterion condensation”. In fact, chain extension and/or disaggregation of aggregated chains may occur or change upon the variation of charge density, and the energetic instability effectively becomes a function of the thermodynamic state of the polyelectrolytic chain. The range of theoretical and experimental approaches has been, in particular, addressed to the problem of conformational transitions between two different states, provided they have different charge densities. For thermally induced, conformational transitions be-
12.2 Chain conformations of polysaccharides
731
tween states i and f of a polyelectrolyte, characterized by a set of ξi and ξf (i.e., bi and bf) values, polyelectrolyte theory predicts a simple relationship82 between the values of the melting temperatures (TM, the temperature of transition midpoint) and the logarithm of the ionic strength, I:
( ) = − 9.575F (ξ)
d TM−1
d (log I)
∆M H
where ∆ M H is the value of the enthalpy of transition (in J per mole of charged groups) determined calorimetrically. This linearity implies, indeed, that the enthalpy change is essentially due to non-ionic contributions and largely independent of I. The function F(ξ) depends on the charge density of both the final state (subscript f) and the initial state (subscript i), within the common condition that ξ f < ξ i, that is the final state is characterized by a smaller value of the charge density. The value of F(ξ) is given in the literature. This relation has been successfully applied first to the transition processes of DNA,83 polynucleotides,84 but also to many ionic polysaccharides (carrageenans,85 xanthan,86 succinoglycan,87) of great industrial interest. Accurate determination of the TM values of the polysaccharide as a function of the ionic strength is necessary. 12.2.6.4 Conformational calculations of charged polysaccharides The major problem for conformational calculations of ionic polysaccharides arises from the correct evaluation of the electrostatic potential energy due to the charged groups along the chain and to the all other ions in solutions. The interaction between the polyion charges and the counterions is formally non-conformational but it largely affects the distribution of the conformational states. Ionic polymers are often simplistically treated either in the approximation of full screening of the charged groups or in the approximation of rigid conformational states (regular rod-like polyelectrolyte models). A combination of the molecular polyelectrolyte theory82,83 with the methods of statistical mechanics can be used at least for the description of the chain expansion due to charges along the polysaccharide chain. The physical process of the proton dissociation of a (weak) polyacid is a good way to assess the conformational role of the polyelectrolytic interactions, since it is possible of tuning polyelectrolyte charge density on an otherwise constant chemical structure. An amylose chain, selectively oxidized on carbon 6 to produce a carboxylic (uronic) group, has proved to be a good example to test theoretical results.81 If the real semi-flexible chain of infinite length is replaced by a sequence of segments, the average end-to-end distance
r N
[12.2.8]
where N is the number of charges in the segment. The distance between charges fluctuates within the limits of the conformational flexibility of the chain, as calculated by the proper non-bonding inter-residue interactions. The probability function W'(r) of the end-to-end displacement r of a charged segment can be obtained by multiplying its a priori (non-ionic) probability W(r) with the Boltzmann
732
Ranieri Urbani and Attilio Cesàro
term involving the excess electrostatic free energy (Figure 12.2.15). The probability theory guarantees both that the components (repeating units) of the segment vectors be distributed in a Gaussian way along the chain segment, and that high molecular weight polymers be composed by a statistical sequence of those segments. Consequence of the above approximation is that the distance r between any two points of the chain (separated by a sufficiently large number of residues, n) does not depend on the specific sequence and values of conformational angles and energies, but only upon the average potential summed over the number of residues n. The calculation of the averaged (electrostatic) functions is reached in two steps. At the first, the proper flexibility of the Figure 12.2.15. Dependence of the probability distribuis evaluated either from tion function of a model semi-flexible chain, (a) un- polymer charged and charged (a'), on the end-to-end distance and conformational calculation or from suitable the respective total free energies (c and c'). The electro- models, then the mean value of each propstatic contribution (b) is also reported. erty is calculated through the averaging procedure described below. The computational procedure is the following: • the conformational energy surface of the uncharged polymer is evaluated by the standard methods the conformational analysis;65 • the end-to-end distribution distance Wn(r) for the (uncharged) polymer segments is determined by numerical Monte Carlo methods;64 • the dependence of the total (conformational) energy G(r) upon chain extension r is therefore estimated from the distribution of segment lengths; a Boltzmannian distribution is assumed. In most cases the distribution function is Gaussian (or approximately so) and the corresponding free energy function can be approximated by a simple parabolic equation (Figure 12.2.15). In this case, we assume a Hookean energy (which is correct at least for the region around the maximum of the distribution curve), so we have: G 0 W (r ) = A exp − , G(r ) = k r − r RT
(
)
[12.2.9]
where: r0 k
average segment length a constant which determines the flexibility of the chain
The ionic energy, that results from the process of charging the polymer groups, changes the probability of the end-to-end distance for the i-th segment, W'(r), to the probability of the average inter-charge separation distance , W(b), following the definition of equation [12.2.8] and [12.2.9].
12.2 Chain conformations of polysaccharides
733
The conformational (non-ionic) free energy, obtained from the radial distribution function for non-ionic chains by Monte Carlo calculations, was used in conjunction with the electrostatic free energy to calculate the actual distribution function of the charged chain segments. The resulting expansion justifies almost quantitatively in many cases the experimental thermodynamic properties (such as pKa, Hdil, etc.) and the dimensional properties (viscosity) of the ionic polysaccharides to which the approach has been applied. 12.2.7 CONCLUSIONS Only some aspects of the solvent perturbation on the conformational properties of carbohydrate polymers have been covered in this chapter. One of the major concerns has been to develop a description of these “solvent effects” starting with the complex conformational equilibria of simple sugars. In fact, only recently it has been fully appreciated the quantitative relationship between conformational population and physical properties, e.g. optical rotation. The chapter, however, does not give extensive references to the experimental determination of the polysaccharide shape and size in different solvents, but rather it attempts to focus on the molecular reasons of these perturbations. A digression is also made to include the electrostatic charges in polyelectrolytic polysaccharides, because of their diffusion and use and because of interesting variations occurring in these systems. Thus, provided that all the interactions are taken into account, the calculation of the energetic state of each conformation provides the quantitative definition of the chain dimensions. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
J.R. Brisson and J.P. Carver, Biochemistry, 22, 3671 (1983). R.C. Hughes and N. Sharon, Nature, 274, 637 (1978). D.A. Brant, Q. Rev. Biophys., 9, 527 (1976). B.A. Burton and D.A. Brant, Biopolymers, 22, 1769 (1983). G.S. Buliga and D.A. Brant, Int. J. Biol. Macromol., 9, 71 (1987). V. S. R. Rao, P. K. Qasba, P. V. Balaji and R. Chandrasekaran, Conformation of Carbohydrates, Harwood Academic Publ., Amsterdam, 1998, and references therein. R. H. Marchessault and Y. Deslandes, Carbohydr. Polymers, 1, 31 (1981). D. A. Brant, Carbohydr. Polymers, 2, 232 (1982). P.R. Straub and D.A. Brant, Biopolymers, 19, 639 (1980). A. Cesàro in Thermodynamic Data for Biochemistry and Biotechnology, H.J. Hinz (Ed.), Springer-Verlag, Berlin, 1986, pp. 177-207. Q. Liu and J.W. Brady, J. Phys.Chem. B, 101, 1317 (1997). J.W. Brady, Curr. Opin. Struct.Biol., 1, 711 (1991). Q. Liu and J.W. Brady, J. Am. Chem. Soc., 118, 12276 (1996). I. Tvaroška, Biopolymers, 21, 188 (1982). I. Tvaroška, Curr. Opin. Struct.Biol., 2, 661 (1991). K. Mazeau and I. Tvaroška, Carbohydr. Res., 225, 27 (1992). Perico, A., Mormino, M., Urbani, R., Cesàro, A., Tylianakis, E., Dais, P. and Brant, D. A., Phys. Chem. B, 103, 8162-8171(1999). K.D. Goebel, C.E. Harvie and D.A. Brant, Appl. Polym. Symp., 28, 671 (1976). M. Ragazzi, D.R. Ferro, B. Perly, G. Torri, B. Casu, P. Sinay, M. Petitou and J. Choay, Carbohydr. Res., 165, C1 (1987). P.E. Marszalek, A.F. Oberhauser, Y.-P. Pang and J.M. Fernandez, Nature, 396, 661 (1998). S.J. Angyal, Aust. J.Chem., 21, 2737 (1968). S.J. Angyal, Advan. Carbohyd. Chem. Biochem., 49, 35 (1991). D.A. Brant and M.D. Christ in Computer Modeling of Carbohydrate Molecules, A.D. French and J.W. Brady, Eds., ACS Symposium Series 430, ACS, Washington, DC, 1990, pp. 42-68. R. Harris, T.J. Rutherford, M.J. Milton and S.W. Homans, J. Biomol. NMR, 9, 47 (1997). M. Kadkhodaei and D.A. Brant, Macromolecules, 31, 1581 (1991).
734
Ranieri Urbani and Attilio Cesàro
I. Tvaroška and J. Gajdos, Carbohydr. Res., 271, 151 (1995). F.R. Taravel , K. Mazeau and I. Tvaroška, Biol. Res., 28, 723 (1995). J.L. Asensio and J. Jimenez-Barbero, Biopolymers, 35, 55 (1995). P. Dais, Advan. Carbohyd. Chem. Biochem., 51, 63 (1995). F. Cavalieri, E. Chiessi, M. Paci, G. Paradossi, A. Flaibani and A. Cesàro, Macromolecules, (submitted). I. Tvaroška and F.R. Taravel, J. Biomol. NMR, 2, 421 (1992). R.E. Wasylishen and T. Shaefer, Can. J. Chem., 51, 961 (1973). A.A. van Beuzekom, F.A.A.M. de Leeuw and C. Altona, Magn. Reson. Chem., 28, 888 (1990). I. Tvaroška, Carbohydr. Res., 206, 55 (1990). I. Tvaroška and F.R. Taravel, Carbohydr. Res., 221, 83 (1991). J.P. Carver, D. Mandel, S.W. Michnick, A. Imberty, J.W. Brady in Computer Modeling of Carbohydrate Molecules, A.D. French and J.W. Brady, Eds., ACS Symposium Series 430, ACS, Washington, DC, 1990, pp. 267-280. 37 D. Rees and D. Thom, J. Chem. Soc. Perkin Trans. II, 191 (1977). 38 E.S. Stevens, Carbohydr. Res., 244, 191 (1993). 39 D.H. Whiffen, Chem. Ind., 964 (1956). 40 J.H. Brewster, J. Am. Chem. Soc., 81, 5483 (1959). 41 D. A. Rees, J. Chem. Soc. B, 877 (1970). 42 E.S. Stevens and B.K. Sathyanarayana, J.Am.Chem.Soc., 111, 4149 (1989). 43 C.A. Duda and E.S. Stevens, Carbohydr. Res., 206, 347 (1990). 44 R. Urbani, A. Di Blas and A. Cesàro, Int. J. Biol. Macromol., 15, 24 (1993). 45. I. Tvaroška, S. Perez, O. Noble and F. Taravel, Biopolymers, 26, 1499 (1987). 46 C. Gouvion, K. Mazeau, A. Heyraud, F. Taravel, Carbohydr. Res., 261, 261 (1994). 47 R.A. Pierotti, Chem. Rev., 76, 717 (1976). 48 R.J. Abraham and E. Bretschneider in Internal Rotation in Molecules, W.J. Orville-Thomas, Ed., Academic Press, London, 1974, ch. 13. 49 D.L. Beveridge, M.M. Kelly and R.J. Radna, J. Am. Chem. Soc., 96, 3769 (1974). 50 M. Irisa, K. Nagayama and F. Hirata, Chem. Phys. Letters, 207, 430 (1993). 51 R.R. Birge, M.J. Sullivan and B.J. Kohler, J. Am. Chem.Soc., 98, 358 (1976). 52 I. Tvaroška and T. Kozar, J. Am. Chem. Soc., 102, 6929 (1980). 53 R. Urbani and A. Cesàro, Polymers, 32, 3013 (1991). 54 J.W. Brady, J. Am. Chem. Soc., 108, 8153 (1986). 55 C.B. Post, B.R. Brooks, M. Karplus, C.M. Dobson, P.J. Artymiuk, J.C. Cheetham and D.C. Phillips, J. Mol. Biol., 190, 455 (1986). 56 K. Ueda, A. Imamura and J.W. Brady, J. Phys. Chem., 102, 2749 (1998). 57 R. Grigera, Advan. Comp. Biol., 1, 203 (1994). 58 S.N. Ha, A. Giammona, M. Field and J.W. Brady, Carbohydr. Res., 180, 207 (1988). 59 C.L. Brooks, M. Karplus, B.M. Pettitt, Proteins: A Theoretical Perspective of Dynamics, Structure and Thermodynamics, Wiley Interscience, New York, 1988, vol. LXXI. 60 R. Schmidt, B. Teo, M. Karplus and J.W. Brady, J. Am. Chem. Soc., 118, 541 (1996). 61 Q. Liu, R. Schmidt, B. Teo, P.A. Karplus and J.W. Brady, J. Am. Chem. Soc., 119, 7851 (1997). 62 V.H. Tran and J.W. Brady, Biopolymers, 29, 977 (1990). 63 M. Karplus and J.N. Kushick, Macromolecules, 31, 1581 (1981). 64 R.C. Jordan, D.A. Brant and A. Cesàro, Biopolymers, 17, 2617 (1978). 65 P.J. Flory, Statistical Mechanics of Chain Molecules, Wiley-Interscience, New York, 1969. 66 D.W. Tanner and G.C. Berry, J. Polym. Sci., Polym. Phys. Edn., 12, 441 (1974). 67 R.C. Jordan and D.A. Brant, Macromolecules, 13, 491 (1980). 68 Y. Nakanishi, T. Norisuye, A. Teramoto and S. Kitamura, Macromolecules, 26, 4220 (1993). 69 T. Norisuye, Food Hydrocoll., 10, 109 (1996). 70 S.G. Ring, K.J. Anson and V.J. Morris, Macromolecules, 18, 182 (1985). 71 T. Hirao, T. Sato, A. Teramoto, T. Matsuo and H. Suga, Biopolymers, 29, 1867 (1990). 72 C. Sagui and T.A. Darden, Ann. Rev. Biophys. Biomol. Struct., 28, 155 (1999). 73 J.R. Ruggiero, R. Urbani and A. Cesàro, Int. J. Biol. Macromol., 17, 205 (1995). 74 O. Smidsrød and A. Haug, Biopolymers, 10, 1213 (1971). 75 G.S. Manning and S. Paoletti, in Industrial Polysaccharides, V. Crescenzi, I.C.M. Dea and S.S. Stivala, Eds., Gordon & Breach, N Y, 1987, pp. 305-324. 76 S. Paoletti, A. Cesàro, F. Delben, V. Crescenzi and R. Rizzo, in Microdomains in Polymer Solutions, P. Dubin Ed., Plenum Press, New York, (1985) pp 159-189. 26 27 28 29 30 31 32 33 34 35 36
12.2 Chain conformations of polysaccharides
77 78 79 80 81 82 83 84 85 86 87
735
H. Morawetz, Macromolecules in Solution, Interscience, New York, 1975, Ch. 7. M. Bohdanecký and J. Kovár, Viscosity of Polymer Solutions, Elsevier, Amsterdam, 1982, p. 108. M.W. Anthonsen, K.M. Vårum and O. Smidsrød, Carbohydr. Polymers, 22, 193 (1993). R. Geciova, A. Flaibani, F. Delben, G. Liut, R. Urbani and A. Cesàro, Macromol. Chem. Phys., 196, 2891 (1995). A. Cesàro, S. Paoletti, R. Urbani and J.C. Benegas, Int. J. Biol. Macromol., 11, 66 (1989). G.S. Manning, Acc. Chem. Res., 12, 443 (1979). G.S. Manning, Quart. Rev. Biophys., 11, 179 (1978). M.T. Record, C.F. Anderson and T.M. Lohman, Quart. Rev. Biophys., 11, 103 (1978). S. Paoletti, F. Delben, A. Cesàro and H. Grasdalen, Macromolecules, 18, 1834 (1985). S. Paoletti, A. Cesàro and F. Delben, Carbohydr. Res., 123, 173 (1983). T.V. Burova, I.A. Golubeva, N.V. Grinberg, A.Ya. Mashkevich, V.Ya. Grinberg, A.I.Usov, L. Navarini and A. Cesàro, Biopolymers, 39, 517 (1996).
13
Effect of Solvent on Chemical Reactions and Reactivity 13.1 SOLVENT EFFECTS ON CHEMICAL REACTIVITY Roland Schmid Technical University of Vienna Institute of Inorganic Chemistry, Vienna, Austria
13.1.1 INTRODUCTION About a century ago, it was discovered that the solvent can dramatically change the rate of chemical reactions.1 Since then, the generality and importance of solvent effects on chemical reactivity (rate constants or equilibrium constants) has been widely acknowledged. It can be said without much exaggeration that studying solvent effects is one of the most central topics of chemistry and remains ever-increasingly active. In the course of development, there are few topics in chemistry in which so many controversies and changes in interpretation have arisen as in the issue of characterizing solute-solvent interactions. In a historical context, two basic approaches to treating solvent effects may be distinguished: a phenomenological approach and a physical approach. The former may be subdivided further into the dielectric approach and the chemical approach. • Phenomenological approach Dielectric Chemical • Physical approach That what follows is not intended just to give an overview of existing ideas, but instead to filter seminal conceptions and to take up more fundamental ideas. It should be mentioned that solvent relaxation phenomena, i.e., dynamic solvent effects, are omitted. 13.1.2 THE DIELECTRIC APPROACH It has soon been found that solvent effects are particularly large for reactions in which charge is either developed or localized or vice versa, that is, disappearance of charge or spreading out of charge. In the framework of electrostatic considerations, which have been around since Berzelius, these observations led to the concept of solvation. Weak electrostatic interactions simply created a loose solvation shell around a solute molecule. It was in this climate of opinion that Hughes and Ingold2 presented the first satisfactory qualitative account of solvent effects on reactivity by the concept of activated complex solvation.
738
Roland Schmid
The first solvent property applied to correlate reactivity data was the static dielectric constant ε (also termed εs) in the form of dielectric functions as suggested from elementary electrostatic theories as those by Born (1/ε), Kirkwood (ε-1)/(2ε+1), ClausiusMosotti (ε-1)/(ε+2), and (ε-1)/(ε+1). A successful correlation is shown in Figure 13.1.1 for the rate of the SN2 reaction of p-nitrofluorobenzene with piperidine.3 The classical dielecFigure 13.1.1. Relationship between second-order rate constants at tric functions predict that reactiv50°C of the reaction of p-nitrofluorobenzene and piperidine and the ity changes level out for dielectric solvent dielectric properties [from ref. 21]. constants say above 30. For instance, the Kirkwood function has an upper limiting value of 0.5, with the value of 0.47 reached at ε = 25. The insert in Figure 13.1.1 illustrates this point. Therefore, since it has no limiting value, the log ε function may be preferred. A theoretical justification can be given in the framework of the dielectric saturation model of Block and Walker.4 Picturing the solvent as a homogeneous dielectric continuum means in essence that the solvent molecules have zero size and that the molecules cannot move. The most adequate physical realization would be a lattice of permanent point dipoles that can rotate but cannot translate. 13.1.3 THE CHEMICAL APPROACH Because of the often-observed inadequacies of the dielectric approach, that is, using the dielectric constant to order reactivity changes, the problem of correlating solvent effects was next tackled by the use of empirical solvent parameters measuring some solvent-sensitive physical property of a solute chosen as the model compound. Of these, spectral properties such as solvatochromic and NMR shifts have made a spectacular contribution. Other important scales are based on enthalpy data, with the best-known example being the donor number (DN) measuring solvent’s Lewis basicity. In the intervening years there is a proliferation of solvent scales that is really alarming. It was the merit particularly of Gutmann and his group to disentangle the great body of empirical parameters on the basis of the famous donor-acceptor concept or the coordination-chemical approach.5 This concept has its roots in the ideas of Lewis going back to 1923, with the terms donor and acceptor introduced by Sidgwick.6 In this framework, the two outstanding properties of a solvent are its donor (nucleophilic, basic, cation-solvating) and acceptor (electrophilic, acidic, anion-solvating) abilities, and solute-solvent interactions are considered as acid-base reactions in the Lewis’ sense. Actually, many empirical parameters can be lumped into two broad classes, as judged from the rough interrelationships found between various scales.7 The one class is more concerned with cation (or positive dipole’s end) solvation, with the most popular solvent basic-
13.1 Solvent effects on chemical reactivity
739
ity scales being the Gutmann DN, the Kamlet and Taft β, and the Koppel and Palm B. The other class is said to reflect anion (or negative dipole’s end) solvation. This latter class includes the famous scales π*, α, ET(30), Z, and last but not least, the acceptor number AN. Summed up: Cation (or positive dipole’s end) solvation DN • Gutmann • Kamlet and Taft β B (B*) • Koppel and Palm Anion (and negative dipole’s end) solvation Gutmann AN Dimroth and Reichardt ET(30) Kosower Z Kamlet and Taft α, π* These two sets of scales agree in their general trend, but are often at variance when values for any two particular solvents are taken. Some intercorrelations have been presented by Taft et al., e.g., the parameters ET, AN and Z can be written as linear functions of both α and π*.8 Originally, the values of ET and π* were conceived as microscopic polarity scales reflecting the “local” polarity of the solvent in the neighborhood of solutes (“effective” dielectric constant in contrast to the macroscopic one). In the framework of the donor-acceptor concept, however, they obtained an alternative meaning, based on the interrelationships found between various scales. Along these lines, the common solvents may be separated into six classes as follows. 1 nonpolar aliphatic solvents 2 protics or protogenetic solvents (at least one hydrogen atom is bonded to oxygen) 3 aromatic solvents 4 (poly)halogenated solvents 5 (perhaps) amines 6 select (or “normal” according to Abraham) solvents defined as non-protonic, non-chlorinated, aliphatic solvents with a single dominant bond dipole. A case study is the plot of AN versus ET shown in Figure 13.1.2. While there is a quite good correspondence for the select solvents (and likely for the nonpolar aliphatic solvents), the other classes are considerably off-line.9 This behavior may be interpreted in terms of the operation of different solvation mechanisms such as electronic polarizability, dipole density, and/or hydrogen-bonding (HB) ability. For instance, the Figure 13.1.2. Relationship between the ET(30) values and the ac- main physical difference between ceptor number [from ref. 21]. Triangles: protic solvents, squares: ar- π* and ET(30), in the absence of
• • • •
omatic and chlorinated solvents.
740
Roland Schmid
HB interactions, is claimed to lie in different responses to solvent polarizability effects. Likewise, in the relationship between the π* scale and the reaction field functions of the refractive index (whose square is called the optical dielectric constant e∞) and the dielectric constant, the aromatic and the halogenated solvents were found to constitute special cases.10 This feature is also reflected by the polarizability correction term in eq. [13.1.2] below. For the select solvents, the various “polarity” scales are more or less equivalent. A recent account of the various scales has been given by Marcus,11 and in particular of π* by Laurence et al.,12 and of ET by Reichardt.13 However, solvation is not the only mode of action taken by the solvent on chemical reactivity. Since chemical reactions typically are accompanied by changes in volume, even reactions with no alteration of charge distribution are sensitive to the solvent. The solvent dependence of a reaction where both reactants and products are neutral species (“neutral” pathway) is often treated in terms of either of two solvent properties. The one is the cohesive energy density ε c or cohesive pressure measuring the total molecular cohesion per unit volume, ε c = (∆Hv − RT ) / V
[13.1.1]
where: ∆Hv V
molar enthalpy of vaporization molar liquid volume
The square root of ε c is termed the Hildebrand solubility parameter δH, which is the solvent property that measures the work necessary to separate the solvent molecules (disrupt and reorganize solvent/solvent interactions) to create a suitably sized cavity for the solute. The other quantity in use is the internal pressure Pi which is a measure of the change in internal energy U of the solvent during a small isothermal expansion, Pi = (∂U/∂V)T. Interesting, and long-known, is the fact that for the highly dipolar and particular for the protic solvents, values of ε c are far in excess of Pi.14 This is interpreted to mean that a small expansion does not disrupt all of the intermolecular interactions associated with the liquid state. It has been suggested that Pi does not detect hydrogen bonding but only weaker interactions. At first, solvent effects on reactivity were studied in terms of some particular solvent parameter. Later on, more sophisticated methods via multiparameter equations were applied such as15 XYZ = XYZ 0 + s(π * +dδ) + aα + bβ + hδ H
[13.1.2]
where XYZ0, s, a, b, and h are solvent-independent coefficients characteristic of the process and indicative of its sensitivity to the accompanying solvent properties. Further, δ is a polarizability correction term equal to 0.0 for nonchlorinated aliphatic solvents, 0.5 for polychlorinated aliphatics, and 1.0 for aromatic solvents. The other parameters have been given above, viz. π*, α, β, and δH are indices of solvent dipolarity/polarizability, Lewis acidity, Lewis basicity, and cavity formation energy, respectively. For the latter, instead of δ H, δH2 should be preferred as suggested from regular solution theory.16 Let us just mention two applications of the linear solvation energy relationship (LSER). The one concerns the solvolysis of tertiary butyl-halides17 log k(ButCl) = -14.60 + 5.10π* + 4.17α + 0.73β + 0.0048δ2H
13.1 Solvent effects on chemical reactivity
741
n = 21, r = 0.9973, s = 0.242 and the other deals with the transfer of tetramethylammonium iodide through solvents with methanol as the reference solvent,16 ∆G tr0 = 10.9 − 15.6π∗−6.2α + 0.022δ H n = 18, r = 0.997, s = 0.3 where: n r s
number of solvents correlation coefficient standard deviation
Figure 13.1.3. Relative orbital energy levels for Cu2+ in square planar, tetragonal, and octahedral environments [adapted from ref. 18].
DN = 195.5 - 0.0102ν 0
We will not finish this section without noting that there are also metal complexes available functioning as color indicators of the coordination properties of solvents.18 Thus, Cu(tmen)(acac) ClO4, where tmen = N,N,N’,N’tetramethylethylenediamine and acac = acetylacetonate, can be used as a Lewis-basicity indicator, and Fe(phen)2(CN)2, where phen = 1,10-phenanthroline, as a Lewisacidity indicator. The physical origin of the underlying color changes is sketched in the Figures 13.1.3 and 13.1.4, as modified from ref. 18. These color indicators can be used as a quick method for assessing the coordination properties of solvents, solvent mixtures, and solutes not yet measured. This is very expedient since some classical parameters, particularly the donor numbers, are arduously amenable. The following equation [13.1.3]
n = 12, r = 0.990, s = 1.37 correlates the wave numbers ν 0 (in cm-1) of the visible band of Cu(tmen)(acac)+ and the solvent donor numbers. Similarly, the acceptor numbers are expressed as a function of the wave numbers of the long wavelength absorption of Fe(phen)2(CN)2,
742
Roland Schmid
AN = -133.8 - 0.00933ν 0
[13.1.4]
n = 12, r = 0.980, s = 4.58 13.1.4 DIELECTRIC VS. CHEMICAL APPROACH Although the success of the empirical solvent parameters has tended to downgrade the usefulness of the dielectric approach, there are correlations that have succeeded as exemplified by Figure 13.1.1. It is commonly held that the empirical solvent parameters are superior to dielectric estimates because they are sensitive to short-range phenomena not captured in dielectric measurements. This statement may not be generalized, however, since it depends strongly on the chemical reaction investigated and the choice of solvents. For instance, the rate of the Menschutkin reaction between tripropylamine and methyl Figure 13.1.4. Simplified orbital scheme for the charge transfer tran- iodide in select solvents correlates sition in Fe(phen)2(CN)2 varying with solvation. The diagram, not better with the log ε function than drawn to scale, is adjusted so that π* is constant [adapted from ref. with the solvent acceptor num18]. ber.19 Thus the solution chemists were puzzled for a long time over the question about when and when not the dielectric approach is adequate. In the meantime, this issue has been unraveled, in that dielectric estimates have no relevance to the solvation of positive (partial) charge. Thus, there is no relationship between the free energies of transfer for cations and the dielectric constant.7 Likewise, note the solvent-dependence of the solubilities of sodium chloride (Table 13.1.1) taken from Mayer’s work.19 For instance, the pairs of solvents H2O/PC and DMF/MeCN have similar ε′s but vastly different abilities to dissolve NaCl. In similar terms, the inclusion of a donor number term improves somewhat the correlation in Figure 13.1.1, as may be seen in Figure 13.1.5. This would suggest that the hydrogen of piperidine in the activated complex becomes acidic and is attacked by the strong donor solvents DMF, DMA and DMSO (Scheme 13.1.1).
13.1 Solvent effects on chemical reactivity
743
Table 13.1.1. Standard free energies of solution of sodium chloride in various solvents at 25°C. Data of ∆G0solv are from reference 19 ∆G0solv , kJ mol-1
εs
DN
AN
H 2O
-9.0
78.4
18
55
FA
-0.4
109
24
40
NMF
+3.8
182
27
32
MeOH
+14.1
32.6
19
41
DMSO
+14.9
46.7
30
19
DMF
+26.8
36.7
26
16
PC
+44.7
65
15
18
MeCN
+46.8
36
14
19
Solvent
On the other hand, if negative charge is solvated in the absence of positive charge capable of solvation, the dielectric constant is often a pretty good guide to ranking changes in reactivity. As a consequence, the dielectric approach has still its place in organic chemistry while it is doomed to complete failure in inorganic reactions where typically cation solvation is involved. For select solvents, ultimately, the dielectric constant is related to the anion-solvating properties of solvents according to the regression equation4 Figure 13.1.5. Correlation diagram for the same reaction as in Figure 13.1.1 [from ref. 21].
log ε = 0.32 + 0.073 (ANE)
[13.1.5]
n = 31, r = 0.950, s = 0.129 where: ANE
ET-based acceptor numbers, ANE = - 40.52 + 1.29 ET
This equation works also quite well for the aromatics and the halogenated solvents, but it does not hold for the protic solvents. For these, the predicted values of the dielectric constants are orders of magnitude too large, revealing how poorly the associates are dissoci-
744
Roland Schmid
ated by the macroscopically attainable fields. A correlation similar to [13.1.5] has been proposed20 between the gas phase dipole moment and π* µ(D) = 4.3π * −01 .
[13.1.6]
n = 28, r = 0.972, s = 0.3 Along these lines the dielectric and the chemical approach are brought under one roof.4,21 The statement, however, that the terms “good acceptor solvent” and “highly polar solvent” may be used synonymously would seem, though true, to be provocative. 13.1.5 CONCEPTUAL PROBLEMS WITH EMPIRICAL SOLVENT PARAMETERS A highly suspect feature behind the concept of empirical solvent parameters lies in the interpretation of the results in that condensed phase matters are considered from the narrow viewpoint of the solute only with the solvent’s viewpoint notoriously neglected. However, the solute is actually probing the overall action of the solvent, comprising two modes of interactions: solute-solvent (solvation) and solvent-solvent (restructuring) effects of unknown relative contribution. Traditionally, it is held that solvent structure only assumes importance when highly structured solvents, such as water, are involved.22 But this view increasingly turns out to be erroneous. In fact, ignoring solvent-solvent effects, even in aprotic solvents, can lead to wrong conclusions as follows. In the donor-acceptor approach, solutes and solvents are divided into donors and acceptors. Accordingly, correlations found between some property and the solvent donor (acceptor) ability are commonly thought to indicate that positive (negative) charge is involved. In the case of solvent donor effects this statement is actually valid. We are unaware, in fact, of any exception to the rule saying: “Increase in reaction rate with increasing solvent DN implies that positive charge is developed or localized and vice versa”.21 In contrast, correlations with the acceptor number or related scales do not simply point to anion solvation, though this view is commonly held. An example for such type of reasoning concerns the medium effect on the intervalence transition (IT) energy within a certain binuclear, mixed-valence, 5+ cation.23 As the salt effect was found to vary with the solvent AN, anion, that is counterion, solvation in ion pairs was invoked to control the IT energy. A conceptual problem becomes obvious by the at first glance astonishing result that the reduction entropies of essentially non-donor cationic redox couples such as Ru(NH3)63+/2+ are correlated with the solvent AN.24 These authors interpreted this solvent dependence as reflecting changes in solvent-solvent rather than solvent-ligand interactions. That the acceptor number might be related to solvent structure is easy to understand since all solvents of high AN always are good donors (but not vice versa!) and therefore tend to be increasingly self-associated.21 There is since growing evidence that the solvent’s AN and related scales represent ambiguous solvent properties including solvent structural effects instead of measuring anion solvation in an isolated manner. Thus, correlations between Gibbs energies of cation transfer from water to organic solvents and the solvent DN are improved by the inclusion of a term in ET (or a combination of α and π*).25 Consequently Marcus et al. rightly recognized that “ET does not account exclusively for the electron pair acceptance capacity of solvents”.26 In more recent work27 a direct relationship has been
13.1 Solvent effects on chemical reactivity
745
found between the solvent reorganizational energy accompanying the excitation of ruthenium(II) cyano complexes and the solvent acceptor number. In the basicity scales, on the other hand, complications by solvent structure are not as obvious. If restriction is to aprotic solvents, as is usual, various scales though obtained under different conditions, are roughly equivalent.21,4 There is for instance a remarkably good relationship between the DN scale (obtained in dilute dichloromethane solution, i.e., with medium effects largely excluded) and the B scale (derived from measurements performed with 0.4 M solutions of MeOD in the various solvents4). The relationship between β and B, on the other hand, separates out into families of solvents.20 Donor measures for protic solvents eventually are hard to assess and often are at considerable variance from one scale to another.28,29 To rationalize the discrepancies, the concept of “bulk donicity” was introduced7 but with little success. Instead, the consideration of structure changes accompanying solvation might better help tackle the problem. Another suspect feature of the common method of interpreting solvent-reactivity correlations is that it is notoriously done in enthalpic (electronic, bond-strength etc.) terms. This way of thinking goes back to the Hughes-Ingold theory. However, many reactions in solution are not controlled by enthalpy changes but instead by entropy. Famous examples are the class of Menschutkin reactions and the solvolysis of t-butyl halides. Both these reaction types are characterized by the development of halide ions in the transition state, which can be considered as ion-pair like. In view of this, rate acceleration observed in good acceptor (or, alternatively, highly polar) solvents seems readily explainable in terms of solvation of the developing halide ion with concomitant carbon-halogen bond weakening. If this is true, most positive activation entropies and highest activation enthalpies should be expected to occur for the poor acceptor solvents. However, a temperature dependence study of the t-butyl halide solvolysis revealed just the opposite.17 This intriguing feature points to changes in solvent structure as a major determinant of the reaction rate with the ionic transition state acting as a structure maker in poor acceptor solvents, and as a structure maker in the protic solvents. It is rather ironic that the expected increase in rate with increasing solvent acceptor strength is a result of the coincidence of two, from the traditional point of view, unorthodox facts: (i) The intrinsic solvation of the developing halide ion disfavors the reaction via the entropy term. However, (ii), the extent of that solvation is greater in the poorly coordinating solvents (providing they are polarizable such as the aromatic solvents and the polyhalogenated hydrocarbons). In keeping with this interpretation, the Menschutkin reaction between benzyl bromide and pyridine is characterized by more negative activation volumina (i.e., stronger contraction of the reacting system in going to the activated complex) in poor acceptor (but polarizable) solvents.30 The importance is evident of studying temperature or pressure dependencies of solvent effects on rate in order to arrive at a physically meaningful interpretation of the correlations. Another problem with the interpretation of multiparameter equations such as [13.1.2] arises since some of the parameters used are not fully independent of one another. As to this, the trend between π* and α has already been mentioned. Similarly, the δH parameter displays some connection to the polarity indices.31,32 Virtually, the various parameters feature just different blends of more fundamental intermolecular forces (see below). Because of this, the interpretations of empirical solvent-reactivity correlations are often based more on intuition or preconceived opinion than on physically defined interaction mechanisms. As it
746
Roland Schmid
turns out, polar solvation has traditionally been overemphasized relative to nonpolar solvation (dispersion and induction), which is appreciable even in polar solvents. The conceptual problems of the empirical solvent parameters summarized: • The solvent acceptor number and other “polarity” scales include appreciable, perhaps predominant, contributions from solvent structure changes rather than merely measuring anion solvation. • Care is urged in a rash interpretation of solvent-reactivity correlations in enthalpic terms, instead of entropic, before temperature-dependence data are available. Actually, free energy alone masks the underlying physics and fails to provide predictive power for more complex situations. • Unfortunately, the parameters used in LSER’s sometimes tend to be roughly related to one another, featuring just different blends of more fundamental intermolecular forces. Not seldom, fortuitous cancellations make molecular behavior in liquids seemingly simple (see below). Further progress would be gained if the various interaction modes could be separated by means of molecular models. This scheme is in fact taking shape in current years giving rise to a new era of tackling solvent effects as follows. 13.1.6 THE PHYSICAL APPROACH There was a saying that the nineteenth century was the era of the gaseous state, the twentieth century of the solid state, and that perhaps by the twenty-first century we may understand something about liquids.33 Fortunately, this view is unduly pessimistic, since theories of the liquid state have actively been making breath-taking progress. In the meantime, not only equations of state of simple liquids, that is in the absence of specific solvent-solvent interactions,34-36 but also calculations of simple forms of intermolecular interactions are becoming available. On this basis, a novel approach to treating solvent effects is emerging, which we may call the physical approach. This way of description is capable of significantly changing the traditionally accepted methods of research in chemistry and ultimately will lay the foundations of the understanding of chemical events from first principles. A guiding principle of these theories is recognition of the importance of packing effects in liquids. It is now well-established that short-ranged repulsive forces implicit in the packing of hard objects, such as spheres or dumbbells, largely determine the structural and dynamic properties of liquids.37 It may be noted in this context that the roots of the idea of repulsive forces reach back to Newton who argued that an elastic fluid must be constituted of small particles or atoms of matter, which repel each other by a force increasing in proportion as their distance diminishes. Since this idea stimulated Dalton, we can say that the very existence of liquids helped to pave the way for formulating modern atomic theory with Newton granting the position of its “grandfather”.38 Since the venerable view of van der Waals, an intermolecular potential composed of repulsive and attractive contributions is a fundamental ingredient of modern theories of the liquid state. While the attractive interaction potential is not precisely known, the repulsive part, because of changing sharply with distance, is treatable by a common formalism in terms of the packing density η, that is the fraction of space occupied by the liquid molecules. The packing fraction is a key parameter in liquid state theories and is in turn related in a simple way to the hard sphere (HS) diameter σ in a spherical representation of the molecules comprising the fluid:
13.1 Solvent effects on chemical reactivity
η = πρσ 3 / 6 = ρVHS
747
[13.1.7]
where: η ρ σ VHS
packing density number density N/V= number of particles per unit volume HS diameter HS volume
For the determination of σ (and hence η), the most direct method is arguably that based on inert gas solubility data.39,40 However, in view of the arduousness involved and the uncertainties in both the extrapolation procedure and the experimental solubilities, it is natural to look out for alternatives. From the various suggestions,41,42 a convenient way is to adjust σ such that the computed value of some selected thermodynamic quantity, related to σ, is consistent with experiment. The hitherto likely best method43 is the following: To diminish effects of attraction, the property chosen should probe primarily repulsive forces rather than attractions. Since the low compressibility of the condensed phase is due to short-range repulsive forces, the isothermal compressibility βT = -(1/V)(∂V/∂P)T might be a suitable candidate, in the framework of the generalized van der Waals (vdW) equation of state βT (RT / V)Q r = 1
[13.1.8]
where Qr is the density derivative of the compressibility factor of a suitable reference system. In the work referred to, the reference system adopted is that of polar-polarizable spheres in a mean field, 5η 2 − 2η 3 Q r = 2 − 1 − 2Z µ 4 (1 − η)
[13.1.9]
where Zµ = compressibility factor due to dipole-dipole forces,43 which is important only for a few solvents such as MeCN and MeNO2. The HS diameters so determined are found to be in excellent agreement with those derived from inert gas solubilities. It may be noted that the method of Ben-Amotz and Willis,44 also based on βT, uses the nonpolar HS liquid as the reference and, therefore, is applicable only to liquids of weak dipole-dipole forces. Of course, as the reference potential approaches that of the real liquid, the HS diameter of the reference liquid should more closely approximate the actual hard-core length. Finally, because of its popularity, an older method should be mentioned that relies on the isobaric expansibility αp as the probe, but this method is inadequate for polar liquids. It turns out that solvent expansibility is appreciably determined by attractions. Some values of η and σ are shown in Table 13.1.2 including the two extreme cases. Actually, water and n-hexadecane have the lowest and highest packing density, respectively, of the common solvents. As is seen, there is an appreciable free volume, which may be expressed by the volume fraction η − η 0 , where η 0 is the maximum value of η calculated for the face-centered cubic packing of HS molecules where all molecules are in contact with each other is η 0 = π 2 / 6 = 0.74. Thus, 1 - η 0 corresponds to the minimum of unoccupied volume. Since η typically is around 0.5, about a quarter of the total liquid volume is empty enabling solvent molecules to change their coordinates and hence local density fluctuations to occur.
748
Roland Schmid
Packing density η minimum ≈0 perfect gas
0.4 - 0.6 liquids
maximum 0.74 cubic close packed
Table 13.1.2. Packing densities in some liquids
These considerations ultimately offer the basis of a genuinely Liquid free volume, % molecular theory of solvent effects, η as compared to a mean-field theory. 0.41 59 H2O Thus, packing and repacking effects 0.50 50 n-C6 accompanying chemical reactions have to be taken into account for any Benzene 0.51 49 realistic view of the solvent’s role MeOH 0.41 59 played in chemical reactions to be at0.47 53 Et2O tained. The well-known cavity formation energy is the work done 0.62 38 n-C16 against intermolecular repulsions. At present, this energy is calculated for spherical cavities by the Boublik-Mansoori-Carnahan-Starling-Leland (BMCSL) mixed HS equation of state45,46 ∆G rep RT
=2
ηd ( −d 2 + d + 1) ηd 3 ηd 2 +3 +3 + ( −2d 3 + 3d 2 − 1) ln( 1− η) 3 2 ( 1− η) ( 1− η) ( 1− η)
[13.1.10]
where d = σ 0 / σ is the relative solute size (σ0 is the solute HS diameter, and σ is the solvent diameter). Quite recently, a modification of this equation has been suggested for high liquid densities and large solute sizes.47 Notice that under isochoric conditions the free energy of cavity formation is a totally entropic quantity. Ravi et al48 have carried out an analysis of a model dissociation reaction (Br2 → 2Br) dissolved in a Lennard-Jones solvent (Ne, Ar, and Xe). That and the previous work49 demonstrated that solvent structure contributes significantly to both chemical reaction volumes (which are defined as the pressure derivatives of reaction free energies) and free energies, even in systems containing no electrostatic or dispersion long-ranged solvent-solute interactions. Let us now turn to the more difficult case of intermolecular attractive forces. These may be subdivided into: Long-ranged or unspecific • dispersion • induction • dipole-dipole • higher multipole Short-ranged or specific • electron overlap (charge transfer) • H-bonding For the first three ones (dispersion, induction, dipole-dipole forces) adequate calculations are just around the corner. Let us give some definitions.
13.1 Solvent effects on chemical reactivity
749
Dispersion forces are the result of the dipolar interactions between the virtually excited dipole moments of the solute and the solvent, resulting in a nonzero molecular polarizability. Although the average of every induced dipole is zero, the average of the product of two induced dipoles is nonzero (Figure 13.1.6). Induction forces are caused by the interaction of the permanent solvent dipole with the solvent dipoles induced by the solute and solvent field (Figure 13.1.7). Sometimes it is stated that dispersion is a quantum mechanical effect and induction is not. Thus, Figure 13.1.6 some clarifying comments are at place here. From the general viewpoint, all effects including polarizability are quantum mechanical in their origin because the polarizability of atoms and molecules is a quantum mechanical quantity and can be assessed only in the framework of quantum mechanics. However, once calculated, one can think of polarizability in classical terms representing a quantum molecular object as a classical oscillator with the mass equal to the polarizability, which is not specified in the classical framework. This is definitely wrong from a fundamental viewpoint, but, as it usually appears with harmonic models, a quantum mechanical calculation and such a primitive classical model give basically the same results about the induction Figure 13.1.7 matter. Now, if we implement this classical model, we would easily come up with the induction potential. However, the dispersion interaction will be absent. The point is that to get dispersions, one needs to switch back to the quantum mechanical description where both inductions and dispersions naturally appear. Thus the quantum oscillator may be used resulting in both types of potentials.50 If in the same procedure one switches to the classical limit (which is equivalent to putting the Plank constant zero) one would get only inductions. The calculation of the dispersive solvation energy is based on perturbation theories following the Chandler-Andersen-Weeks51 or Barker-Henderson52 formalisms, in which long-range attractive interactions are treated as perturbations to the properties of a hard body reference system. Essentially, perturbative theories of fluids are a modern version of van der Waals theory.53 In the papers reviewed here, the Barker-Henderson approach was utilized with the following input parameters: Lennard-Jones (LJ) energies for the solvent, for which reliable values are now available, the HS diameters of solvent and solute, the solvent polarizability, and the ionization potentials of solute and solvent. A weak point is that in order to get the solute-solvent LJ parameters from the solute and solvent components, some combining rule has to be utilized. However, the commonly applied combining rules appear to be adequate only if solute and solvent molecules are similar in size. For the case of particles appreciably different both in LJ energy and size, the suggestion has been made to use an empirical scaling by introducing empirical coefficients so as to obtain agreement be-
750
Roland Schmid
tween calculated and experimental solvation energies for selected inert gases and nonpolar large solutes.54 In the paper referred to,54 the relevance of the theoretical considerations has been tested on experimental solvation free energies of nitromethane as the solute in select solvents. The total solvation energy is a competition of the positive cavity formation energy and the negative solvation energy of dispersion and dipolar forces, ∆G = ∆G cav + ∆G disp + ∆G dipolar
[13.1.11]
where the dipolar term includes permanent and induced dipole interactions. The nitromethane molecule is represented by the parameters of the HS diameter σ = 4.36 Å, the gas-phase dipole moment µ = 3.57 D, the polarizability α = 4.95 Å , and the LJ energy εLJ/k=391K. Further, the solvent is modeled by spherical hard molecules of spherical polarizability, centered dipole moment, and central dispersion potential. To calculate the dipolar response, the Padé approximation was applied for the chemical potential of solvation in the dipolar liquid and then extended to a polarizable fluid according to the procedure of Wertheim. The basic idea of the Wertheim theory is to replace the polarizable liquid of coupled induced dipoles with a fictitious fluid with an effective dipole moment calculated in a self-consistent manner. Further, the Padé form is a simple analytical way to describe the dependence of the dipolar response on solvent polarity, solvent density, and solute/solvent size ratio. The theory/experiment agreement of the net solvation free energy is acceptable as seen in Table 13.1.3 where solvent ordering is according to the dielectric constant. Note that the contribution of dispersion forces is considerable even in strongly polar solvents. Table 13.1.3. Thermodynamic potentials (kJ/mol) of dissolution of nitromethane at 25°C. Data are from reference 54 εs
∆Gcav
∆Gdisp
∆Gdipolar
∆G(calc)
∆G(exp)
n-C6
1.9
23.0
-32.9
-2.2
-12.1
-12.1
c-C6
2.0
28.1
-38.1
-2.8
-12.7
-12.0
Et3N
2.4
24.2
-33.5
-2.9
-12.2
-15.2
Et2O
4.2
22.6
-33.4
-5.8
-16.5
-17.5
EtOAc
6.0
28.0
-38.5
-9.7
-20.1
-21.2
THF
7.5
32.5
-41.5
-12.2
-21.1
-21.3
c-hexanone
15.5
35.1
-39.8
-17.9
-22.6
-21.8
2-butanone
17.9
28.3
-33.9
-18.9
-24.5
-21.9
Acetone
20.7
27.8
-31.2
-22.1
-25.5
-22.5
DMF
36.7
38.1
-32.1
-28.5
-22.5
-23.7
DMSO
46.7
41.9
-31.1
-31.0
-20.2
-23.6
Solvent
With an adequate treating of simple forms of intermolecular attractions becoming available, there is currently great interest to making a connection between the empirical scales and solvation theory. Of course, the large, and reliable, experimental databases on
13.1 Solvent effects on chemical reactivity
751
empirical parameters are highly attractive for theoreticians for testing their computational models and improving their predictive power. At present, the solvatochromic scales are under considerable scrutiny. Thus, in a recent thermodynamic analysis, Matyushov et al.55 analyzed the two very popular polarity scales, ET(30) and π*, based on the solvent-induced shift of electronic absorption transitions (Scheme 13.1.2) Solvatochromism has its origin in changes in both dipole moment and polarizability of the dye upon electronic excitation provoking differential solvation of the ground and excited states. The dipole moment, µe, of the excited state can be either smaller or larger than the ground state value µg. In the former case one speaks about a negatively solvatochromic dye such as betaine-30, whereas Table 13.1.4. Dye properties used in the the π* dye 4-nitroanisole is positively solvatochromic. Thus, polar calculations. Data are from reference 55 solvent molecules produce a red Molecular parameter Betaine-30 4-Nitroanisole shift (lower energy) in the former and a blue shift (higher energy) in Vacuum energy gap (eV) 1.62 4.49 the latter. On the other hand, 6.4 4.5 R0 (Å) polarizability arguably always increases upon excitation. Disper68 15 α g (Å) sion interactions, therefore, would 61 6 ∆α (Å) produce a red shift proportional to 14.8 4.7 mg (D) ∆α = αe - αg of the dye. In other words, the excited state is stabi6.2 12.9 me (D) lized through strengthening of -8.6 +8.2 ∆µ (D) dispersive coupling. Finally, the relative contributions of dispersion and dipolar interactions will depend on the size of the dye molecules with dispersive forces becoming increasingly important the larger the solute. Along these lines, the dye properties entering the calculations are given in Table 13.1.4. The purpose of the analysis was to determine how well the description in terms of “trivial” dipolar and dispersion forces can reproduce the solvent dependence of the absorption energies (and thereby, by difference to experiment, expose the magnitude of specific forces), [13.1.12] £ωabs = ∆ + ∆E rep + ∆E disp + ∆E dipolar + ∆E ss where
£ω abs ∆ ∆Erep ∆Edisp ∆Edipolar ∆Ess
absorption energy vacuum energy gap shift due to repulsion solute-solvent interactions (taken to be zero) shift due to dispersion interactions shift due to dipolar forces of permanent and induced dipoles solvent reorganization energy
For the detailed and arduous calculation procedure, the reader may consult the paper cited. Here, let us just make a few general comments. The solvent influence on intramolecular optical excitation is treated by implementing the perturbation expansion
752
Roland Schmid
over the solute-solvent attractions. The reference system for the perturbation expansion is chosen to be the HS liquid with the imbedded hard core of the solute. It should be noted for clarity that ∆Edisp and ∆Edipolar are additive due to different symmetries: dispersion force is non-directed (i.e., is a scalar quantity), and dipolar force is directed (i.e., is a vector). In other words, the attractive intermolecular potential can be split into a radial and an angle-dependent part. In modeling the solvent action on the optical excitation, the solute-solvent interactions have to be dissected into electronic (inertialess) (dispersion, induction, charge-transfer) and molecular (inertial) (molecular orientations, molecular packing) modes. The idea is that the inertial modes are frozen on the time scale of the electronic transition. This is the Franck-Condon principle with such types of transitions called vertical transitions. Thus the excited solute is to be considered as a Frank-Condon state, which is equilibrated only to the electronic modes, whereas the inertial modes remain equilibrated to the ground state. According to the frozen solvent configuration, the dipolar contribution is represented as the sum of two terms corresponding to the two separate time scales of the solvent, (i) the variation in the solvation potential due to the fast electronic degrees of freedom, and (ii) the work needed to change the solute permanent dipole moment to the excited state value in a frozen solvent field. The latter is calculated for accommodating the solute ground state in the solvent given by orientations and local packing of the permanent solvent dipoles. Finally, the solvent reorganization energy, which is the difference of the average solvent-solvent interaction energy in going from the ground state to the excited state, is extracted by treating the variation with temperature of the absorption energy. Unfortunately, experimental thermochromic coefficients are available for a few solvents only. The following results of the calculations are relevant. While the contributions of dispersions and inductions are comparable in the π* scale, inductions are overshadowed in the ET(30) values. Both effects reinforce each other in π*, producing the well-known red shift. For the ET(30) scale, the effects due to dispersion and dipolar solvation have opposite signs making the red shift for nonpolar solvents switch to the blue for polar solvents. Furthermore, there is overall reasonable agreement between theory and experiment for both dyes, as far as the nonpolar and select solvents are concerned, but there are also discrepant solvent classes pointing to other kinds of solute-solvent interactions not accounted for in the model. Thus, the predicted ET(30) values for protic solvents are uniformly too low, revealing a decrease in H-bonding interactions of the excited state with lowered dipole moment. Another intriguing observation is that the calculated π* values of the aromatic and chlorinated solvents are throughout too high (in contrast to the ET(30) case). Clearly, these deviations, reminiscent of the shape of the plots such as Figure 13.1.2, may not be explained in terms of polarizability as traditionally done (see above), since this solvent property has been adequately accommodated in the present model via the induction potential. Instead, the theory/experiment discord may be rationalized in either of two ways. One reason for the additional solvating force can be sought in terms of solute-solvent π overlap resulting in exciplex formation. Charge-transfer (CT) interactions are increased between the solvent and the more delocalized excited state.55 The alternative, and arguably more reasonable, view considers the quadrupole moment which Figure 13.1.8
13.1 Solvent effects on chemical reactivity
753
is substantial for both solvent classes.56 Recently, this latter explanation in terms of dipole-quadrupole interactions is favored (Fig 13.1.8). It is well-known that the interaction energy falls off more rapidly the higher the order of the multipole. Thus, for the interaction of an n-pole with an m-pole, the potential energy varies with distance as E ∝ 1/(rn+m+1). The reason for the faster decrease is that the array of charges seems to blend into neutrality more rapidly with distance the higher the number of individual charges contributing to the multipole. Consequently, quadrupolar forces die off faster than dipolar forces. It has been calculated that small solute dipoles are even more effectively solvated by solvent quadrupoles than by solvent dipoles.57 In these terms it is understandable that quadrupolar contributions are more important in the π* than in the ET(30) scale. Similarly, triethylphosphine oxide, the probe solute of the acceptor number scale, is much smaller than betaine(30) and thus might be more sensitive to quadrupolar solvation. Thus, at long last, the shape of Figure 13.1.2 and similar ones seems rationalized. Note by the way that the quadrupole and CT mechanisms reflect, respectively, inertial and inertialess solvation pathways, and hence could be distinguished by a comparative analysis of absorption and fluorescence shifts (Stokes shift analysis). However, for 4-nitroanisole fluorescence data are not available. Reverting once more to the thermodynamic analysis of the π* and ET(30) scales referred to above, it should be mentioned that there are also other theoretical treatments of the solvatochromism of betaine(30). Actually, in a very recent computer simulation,58 the large polarizability change ∆α (nearly 2-fold, see Table 13.1.4) upon the excitation of betaine(30) has been (correctly) questioned. (According to a rule of thumb, the increase in polarizability upon excitation is proportional to the ground state polarizability, on the order ∆α ≈ 0.25α g .50) Unfortunately, Matyushov et al.55 derived this high value of ∆α = 61 Å3 from an analysis of experimental absorption energies based on aromatic, instead of alkane, solvents as nonpolar reference solvents. A lower value of ∆α would diminish the importance of dispersion interactions. Further theoretical and computational studies of betaine(30) of the ET(30) scale are reviewed by Mente and Maroncelli.58 Despite several differences in opinion obvious in these papers, an adequate treatment of at least the nonspecific components of solvatochromism would seem to be “just around the corner”. Finally, a suggestion should be mentioned on using the calculated π* values taken from ref. 55 as a descriptor of nonspecific solvent effects.59 However, this is not meaningful since these values are just a particular blend of inductive, dispersive, and dipole-dipole forces. 13.1.7 SOME HIGHLIGHTS OF RECENT INVESTIGATIONS The like dissolves like rule The buzzword “polarity”, derived from the dielectric approach, is certainly the most popular word dealing with solvent effects. It is the basis for the famous rule of thumb “similia similibus solvuntur” (“like dissolves like”) applied for discussing solubility and miscibility. Unfortunately, this rule has many exceptions. For instance, methanol and toluene, with dielectric constants of 32.6 and 2.4, respectively, are miscible, as are water (78.4) and isopropanol (18.3). The problem lies in exactly what is meant by a “like” solvent. Originally, the term “polarity” was meant to be an abbreviation of “static dipolarity” and was thus associated with solely the dielectric properties of the solvent. Later on, with the advent
754
Roland Schmid
of the empirical solvent parameters, it has assumed a broader meaning, sometimes even that of the overall solvating power.13 With this definition, however, the term “polarity” is virtually superfluous. Clearly, neither the dielectric constant nor the dipole moment is an adequate means to define polarity. The reason is that there are liquids whose constituent molecules have no net dipole moment, for symmetry reasons, but nevertheless have local polar bonds. This class of solvents, already mentioned above, comprises just the notorious troublemakers in solvent reactivity correlations, namely the aromatic and chlorinated solvents. These solvents, called “nondipolar” in the literature,60 stabilize charge due to higher solvent multipoles (in addition to dispersive forces) like benzene (“quadrupolar”) and carbon tetrachloride (“octupolar”). Of this class, the quadrupolar solvents are of primary importance. Thus, the gas-phase binding energy between of K+ and benzene is even slightly greater than that of K+-water. The interaction between the cation and the benzene molecule is primarily electrostatic in nature, with the ion-quadrupole interaction accounting for 60% of the binding energy.61 This effect is size-dependent: Whereas at K+ benzene will displace some water molecules from direct contact with the ion, Na +aq is resistant towards dehydration in an aromatic environment, giving rise to selectivity in some K+ channel proteins.62 For the polarity of the C-H bonds, it should be remembered that electronegativity is not an intrinsic property of an atom, but instead varies with hybridization. Only the C(sp3)-H bond can be considered as truly nonpolar, but not so the C(sp2)-H bond.63 Finally, ethine has hydrogen atoms that are definitely acidic. It should further be mentioned that higher moments or local polarities cannot produce a macroscopic polarization and thus be detected in infinite wavelength dielectric experiments yielding a static dielectric constant close to the squared refractive index. Because of their short range, quadrupolar interactions do not directly contribute to the dielectric constant, but are reflected only in the Kirkwood gK factor that decreases due to breaking the angular dipole-dipole correlations with increasing quadrupolar strength. In these terms it is strongly recommended to redefine the term polarity. Instead of meaning solely dipolarity, it should also include higher multipolar properties, polarity = dipolarity + quadrupolarity + octupolarity
[13.1.13]
This appears to be a better scheme than distinguishing between truly nonpolar and nondipolar solvents.56 A polar molecule can be defined as having a strongly polar bond, but need not necessarily be a dipole. In this framework, the solvating power of the “nondipolar” solvents need no longer be viewed as anomalous or as essentially dependent on specific solvation effects.10 Beyond this it should be emphasized that many liquids have both a dipole moment and a quadrupole moment, water for example. However, for dipolar solvents such as acetonitrile, acetone, and dimethyl sulfoxide, the dipolar solvation mechanism will be prevailing. For less dipolar solvents, like tetrahydrofuran, quadrupoles and dipoles might equally contribute to the solvation energetics.57 Notwithstanding this modified definition, the problem with polarity remains in that positive and negative charge solvation is not distinguished. As already pointed out above, there is no general relationship between polarity and the cation solvation tendency. For example, although nitromethane (MeNO2) and DMF have the same dielectric constant, the extent of ion pairing in MeNO2 is much greater than that in DMF. This observation is
13.1 Solvent effects on chemical reactivity
755
attributed to the weak basicity of MeNO2 which poorly solvates cations. As a result, ion pairing is stronger in MeNO2 in spite of the fact that long-range ion-ion interactions in the two solvents are equal. Finally, a potential problem with polarity rests in the fact that this term is typically associated with enthalpy. But caution is urged in interpreting the like-dissolves-like rule in terms of enthalpy. It is often stated for example that nonpolar liquids such as octane and carbon tetrachloride are miscible because the molecules are held together by weak dispersion forces. However, spontaneous mixing of the two phases is driven not by enthalpy, but by entropy. Water’s anomalies The outstanding properties and anomalies of water have fascinated and likewise intrigued physicists and physical chemists for a long time. During the past decades much effort has been devoted to finding phenomenological models that explain the (roughly ten) anomalous thermodynamic and kinetic properties, including the density maximum at 4°C, the expansion upon freezing, the isothermal compressibility minimum at 46°C, the high heat capacity, the decrease of viscosity with pressure, and the remarkable variety of crystalline structures. Furthermore, isotope effects on the densities and transport properties do not possess the ordinary mass or square-root-mass behavior. Some of these properties are known from long ago, but their origin has been controversial. From the increasingly unmanageable number of papers that have been published on the topic, let us quote only a few that appear to be essential. Above all, it seems to be clear that the exceptional behavior of water is not simply due to hydrogen bonding, but instead due to additional “trivial” vdW forces as present in any liquid. A hydrogen bond occurs when a hydrogen atom is shared between generally two electronegative atoms; vdW attractions arise from interactions among fixed or induced dipoles. The superimposition and competition of both is satisfactorily accommodated in the framework of a “mixture model”. The mixture model for liquid water, promoted in an embryonic form by Röntgen64 over a century ago, but later discredited by Kauzmann65 and others,66 is increasingly gaining ground. Accordingly there are supposed to be two major types of intermolecular bonding configurations, an open bonding form, with a low density, such as occurs in ice-Ih, plus a dense bonding form, such as occurs in the most thermodynamically stable dense forms of ice, e.g., ice-II, -III, -V, and -VI.67 In these terms, water has many properties of the glassy states associated with multiple hydrogen-bond network structures.68 Clearly, for fluid properties, discrete units, (H2O)n, which can move independently of each other are required. The clusters could well be octamers dissociating into tetramers, or decamers dissociating into pentamers.69,70 (Note by the way that the unit cell of ice contains eight water molecules.) However, this mixture is not conceived to be a mixture of ices, but rather is a dynamic (rapidly fluctuating) mixture of intermolecular bonding types found in the polymorphs of ice. A theoretical study of the dynamics of liquid water has shown that there exist local collective motions of water molecules and fluctuation associated with hydrogen bond rearrangement dynamics.68 The half-life of a single H bond estimated from transition theory is about 2x10-10 s at 300K.71 In view of this tiny lifetime it seems more relevant to identify the two mixtures not in terms of different cluster sizes, but rather in terms of two different bonding modes. Thus, there is a competition between dispersion interactions that favor random dense states and hydrogen bonding that favors ordered open states. Experimental verification of the two types of bonding has been reviewed by Cho et al.72
756
Roland Schmid
From X-ray and neutron scattering, the open structure is characterized by an inner tetrahedral cage of four water molecules surrounding a central molecule, with the nearest-neighbor OLO distance of about 2.8 Å. This distance, as well as the nearest-neighbor count of four, remains essentially intact in both in all crystalline ice polymorphs and in the liquid water up to near the boiling point. In this open tetrahedral network the second-neighbor OLO distance is found at 4.5 Å. However, and most intriguing, another peak in the OLO radial distribution function (RDF), derived from a nonstandard structural approach (ITD), is found near 3.4 Å,73 signaling a more compact packing than in an ordinary H-bonding structure. This dense bonding form is affected through dispersive OLO interactions supplanting H-bonding. Note, however, that in this array the H bonds may not be envisaged as being really broken but instead as being only bent. This claim is substantiated by a sophisticated analysis of vibrational Raman spectra74 and mid-IR spectra70 pointing to the existence of essentially two types of H bonds differing in strength, with bent H bonds being weaker than normal (i.e., linear) H bonds. It should be mentioned that the 3.4 Å feature is hidden by the ordinary minimum of open tetrahedral contributions to the RDF. Because of this, the ordinary integration procedure yields coordination numbers greater than four,75 which confuses the actual situation. Instead, it is the outer structure that is changing whereas the inner coordination sphere remains largely invariant. Even liquid water has much of the tetrahedral H-bonding network of ice I. As temperature, or pressure, is raised, the open tetrahedral hydrogen bonding structure becomes relatively less stable and begins to break down, creating more of the dense structure. Actually all the anomalous properties of water can be rationalized on the basis of this open → dense transformation. An extremum occurs if two opposing effects are superimposed. The density maximum, for instance, arises from the increase in density due to the thermal open → dense transformation and the decrease in density due to a normal thermal expansion.72 As early as 1978 Benson postulated that the abnormal heat capacity of water is due to an isomerization reaction.76 A clear explanation of the density anomaly is given by Silverstein et al.77: “The relatively low density of ice is due to the fact that H-bonding is stronger than the vdW interactions. Optimal H-bonding is incommensurate with the tighter packing that would be favored by vdW interactions. Ice melts when the thermal energy is sufficient to disrupt and disorder the H-bonds, broadening the distribution of H-bond angles and lengths. Now among this broadened H-bond distribution, the vdW interactions favor those conformations of the system that have higher density. Hence liquid water is denser than ice. Heating liquid water continues to further deform hydrogen bonds and increase the density up to the density anomaly temperature. Further increase of temperature beyond the density anomaly weakens both H bonds and vdW bonds, thus reducing the density, as in simpler liquids.” The same authors commented on the high heat capacity of water as follows: “Since the heat capacity is defined as CP = (∂H/∂T)P the heat capacity describes the extent to which some kind of bonds are broken (increasing enthalpy) with increasing temperature. Breaking bonds is an energy storage mechanism. The heat capacity is low in the ice phase because thermal energy at those temperatures is too small to disrupt the H bonds. The heat capacity peaks at the melting temperature where the solid-like H bonds of ice are weakened to become the liquid-like H bonds of liquid water. The reason liquid water has a higher heat capacity than vdW liquids have is because water has an additional energy storage mechanism, namely the H bonds, that can also be disrupted by thermal energies.”
13.1 Solvent effects on chemical reactivity
757
Summed up, it appears that any concept to be used in a realistic study of water should have as a fundamental ingredient the competition between expanded, less dense structures, and compressed, more dense ones. Thus, the outer structure in the total potential of water should be characterized by a double minimum: open tetrahedral structure with a second-neighbor OLO distance of 4.5 Å and a bent H-bond structure with an OLO non-H-bonded distance of about 3.4 Å. Actually, according to a quite recent theoretical study, all of the anomalous properties of water are qualitatively explainable by the existence of two competing equilibrium values for the interparticle distance.78 Along these lines the traditional point of view as to the structure of water is dramatically upset. Beyond that, also the classical description of the hydrogen bond needs revision. In contrast to a purely electrostatic bonding, quite recent Compton X-ray scattering studies have demonstrated that the hydrogen bonds in ice have substantial covalent character,79 as already suggested by Pauling in the 1930s.80 In overall terms, a hydrogen bond is comprised of electrostatic, dispersion, charge-transfer, and steric repulsion interactions. Similarly, there are charge-transfer interactions between biological complexes and water81 that could have a significant impact on the understanding of biomolecules in aqueous solution. Finally, we return to the physical meaning of the large difference, for the protic solvents, between the cohesive energy density εc and the internal pressure Pi, quoted in section 13.1.3. For water this difference is highest with the factor εc/Pi equal to 15.3. At first glance this would seem explainable in the framework of the mixture model if H bonding is insensitive to a small volume expansion. However, one should have in mind the whole pattern of the relationship between the two quantities. Thus, εc - Pi is negative for nonpolar liquids, relatively small (positive or negative) for polar non-associated liquids, and strongly positive for H-bonded liquids. A more rigorous treatment41 using the relations, Pi = (∂U/∂V)T = T(∂P/ ∂T)V - P and the thermodynamic identity (∂S/∂V)T = (∂P/∂T)V reveals that the relationship is not as simple and may be represented by the following equation with dispersion detached from the other types of association, ε c − Pi = P −
Udisp RT 2 Z 0 + − ρUass + ρ T (∂S ass / ∂ρ)T V RT
[13.1.14]
where: P V Z0 Udisp Uass ρ Sass
external pressure liquid volume compressibility factor due to intermolecular repulsion potential of dispersion potential of association excluding dispersion liquid number density entropy of association excluding dispersion
With the aid of this equation we readily understand the different ranges of εc - Pi found for the different solvent classes. Thus, for the nonpolar liquids, the last two terms are negligible, and for the usual values, Zo ≈ 10, -Udisp/RT ≈ 8, and V ≈ 150 cm3, we obtain the typical order of εc - Pi ≈ -(300 - 400) atm (equal to -(30 - 40) J cm-3, since 1 J cm-3 ≡ 9.875 atm). For moderately polar liquids, only the last term remains small, while the internal energy of dipolar forces is already appreciable -ρUpolar ≈ (200-500) atm giving the usual magnitude of εc - Pi. For H-bonded liquids, ultimately, the last term turns out to dominate reflecting the large increase in entropy of a net of H-bonds upon a small decrease in liquid density.
758
Roland Schmid
The hydrophobic effect In some respects the hydrophobic effect may be considered as the converse of the like-dissolves-like rule. The term hydrophobic effect refers to the unusual behavior of water towards nonpolar solutes. Unlike simple organic solvents, the insertion of nonpolar solutes into water is (1) strongly unfavorWater CCl4 able though slightly favored by enthalpy, but (2) - 64.4 - 7.1 ∆S*, J/mol K strongly opposed by a large, negative change in entropy at room temperature, and (3) accompanied by 300∆S*, kJ/mol - 19.3 - 2.1 a large positive heat capacity. An example is given - 10.9 - 1.2 ∆H°, kJ/mol in Table 13.1.5 for the thermodynamic properties of + 8.4 + 0.9 ∆G*, kJ/mol methane dissolved in water and in carbon tetrachloride. In dealing with the entropy (and free energy) of 217.5 0 to 42 ∆Cp, J/mol K hydration, a brief remark on the choice of standard states is in order. The standard molar entropy of dissolution, ∆solvS° pertains to the transfer from a 1 atm gas state to a 1 mol L-1 solution and hence includes compression of the gas phase from 1 mol contained in 24.61 L (at 300 K) to 1 mol present in 1 L. Since theoretical calculations disregard volume contributions, it is proper to exclude the entropy of compression equal to -Rln24.61 = -26.63 J K-1 mol-1, and instead to deal with ∆solvS*.82 Thus, Table 13.1.5. Solution thermodynamics of methane in water and carbon tetrachloride at 25°C. [Data from T. Lazaridis and M. E. Paulaitis, J. Phys. Chem., 96, 3847 (1992) and ref. 108]
∆solv S * = ∆solv S o + 26.63 JK −1mol −1
[13.1.15]
∆solv G *300 = ∆solv H o − 300∆solv S *
[13.1.16]
and
Hydrophobicity forms the basis for many important chemical phenomena including the cleaning action of soaps and detergents, the influence of surfactants on surface tension, the immiscibility of nonpolar substances in water,83 the formation of biological membranes and micelles,84,85 the folding of biological macromolecules in water,86 clathrate hydrate formation,87 and the binding of a drug to its receptor.88 Of these, particularly intriguing is the stabilization of protein structure due to the hydrophobicity of nonpolar groups. Hydrophobic interactions are considerably involved in self-assembly, leading to the aggregation of nonpolar solutes, or equivalently, to the tendency of nonpolar oligomers to adopt chain conformations in water relative to a nonpolar solvent.89 Ever-increasing theoretical work within the last years is being lifting the veil of secrecy about the molecular details of the hydrophobic effect, a subject of vigorous debate. Specifically, the scientific community would eagerly like to decide whether the loss in entropy stems from the water-water or the water-solute correlations. There are two concepts. The older one is the clathrate cage model reaching back to the “iceberg” hypothesis of Frank and Evans,90 and the other, newer one, is the cavity-based model. It should be stressed here that the vast literature on the topic is virtually impossible to survey comprehensively. In the following we will cite only a few papers (and references therein) that paved the way to the present state of the art.
13.1 Solvent effects on chemical reactivity
759
The clathrate cage model states that the structure of water is strengthened around a hydrophobic solute, thus causing a large unfavorable entropic effect. The surrounding water molecules adopt only a few orientations (low entropy) to avoid “wasting” hydrogen bonds, with all water configurations fully H-bonded (low energy). There is experimental evidence of structure strengthening, such as NMR and FT-IR studies,91 NMR relaxation,92 dielectric relaxation,93 and HPLC.94 A very common conclusion is that the small solubility of nonpolar solutes in water is due to this structuring process. In the cavity-based model the hard core of water molecules is more important to the hydrophobic effect than H-bonding of water. The process of solvation is dissected into two components, the formation of a cavity in the water to accommodate the solute and the interaction of the solute with the water molecules. The creation of a cavity reduces the volume of the translational motion of the solvent particles. This causes an unfavorable entropic effect. The total entropy of cavity formation at constant pressure54 ∆S cav, P = ρα p (∂∆G cav / ∂ρ)T − ∆G cav / T
[13.1.17]
where ∆Scav,P ρ ∆Gcav
cavity formation entropy at constant pressure liquid number density free energy of cavity formation
is the result of the opposing nature of the (positive) liquid expansibility term and the (negative) chemical potential summand. Along these lines the large and negative entropy of cavity formation in water is traced to two particular properties of water: the small molecular size (σ = 2.87 Å) and the low expansibility (αp = 0.26x10-3 K-1), with the latter having the greater impact. It is interesting to note that in both aspects water is extraordinary. Water’s low expansibility reflects the fact that chemical bonds cannot be stretched by temperature. There is also a recent perturbation approach showing that it is more costly to accommodate a cavity of molecular size in water than in hexane as example.95 Considering the high fractional free volume for water (Table 13.1.2), it is concluded that the holes in water are distributed in smaller packets.96 Compared to a H-bonding network, a hard-sphere liquid finds more ways to configure its free volume in order to make a cavity. In the cavity-based model, large perturbations in water structure are not required to explain hydrophobic behavior. This conclusion arose out of the surprising success of the scaled particle theory (SPT),39 which is a hard-sphere fluid theory, to account for the free energy of hydrophobic transfers. Since the theory only uses the molecular size, density, and pressure of water as inputs and does not explicitly include any special features of H-bonding of water, the structure of water is arguably not directly implicated in the hydration thermodynamics. (However, the effect of H-bonds of water is implicitly taken into account through the size and density of water.) The proponents of this hypothesis argue that the entropic and enthalpic contributions arising from the structuring of water molecules largely compensate each other. In fact, there is thermodynamic evidence of enthalpy-entropy compensation of solvent reorganization.97-100 Furthermore, recent simulations101,102 and neutron scattering data103-105 suggest that solvent structuring might be of much lower extent than previously believed. Also a recent MD study report106 stated that the structure of water is preserved, rather than enhanced, around hydrophobic groups. Finally, the contribution of solute-water correlations to the hydrophobic effect may be displayed, for example, in the framework of the equation
760
Roland Schmid
∆Gsol = ∆G cav + ∆G att
[13.1.18]
where: ∆Gsol ∆Gcav ∆Gatt
free energy of dissolution free energy of cavity formation free energy of attractive interactions
This equation has been used by de Souza and Ben-Amotz107 to calculate values of ∆Gatt from the difference between experimental solubilities of rare gases, corresponding to ∆Gsol, and ∆Gcav assessed from eqn. [13.1.10], i.e., using a hard-sphere fluid (HF) model. The values of ∆Gatt so obtained have been found to correlate with the solute polarizabilities suggesting a dispersive mechanism for attractive solvation. It is interesting to note that, in water, the solubility of the noble gases increases with increasing size, in contrast to the aliphatic hydrocarbons whose solubility decreases with size. This differential behavior is straightforwardly explained in terms of the high polarizability of the heavy noble gases having a large number of weakly bound electrons, which strengthens the vdW interactions with water. It can be shown that for noble gases, on increasing their size, the vdW interactions increase more rapidly than the work of cavity creation, enhancing solubility. On the contrary, for the hydrocarbons, on increasing the size, the vdW interactions increase less rapidly than the work of cavity creation, lowering the solubility.108 We have seen that there is evidence of either model, the clathrate cage model and the cavity-based model. Hence the importance of water structure enhancement in the hydrophobic effect is equivocal. The reason for this may be twofold. First, theoretical models have many adjustable parameters, so their physical bases are not always clear. Second, the free energy alone masks the underlying physics in the absence of a temperature dependence study, because of, amongst other things, the entropy-enthalpy compensation noted above. In place of the free energy, other thermodynamic derivatives are more revealing. Of these, the study of heat capacity changes arguably provides a better insight into the role of changes in water structure upon hydration than analysis of entropy or enthalpy changes alone. Note that heat capacity is the most complex of the four principal thermodynamic parameters describing solvation (∆G, ∆H, ∆S, ∆Cp), with the following connections, ∆Cp =
∂∆H ∂∆S ∂ 2 ∆G =T = −T 2 ∂T ∂T ∂T 2
[13.1.19]
It should be stressed that the negative entropy of hydration is virtually not the main characteristic feature of hydrophobicity, since the hydration of any solute, polar, nonpolar, or ionic, is accompanied by a decrease in entropy.109 The qualitative similarity in hydration entropy behavior of polar and nonpolar groups contrasts sharply with the opposite sign of the heat capacity change in polar and nonpolar group hydration. Nonpolar solutes have a large positive heat capacity of hydration, while polar groups have a smaller, negative one. Thus, the large heat capacity increase might be what truly distinguishes the hydrophobic effect from other solvation effects.110 Recently, this behavioral difference of nonpolar and polar solutes could be reproduced by heat capacity calculations using a combination of Monte Carlo simulations and the random network model (RNM) of water.110-112 It was found that the hydrogen bonds between the water molecules in the first hydration shell of a nonpolar solute are shorter and less bent (i.e., are more ice-like) compared to those in pure water. The opposite effect occurs around
13.1 Solvent effects on chemical reactivity
761
polar solutes (the waters become less ice-like). The increase in H-bond length and angle has been found to decrease the water heat capacity contribution, while decreases in length and angle have been found to cause the opposite effect. Note further that a large heat capacity implies that the enthalpy and entropy are strong functions of temperature, and the free energy vs. temperature is a curved function, increasing at low temperatures and decreasing at higher temperatures. Hence there will be a temperature at which the solubility of nonpolar in water is a minimum. The low solubility of nonpolar species in water at higher temperatures is caused by unfavorable enthalpic interactions, not unfavorable entropy changes. Some light on these features has been shed by using a “simple” statistical mechanical MB model of water in which the water molecules are represented as Lennard-Jones disks with hydrogen bonding arms.113 (the MB model is called this because of the resemblance of each model water to the Mercedes-Benz logo.) As an important result, the insertion of a nonpolar solute into cold water causes ordering and strengthening of the H bonds in the first shell, but the reverse applies in hot water. This provides a physical interpretation for the crossover temperatures TH and TS, where the enthalpy and entropy of transfer equal zero. TH is the temperature at which H-bond reorganizations are balanced by solute-solvent interactions. On the other hand, TS is the temperature at which the relative H-bonding strengths and numbers of shell and bulk molecules reverse roles. Although the large positive free energy of mixing of hydrocarbons with water is dominated by entropy at 25°C, it is dominated by enthalpy at higher temperatures (112°C from Baldwin’s extrapolation for hydrocarbons, or 150°C from the measurements of Crovetto for argon)113 where the disaffinity of oil for water is maximal. Ironically so, where hydrophobicity is strongest, entropy plays no role. For this reason, models and simulations of solutes that focus on cold water, around or below 25°C, miss much of the thermodynamics of the oil/water solvation process. Also, a clathrate-like solvation shell emerged from a recent computer simulation study of the temperature dependence of the structural and dynamical properties of dilute O2 aqueous solutions.114 In the first hydration shell around O2, water-water interactions are stronger and water diffusional and rotational dynamics slower than in the bulk. This calls to one’s mind an older paper by Hildebrand115 showing that at 25°C, methane’s diffusion coefficient in water is 40% less than it is in carbon tetrachloride (D(H2O) = 1.42x10-5 cm2/s vs D(CCl4) = 2.89x10-5 cm2/s). Presumably the loose clathrate water cages serve to inhibit free diffusion of the nonpolar solute. From these data it seems that both the nonpolar solute and the aqueous solvent experience a decrease in entropy upon dissolving in water. It should also be mentioned in this context that pressure increases the solubility. The effect of pressure on the entropy was examined and it was found that increase in the pressure causes a reduction of orientational correlations, in agreement with the idea of pressure as a “structure breaker” in water.116 Actually, frozen clathrate hydrates trapped beneath oceans and arctic permafrost may contain more than 50% of the world’s organic carbon reserves.117,118 Likewise, the solubility of aromatics is increased at high pressure and temperature, with π bond interactions involved.119 Only at first glance, the two approaches, the clathrate cage model and the cavity-based model, looked very different, the former based on the hydrogen bonding of water, and the later on the hard core of water. But taken all results together it would appear that both are just different perspectives on the same physics with different diagnostics reporting consequences of the same shifted balance between H bonds and vdW interactions. Actually, in a
762
Roland Schmid
very recent paper, a unified physical picture of hydrophobicity based on both the hydrogen bonding of water and the hard-core effect has been put forward.120 Hydrophobicity features an interplay of several factors. The structure of liquids A topic of abiding interest is the issue of characterizing the order in liquids which may be defined as the entropy deficit due to preferential orientations of molecular multipoles relative to random orientations (orientational order) and nonuniformly directed intermolecular forces (positional order). Phenomenologically, two criteria are often claimed to be relevant for deciding whether or not a liquid is to be viewed as ordered: the Trouton entropy of vaporization or Trouton quotient and the Kirkwood correlation factor gK.121 Strictly speaking, however, both are of limited relevance to the issue. The Trouton quotient is related to structure of the liquids only at their respective boiling points, of course, which may markedly differ from their structures at room temperature. This should be realized especially for the high-boiling liquids. As these include the highly dipolar liquids such as HMPA, DMSO, and PC, the effect of dipole orientation to produce order in the neat liquids remains obscure. All that can be gleaned from the approximate constancy of the Trouton quotient for all sorts of aprotic solvents is that at the boiling point the entropy of attractions becomes unimportant relative to the entropy of unpacking liquid molecules, that is repulsions.122 In terms of the general concept of separating the interaction potential into additive contributions of repulsion and attraction, the vaporization entropy can be expressed by ∆v H / Tb = ∆v S = ∆S o − S att
[13.1.20]
where: ∆ vH ∆So Satt
vaporization enthalpy at the boiling point Tb entropy of depacking hard spheres = entropy of repulsion entropy of attraction
∆So can be separated into the entropy fo(η) of depacking hard spheres to an ideal gas at the liquid volume V, and the entropy of volume expansion to Vg = RT/P, ∆S o / R = f o (η) + ln(Vg / V )
[13.1.21]
Further, fo(η) can be derived from the famous Charnahan-Starling equation as43
(
)
f o (η) = 4η − 3η 2 / (1 − η)
2
[13.1.22]
It can in fact be shown that, for the nonpolar liquids, ∆vS/R is approximately equal to fo(η).122 Along these lines, Trouton’s rule is traced to two facts: (i) The entropy of depacking is essentially constant, a typical value being ∆So/R ≈ 9.65, due to the small range of packing densities encompassed. In addition, the entropies of HS depacking and of volume change vary in a roughly compensatory manner. (ii) The entropy of attraction is insignificant for all the aprotics. Only for the protics the contributions of Satt may not be neglected. Actually the differences between ∆vS and ∆So reflect largely the entropy of hydrogen bonding. However, the application of the eqns [13.1.20] to [13.1.22] to room temperature data reveals, in con-
13.1 Solvent effects on chemical reactivity
763
trast to boiling point conditions, not unimportant contributions of Satt even for some aprotic liquids (see below). On the other hand, the gK factor is, loosely speaking, a measure of the deviation of the relative dielectric constant of the solvent with the same dipole moment and polarizability would have if its dipoles were not correlated by its structure. However, the gK factor is only the average cosine of the angles between the dipole moments of neighboring molecules. There may thus be orientational order in the vicinity of a molecule despite a gK of unity if there are equal head-to-tail and antiparallel alignments. Furthermore, the gK factor is not related to positional order. The better starting point for assessing order would be experimental room temperature entropies of vaporization upon applying the same method as described above for the boiling point conditions. (Note however, that the packing density, and hence the molecular HS diameter, varies with temperature. Therefore, in the paper122 the packing densities have been calculated for near boiling point conditions.) Thus we choose the simplest fluid as the reference system. This is a liquid composed of spherical, nonpolar molecules, approximated to a HS gas moving in a uniform background or mean field potential provided by the attractive forces.43 Since the mean field potential affects neither structure nor entropy, the excess entropy Sex S ex / R = ln(Vg / V ) + f o (η) − ∆v H / RT
[13.1.23]
may be viewed as an index of orientational and positional order in liquids. It represents the entropy of attractions plus the contributions arising from molecular nonsphericity. This latter effect can be estimated by comparing the entropy deficits for spherical and hard convex body repulsions in the reference system. Computations available for three n-alkanes suggest that only ≈ 20% of Sex are due to nonsphericity effects. Table 13.1.6. Some liquid properties concerning structure. Data are from ref. 55 and 128 (gK) ∆vH/RT
∆vSo/R
-Sex/R
c-C6
18.21
17.89
0.32
THF
12.83
12.48
0.35
CCl4
13.08
12.72
0.36
n-C5
10.65
10.21
0.44
CHCl3
12.62
12.16
0.46
CH2Cl2
11.62
11.14
0.48
Ph-H
13.65
13.08
0.57
Ph-Me
15.32
14.62
0.70
n-C6
12.70
11.89
0.81
Et2O
10.96
10.13
0.83
c-hexanone
18.20
17.30
0.90
Solvent
gK
764
Roland Schmid
∆vH/RT
∆vSo/R
-Sex/R
Py
16.20
15.22
0.98
MeCN
13.40
12.37
1.03
1.18
Me2CO
12.50
11.40
1.10
1.49
MeOAc
13.03
11.79
1.24
EtCN
14.53
13.25
1.28
1.15
PhNO2
22.19
20.89
1.30
1.56
MeNO2
15.58
14.10
1.48
1.38
NMP
21.77
20.25
1.52
1.52
EtOAc
14.36
12.76
1.60
DMSO
21.33
19.30
2.03
1.67
DMF
19.19
17.04
2.15
1.60
DMA
20.26
18.04
2.22
1.89
HMPA
24.65
22.16
2.49
1.44
Solvent
gK
PhCN
21.97
19.41
2.56
NMF
22.69
19.75
2.94
n-C11
22.76
19.66
3.10
FA
24.43
21.26
3.17
2.04
H 2O
17.71
14.49
3.22
2.79
MeOH
15.10
11.72
3.38
2.99
n-C13
26.72
22.51
4.21
EtOH
17.07
12.85
4.22
3.08
PC
26.33
22.06
4.27
1.86
n-PrOH
19.14
14.29
4.85
3.23
n-BuOH
21.12
15.90
5.22
3.26
4.52
The calculations for some common liquids are given in Table 13.1.6 ordered according to decreasing Sex. An important result is the appreciable order produced by the hydrocarbon chain relative to polar groups and hydrogen-bonding effects. For instance, Sex would project for water the same degree of order as for undecane. In like terms, ethanol is comparable to tridecane. However, the same magnitude of the excess entropy does by no means imply that ordering is similar. Much of the orientational ordering in liquids composed of elongated molecules is a consequence of efficient packing such as the intertwining of chains. In contrast, the structure of water is largely determined by strong electrostatic interactions leading to sharply-defined directional correlations characteristic of H-bonding. Although contrary to chemical tradition, there are other indications that the longer-chain hydrocarbon liquids are to be classified as highly structured as judged from
13.1 Solvent effects on chemical reactivity
765
thermodynamic123,124 and depolarized Rayleigh scattering data,125,126 and vibrational spectra.127 For nonprotic fluids, as Table 13.1.6 further shows, the vaporization entropy is strongly dominated by the entropy of HS depacking. This is an at least qualitative representation of the longstanding claim that repulsions play the major role in the structure of dense fluids.37 This circumstance is ultimately responsible for the striking success of the description of neutral reactions in the framework of a purely HS liquid, as discussed in Section 13.1.6. Also included in the Table are values of gK as determined in the framework of a generic mean spherical approximation.128 Since these values differ from those from other sources,129,121 because of differences in theory, we refrain from including the latter. It is seen that the gK parameter is unsuited to scale order, since positional order is not accounted for. On the other hand, values of gK exceed unity for the highly dipolar liquids and thus both Sex and gK attest to some degree of order present in them. Solvent reorganization energy in ET Electron transfer (ET) reactions in condensed matter continue to be of considerable interest to a wide range of scientists. The reasons are twofold. Firstly, ET plays a fundamental role in a broad class of biological and chemical processes. Secondly, ET is rather simple and very suitable to be used as a model for studying solvent effects and to relate the kinetics of ET reactions to thermodynamics. Two circumstances make ET reactions particularly appealing to theoreticians: • Outer-sphere reactions and ET within rigid complexes of well-defined geometry proceed without changes in the chemical structure, since bonds are neither formed nor broken. • The long-ranged character of interactions of the transferred electron with the solvent’s permanent dipoles. As a consequence of the second condition, a qualitative (and even quantitative) description can be achieved upon disregarding (or reducing through averaging) the local liquid structure changes arising on the length of molecular diameter dimensions relative to the charge-dipole interaction length. Because of this, it becomes feasible to use for outer-sphere reactions in strongly polar solvents the formalism first developed in the theory of polarons in dielectric crystals.130 In the treatment, the polar liquid is considered as a dielectric continuum characterized by the high-frequency ε ∞ and static εs dielectric constants, in which the reactants occupy spherical cavities of radii Ra and Rd, respectively. Electron transitions in this model are supposed to be activated by inertial polarization of the medium attributed to the reorientation of permanent dipoles. Along these lines Marcus131 obtained his well-known expression for the free energy ∆F of ET activation ∆F =
(∆Fo
+Er)
4E r
2
[13.1.24]
where ∆Fo is the equilibrium free energy gap between products and reactants and Er is the reorganization energy equal to the work applied to reorganize inertial degrees of freedom changing in going from the initial to the final charge distribution and can be dissected into inner-sphere and solvent contributions:
766
Roland Schmid
Er = Ei + Es
[13.1.25]
where: Ei Es
inner-sphere reorganization energy solvent reorganization energy
For outer-sphere ET the solvent component Es of the reorganization energy Es = e2cog
[13.1.26]
is the product of the medium-dependent Pekar factor co = 1/ε ∞ - 1/εs and a reactant-dependent (but solvent independent) geometrical factor g = 1/2Ra + 1/2Rd - 1/R
[13.1.27]
where: R e
the donor-acceptor separation the electron charge
Further advancements included calculations of the rate constant preexponent for nonadiabatic ET,132 an account of inner-sphere133 and quantum intramolecular134-136 vibrations of reactants and quantum solvent modes.137,138 The main results were the formulation of the dependencies of the activation energy on the solvent dielectric properties and reactant sizes, as well as the bell-shaped relationship between ∆F and ∆Fo. The predicted activation energy dependence on both the solvent dielectric properties139,140 and the donor-acceptor distance141 has, at least qualitatively, been supported by experiment. A bell-shaped plot of ∆F vs ∆Fo was obtained for ET in exciplexes,142 ion pairs,143 intramolecular144 and outer-sphere145 charge shift reactions. However, the symmetric dependence predicted by eq. [13.1.24] has not yet been detected experimentally. Instead, always asymmetric plots of ∆F against ∆Fo are obtained or else, in the inverted region (∆Fo < -Er), ∆F was found to be nearly invariant with ∆Fo.146 A couple of explanations for the asymmetric behavior are circulating in the literature (see, e.g., the review by Suppan147). The first one148,149 considered vibrational excitations of high-frequency quantum vibrational modes of the donor and acceptor centers. Another suggestion150 was that the frequencies of the solvent orientational mode are significantly different around the charged and the neutral reactants. This difference was supposed to be brought about by dielectric saturation of the polar solvent. This model is rightly questioned151 since dielectric saturation cannot affect curvatures of the energy surface at the equilibrium point. Instead, dielectric saturation displays itself in a nonlinear deviation of the free energy surface from the parabolic form far from equilibrium. Hence, other sources of this behavior should be sought. Nevertheless, both concepts tend to go beyond the structureless description advocating a molecular nature of either the donor-acceptor complex or the solvent. Nowadays, theories of ET are intimately related to the theories of optical transitions. While formerly both issues have developed largely independently, there is now growing desire to get a rigorous description in terms of intermolecular forces shifting the research of ET reactions toward model systems amendable to spectroscopic methods. It is the combination of steady state and transient optical spectroscopy that becomes a powerful method of studying elementary mechanisms of ET and testing theoretical concepts. The classical treatments of ET and optical transition have been facing a serious problem when extended to weakly polar and eventually nonpolar solvents. Values of Eop (equal to Er in eq [13.1.25]) as
13.1 Solvent effects on chemical reactivity
767
extracted from band-shape analyses of absorption spectra were found to fall in the range 0.2 - 0.4 eV. Upon partitioning these values into internal vibrations and solvent degrees of freedom, although this matter is still ambiguous, the contribution of the solvent could well be on the order of 0.2 - 0.3 eV.152-154 Unfortunately, all continuum theories predict zero solvent reorganization energies for ET in nonpolar liquids. It is evident that some new mechanisms of ET, alternatively to permanent dipoles’ reorientation, are to be sought. It should be emphasized that the problem cannot be resolved by treatments of fixed positions of the liquid molecules, as their electronic polarization follows adiabatically the transferred electron and thus cannot induce electronic transitions. On the other hand, the displacement of molecules with induced dipoles are capable of activating ET. In real liquids, as we have stated above, the appreciable free volume enables the solvent molecules to change their coordinates. As a result, variations in charge distribution in the reactants concomitantly alter the packing of liquid molecules. This point is corroborated by computer simulations.155 Charging a solute in a Stockmayer fluid alters the inner coordination number from 11.8 for the neutral entity to 9.5 for the positively charged state, with the process accompanied by a compression of the solvation shell. It is therefore apparent that solvent reorganization involves reorganization of liquid density, in addition to the reorientational contribution. This concept has been introduced by Matyushov,156 who dissected the overall solvent reorganization energy Es into a dipole reorganization component Ep and a density reorganization component Ed, Es = Ep + Ed
[13.1.28]
It should be mentioned that the two contributions can be completely separated because they have different symmetries, i.e., there are no density/orientation cross terms in the perturbation expansion involved in the calculations. The density component comprises three mechanisms of ET activation: (i) translations of permanent dipoles, (ii) translations of dipoles induced by the electric field of the donor-acceptor complex (or the chromophore), and (iii) dispersion solute-solvent forces. On the other hand, it appears that in the orientational part only the permanent dipoles (without inductions) are involved. With this novel molecular treatment of ET in liquids the corundum of the temperature dependence of the solvent reorganization energy is straightforwardly resolved. Dielectric continuum theories predict an increase of Es with temperature paralleling the decrease in the dielectric constants. In contrast, experimental results becoming available quite recently show that Es decreases with temperature. Also curved Arrhenius plots eventually featuring a maximum are being reported, in weakly polar157 and nonpolar158 solvents. The bell-shaped temperature dependence in endergonic and moderately exergonic regions found for ET quenching reactions in acetonitrile159 was attributed to a complex reaction mechanism. Analogously, the maximum in the Arrhenius coordinates, peculiar to the fluorescence of exciplexes formed in the intramolecular160 and bimolecular161 pathways, is commonly attributed to a temperature dependent competition of exciplex formation and deactivation rates. A more reasonable explanation can be given in terms of the new theory as follows. A maximum in the Arrhenius coordinates follows from the fact that the two terms in eq. [13.1.28] depend differently on temperature. Density fluctuation around the reacting pair is determined mainly by the entropy of repacking hard spheres representing the repulsive part of the intermolecular interaction. Mathematically, the entropy of activation arises
768
Roland Schmid
from the explicit inverse temperature dependence Ed ∝1/T. Since the liquid is less packed at higher temperature, less energy is needed for reorganization. Repacking of the solvent should lead to larger entropy changes than those of dipoles’ reorientation that is enthalpic in nature due to the long-range character of dipole-dipole forces. The orientational component increases with temperature essentially as predicted by continuum theories. In these ways the two solvent modes play complementary roles in the solvent’s total response. This feature would lead to curved Arrhenius plots of ET rates with slight curvatures in the normal region of ET (-∆Fo < Er), but even a maximum in the inverted region (-∆Fo > Er).162,163 The maximum may however be suppressed by intramolecular reorganization and should therefore be discovered particularly for rigid donor-acceptor pairs. Photoinduced ET in binuclear complexes with localized electronic states provides at the moment the best test of theory predictions for the solvent dependent ET barrier. This type of reaction is also called metal-metal charge-transfer (MMCT) or intervalence transfer (IT). The application of the theory to IT energies for valence localized biruthenium complexes and the acetylene-bridged biferricenium monocation164 revealed its superiority to continuum theories. The plots of Es vs. Eop are less scattered, and the slopes of the best-fit lines are closer to unity. As a major merit, the anomalous behavior of some solvents in the continuum description - in particular HMPA and occasionally water - becomes resolved in terms of the extreme sizes, as they appear at the opposite ends of the solvent diameter scale. Recently, it became feasible for the first time, to measure experimentally for a single chemical system, viz. a rigid, triply linked mixed-valence binuclear iron polypyridyl complex, [Fe(440)3Fe]5+, the temperature dependencies of both the rate of thermal ET and the optical IT energy (in acetonitrile-d3).165 The net Er associated with the intramolecular electron exchange in this complex is governed exclusively by low frequency solvent modes, providing an unprecedented opportunity to compare the parameters of the theories of thermal and optical ET in the absence of the usual complications and ambiguities. Acceptable agreement was obtained only if solvent density fluctuations around the reacting system were taken into account. In these ways the idea of density fluctuations is achieving experimental support. The two latest reports on negative temperature coefficients of the solvent reorganization energy (decrease in Es with temperature) should also be mentioned.166,167 Thus, two physically important properties of molecular liquids are absent in the continuum picture: the finite size of the solvent molecules and thermal translational modes resulting in density fluctuations. Although the limitations of the continuum model are long known, the necessity for a molecular description of the solvent, curiously, was first recognized in connection with solvent dynamic effects in ET. Solvent dynamics, however, affects the preexponent of the ET rate constant and, therefore, influences the reaction rate much less than does the activation energy. From this viewpoint it is suspicious that the ET activation energy has so long been treated in the framework of continuum theories. The reason of this affection is the otherwise relative success of the latter, traceable to two main features. First, the solvents usually used are similar in molecular size. Second, there is a compensation because altering the size affects the orientational and translational parts of the solvent barrier in opposite directions. The solution ionic radius The solution ionic radius is arguably one of the most important microscopic parameters. Although detailed atomic models are needed for a full understanding of solvation, simpler phenomenological models are useful to interpret the results for more complex systems. The
13.1 Solvent effects on chemical reactivity
769
most famous model in this respect is that of Born, originally proposed in 1920,168 representing the simplest continuum theory of ionic solvation. For a spherical ion, the Born excess free energy ∆GB of solvation was derived by considering the free energy change resulting from the transfer of an ion from vacuum to solvent. The equation has a very simple dependence on the ionic charge z, the radius rB, and the solvent dielectric constant ε (for the prime see eq. [13.1.16]): ∆G *B =
−e 2 z 2 1 1− 2rB ε
[13.1.29]
While Born assumes that the dielectric response of the solvent is linear, nonlinear effects such as dielectric saturation and electrostriction should occur due to the high electric field near the ion.169 Dielectric saturation is the effect that the dipoles are completely aligned in the direction of the field so that any further increase in the field cannot change the degree of alignment. Electrostriction, on the other hand, is defined as the volume change or compression of the solvent caused by an electric field, which tends to concentrate dipoles in the first solvation shell of an ion. Dielectric saturation is calculated to occur at field intensities exceeding 104 V/cm while the actual fields around monovalent ions are on the order of 108 V/cm.170 In the following we concentrate on ionic hydration that is generally the focus of attention. Unaware of nonlinear effects, Latimer et al.171 showed that the experimental hydration free energies of alkali cations and halide anions were consistent with the simple Born equation when using the Pauling crystal radii rP increased by an empirical constant ∆ equal to 0.85 Å for the cations and 0.1 Å to the anions. In fact three years earlier a similar relationship was described by Voet.172 The distance rP + ∆ was interpreted as the radius of the cavity formed by the water dipoles around the ion. For cations, it is the ion-oxygen distance while for anions it is the ion-hydrogen distance of the neighboring water molecules. From those days onwards, the microscopic interpretation of the parameters of the Born equation has continued to be a corundum because of the ambiguity of using either an effective radius (that is a modification of the crystal radii) or an effective dielectric constant. Indeed, the number of modifications of the Born equation is hardly countable. Rashin and Honig,173 as example, used the covalent radii for cations and the crystal radii for anions as the cavity radii, on the basis of electron density distributions in ionic crystals. On the other hand, Stokes174 put forward that the ion’s radius in the gas-phase might be appreciably larger than that in solution (or in a crystal lattice of the salt of the ion). Therefore, the loss in self-energy of the ion in the gas-phase should be the dominant contributor. He could show indeed that the Born equation works well if the vdW radius of the ion is used, as calculated by a quantum mechanical scaling principle applied to an isoelectronic series centering around the crystal radii of the noble gases. More recent accounts of the subject are available.175,176 Irrespective of these ambiguities, the desired scheme of relating the Born radius with some other radius is facing an awkward situation: Any ionic radius depends on arbitrary divisions of the lattice spacings into anion and cation components, on the one hand, and on the other, the properties of individual ions in condensed matter are derived by means of some extra-thermodynamic principle. In other words, both properties, values of r and ∆G*, to be compared with one another, involve uncertain apportionments of observed quantities. Con-
770
Roland Schmid
sequently, there are so many different sets of ionic radii and hydration free energies available that it is very difficult to decide which to prefer. In a most recent paper,82 a new table of absolute single-ion thermodynamic quantities of hydration at 298 K has been presented, based on conventional enthalpies and entropies upon implication of the thermodynamics of water dissociation. From the values of ∆hydG* the Born radii were calculated from rB (Å) = -695 z2/∆hydG*(kJ)
[13.1.30]
as given in Table 13.1.7. This is at first a formal definition whose significance may be tested in the framework of the position of the first maximum of the radial distribution function (RDF) measured by solution X-ray and neutron diffraction.177 However, the procedure is not unambiguous as is already reflected by the names given to this quantity, viz. (for the case of a cation) ion-water178 or ion-oxygen Table 13.1.7. Some radii (Å). Data are from distance. The ambiguity of the underlying ref. 82 interpretation resides in the circumstance that the same value of 1.40 Å is assigned in a b raq rmetal Atom rB the literature to the radius of the oxide anion, the water molecule and the vdW radius Li 1.46 1.50 1.52 of the oxygen atom. Na 1.87 1.87 1.86 It seems that many workers would K 2.33 2.32 2.27 tend to equate the distance (d) corresponding to the first RDF peak with the average Rb 2.52 2.48 distance between the center of the ion and Cs 2.75 2.58 2.65 the centers of the nearest water molecules, Be 1.18 1.06 1.12 d=rion + rwater. Actually, Marcus179,180 presented a nice relationship between d, averMg 1.53 1.52 1.60 aged over diffraction and simulation data, Ca 1.86 1.86 1.97 and the Pauling crystal radius in the form Sr 2.03 2.02 2.15 d=1.38 + 1.102 rp. Notwithstanding this success, it is preferable to implicate not the waBa 2.24 2.27 2.17 ter radius but instead the oxygen radius. F 1.39 1.29 This follows from the close correspondence between d and the metal-oxygen bond Cl 1.86 1.85 lengths in crystalline metal hydrates.82 Br 2.00 2.00 The gross coincidence of the solid and I 2.23 2.30 solution state distances is strong evidence that the value of d measures the distance bea From eqn. [13.1.30], beqns. [13.1.31] and [13.1.32]. tween the nuclei of the cation ad the oxygen rather than the center of the electron cloud of the whole ligand molecule. Actually, first RDF peaks for ion solvation in water and in nonaqueous oxygen donor solvents are very similar despite the different ligand sizes. Examples include methanol, formamide and dimethyl sulfoxide.180 Nevertheless, the division of d into ion and ligand components is still not unequivocal. Since the traditional ionic radius is often considered as a literal measure of size, it is usual to
13.1 Solvent effects on chemical reactivity
771
interpret crystallographic metal-oxygen distances in terms of the sum of the vdW radius of oxygen and the ionic radius of the metal. It should b emphasized, however, that the division of bond lengths into “cation” and “anion” components is entirely arbitrary. If the ionic radius is retained, the task remains to seek a connection to the Born radius, an issue that has a long-standing history beginning with the work of Voet.172 Of course, any addition to the ionic radius necessary to obtain good results from the Born equation needs a physical explanation. This is typically done in terms of the water radius, in addition to other correction terms such as a dipolar correlation length in the MSA (mean spherical approximation).181-183 In this case, however, proceeding from the first RDF peak, the size of the water moiety is implicated twice. It has been shown82 that the puzzle is unraveled if the covalent (atomic) radius of oxygen is subtracted from the experimental first peak position of the cation-oxygen radial distribution curve (strictly, the upper limits instead of the averages). The values of raq so obtained are very close to the Born radius, d(cation-O) - rcov(O) = raq ≈ rB
[13.1.31]
Similarly, for the case of the anions, the water radius, taken as 1.40 Å, is implicated, d(anion-O) - r(water) = raq ≈ rB
[13.1.32]
Furthermore, also the metallic radii (Table 13.1.7) are similar to values of raq. This correspondence suggests that the positive ion core dimension in a metal tends to coincide with that of the corresponding rare gas cation. The (minor) differences between raq and rmetal for the alkaline earth metals may be attributed, among other things, to the different coordination numbers (CN) in the metallic state and the solution state. The involvement of the CN is apparent in the similarity of the metallic radii of strontium and barium which is obviously a result of cancellation of the increase in the intrinsic size in going from Sr to Ba and the decrease in CN from 12 to 8. Along these lines a variety of radii are brought under one umbrella, noting however a wide discrepancy to the traditional ionic radii. Cation radii larger than the traditional ionic radii would imply smaller anion radii so as to meet the (approximate) additivity rule. In fact, the large anion radii of the traditional sets give rise to at least two severe inconsistencies: (i) The dramatic differences on the order of 1 Å between the covalent radii and the anion radii are hardly conceivable in view of the otherwise complete parallelism displayed between ionic and covalent bonds.184 (ii) It is implausible that non-bonded radii185 should be smaller than ionic radii. For example, the ionic radius of oxygen of 1.40 Å implies that oxygen ions should not approach each other closer than 2.80 Å. However, non-bonding oxygen-oxygen distances as short as 2.15 Å have been observed in a variety of crystalline environments. “(Traditional) ionic radii most likely do not correspond to any physical reality,” Baur notes.186 It should be remarked that the scheme of reducing the size of the anion at the expense of that of the cation has been initiated by Gourary and Adrian, based on the electron density contours in crystals.187 The close correspondence seen between rB and raq supports the idea that the Born radius (in aqueous solution) is predominantly a distance parameter without containing dielectric, i.e., solvent structure, contributions. This result could well be the outcome of a cancellation of dielectric saturation and electrostriction effects as suggested recently from
772
Roland Schmid
simulations.188-191 It should be emphasized, however, that the present discussion might be confined to water as the solvent. Recent theoretical treatments advise the cavity radius not to be considered as an intrinsic property of the solute, but instead to vary with solvent polarity, with orientational saturation prevailing at low polarity and electrostriction at high polarity.191,192 It would appear that the whole area of nonaqueous ion solvation deserves more methodical attention. It should in addition be emphasized that the cavity radius is sensitive to temperature. Combining eq. [13.1.29] with ∂∆G * ∆H o = ∆G * +T ∂T
[13.1.33]
one obtains177 T ∂ε T ∂r ∆H o = ∆G * 1 + − (ε − 1)ε ∂T P r ∂T P
[13.1.34]
The derivation of (∂r/∂T)P from reliable values of ∆H° and ∆G* is interesting, in that nominally the dielectric effect (-0.018 for water) is smaller than the size effect (-0.069 for chloride), a result that has not given previously the attention due to it. Consequently, as Roux et al.177 stated, unless a precise procedure for evaluating the dependence of the radius on the temperature is available, the Born model should be restricted to the free energy of solvation. Notwithstanding this, beginning with Voet,172 the Born model has usually been tested by considering the enthalpies of hydration.193 The reason for the relative success lies in the fact that ion hydration is strongly enthalpy controlled, i.e., ∆H° ~ ∆G*. The discussion of radii given here should have implications to all calculations involving aqueous ionic radii, for instance the solvent reorganization energy in connection with eq. [13.1.26]. Thus, treatments using the crystal radii as an input parameter194-197 may be revisited. 13.1.8 THE FUTURE OF THE PHENOMENOLOGICAL APPROACH Originally, the empirical solvent parameters have been introduced to provide guidelines for the comparison of different solvent qualities and for an orientation in the search for an understanding of the complex phenomena in solution chemistry. Indeed, the choice of the right solvent for a particular application is an everyday decision for the chemist: which solvent should be the best to dissolve certain products, and what solvent should lead to increased reaction yields and/or rates of a reaction? In the course of time, however, a rather sophisticated scheme has developed of quantitative treatments of solute-solvent interactions in the framework of LSERs.198 The individual parameters employed were imagined to correspond to a particular solute-solvent interaction mechanism. Unfortunately, as it turned out, the various empirical polarity scales feature just different blends of fundamental intermolecular forces. As a consequence, we note at the door to the twenty-first century, alas with melancholy, that the era of combining empirical solvent parameters in multiparameter equations, in a scientific context, is beginning to fade away. As a matter of fact, solution chemistry research is increasingly being occupied by theoretical physics in terms of molecular dynamics (MD) and Monte Carlo (MC) simulations, the integral equation approach, etc.
13.1 Solvent effects on chemical reactivity
773
In the author’s opinion, it seems that further usage of the empirical parameters should more return to the originally intended purpose, emphasizing more the qualitative aspects rather than to devote too much effort to multilinear regression analyses based on parameters quoted to two decimal places. Admittedly, such a scheme may nevertheless still be used to get some insight concerning the nature of some individual solvent effect as in the recent case of an unprecedented positive wavelength shift in the solvatochromism of an aminobenzodifuranone.199 The physical approach, though still in its infancy, has been helping us to see the success of the phenomenological approach in a new light. Accordingly, the reason for this well documented and appreciated success can be traced back to the following features • The molecular structure and the molecular size of many common solvents are relatively similar. The majority belongs to the so-called select solvents having a single dominant bond dipole, which, in addition, is typically hard, viz. an O- or N-donor. For example, if also soft donors (e.g., sulfur) had been employed to a larger extent, no general donor strength scale could have been devised. Likewise, we have seen that solvents other than the select ones complicate the issue. • As it runs like a thread through the present treatment, various cancellations and competitions (enthalpy/entropy, repulsion/attraction, etc.) appear to be conspiring to make molecular behavior in complex fluids seemingly simple. Notwithstanding this, the phenomenological approach will remain a venerable cornerstone in the development of unraveling solvent effects. Only time will tell whether a new generation of solvent indices will arise from the physical approach. REFERENCES 1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
The reaction studied was the quaternarization of triethylamine by ethyl iodide at 100 °C [N. Menschutkin, Z. Phys. Chem., 6, 41 (1890)]. Menschutkin’s first discussion on solvent effects dealt with the reactions between acetic anhydride and alcohols [Z. Phys. Chem., 1, 611 (1887)]. The catalytic role of solvents was already recognized in 1862 by Berthelot and Péan de Saint Gilles in their Recherches sur les Affinités [see, e. g., H. G. Grimm, H. Ruf, and Wolff, Z. Phys. Chem., B13, 301 (1931)]. E. D. Hughes and C. K. Ingold, J. Chem. Soc., 244 (1935). H. Suhr, Ber. Bunsenges. Phys. Chem., 67, 893 (1963). R. Schmid, J. Solution Chem., 12, 135 (1983). V. Gutmann, Donor-Acceptor Approach to Molecular Interactions, Plenum Press, New York, 1978. N. V. Sidgwick, The Electronic Theory of Valency, Clarenton Press, Oxford, 1927. R. Schmid, Rev. Inorg. Chem., 11, 255 (1991). R. W. Taft, N. J. Pienta, M. J. Kamlet, and E. M. Arnett, J. Org. Chem., 46, 661 (1981). W. Linert and R. F. Jameson, J. Chem. Soc. Perkin Trans. 2, 1993, 1415. V. Bekárek, J. Phys. Chem., 85, 722 (1981). Y. Marcus, Chem. Soc. Rev., 409 (1993). C. Laurence, P. Nicolet, M. T. Dalati, J. M. Abboud, and R. Notario, J. Phys. Chem., 98, 5807 (1994). C. Reichardt, Chem. Rev., 94, 2319 (1994). M. R. J. Dack, Aust. J. Chem., 28, 1643 (1975). M. J. Kamlet, J. M. Abboud, M. H. Abraham, and R. W. Taft, J. Org. Chem., 48, 2877 (1983). R. W. Taft, M. H. Abraham, R. M. Doherty, and M. J. Kamlet, J. Am. Chem. Soc., 107, 3105 (1985). M. H. Abraham, P. L. Grellier, A. Nasehzadeh, and R. A. C. Walker, J. Chem. Soc. Perkin Trans. II, 1988, 1717. R. W. Soukup and R. Schmid, J. Chem. Educ., 62, 459 (1985). U. Mayer, Pure Appl. Chem., 51, 1697 (1979). M. J. Kamlet, J. L. M. Abboud, and R. W. Taft, Prog. Phys. Org. Chem., 13, 485 (1981). R. Schmid and V. N. Sapunov, Non-Formal Kinetics in Search for Chemical Reaction Pathways, Verlag Chemie, Weinheim, 1982. M. H. Abraham, R. M. Doherty, M. J. Kamlet, and R. W. Taft, Chem. Br., 22, 551 (1986).
774
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
Roland Schmid N. A. Lewis, Y. S. Obeng, and W. L. Purcell, Inorg. Chem., 28, 3796 (1989). J. T. Hupp and M. J. Weaver, Inorg. Chem., 23, 3639 (1984). S. Glikberg and Y. Marcus, J. Solution Chem., 12, 255 (1983). Y. Marcus, M. J. Kamlet, and R. W. Taft, J. Phys. Chem., 92, 3613 (1988). C. J. Timpson, C. A. Bignozzi, B. P. Sullivan, E. M. Kober, and T. J. Meyer, J. Phys. Chem., 100, 2915 (1996). U. Mayer and V. Gutmann, Structure and Bonding, 12, 113 (1972). Y. Marcus, J. Solution Chem., 13, 599 (1984). Y. Kondo, M. Ohnishi, and N. Tokura, Bull. Chem. Soc. Japan, 45, 3579 (1972). H. F. Herbrandson and F. R. Neufeld, J. Org. Chem., 31, 1140 (1966). M. J. Kamlet, P. W. Carr, R. W. Taft, and M. H. Abraham, J. Am. Chem. Soc., 103, 6062 (1981). B. Widom, Science, 157, 375 (1967). J. A. Barker and D. Henderson, J. Chem. Educ., 45, 2 (1968). Y. Song, R. M. Stratt, and E. A. Mason, J. Chem. Phys., 88, 1126 (1988). S. Phan, E. Kierlik, M. L. Rosinberg, H. Yu, and G. Stell, J. Chem. Phys., 99, 5326 (1993). C. G. Gray and K. E. Gubbins, Theory of Molecular Fluids, Vol.1: Fundamentals, Clarendon Press, Oxford, 1984. L. K. Nash in Harvard Case Histories in Experimental Science, J. B. Conant, Ed., Harvard University Press, Harvard, 1950, p. 17. R. A. Pierotti, J. Phys. Chem., 67, 1840 (1963); Chem. Rev., 76, 717 (1976). E. Wilhelm and R. Battino, J. Chem. Phys., 55, 4021 (1971); J. Chem. Thermodyn., 3, 761 (1971); J. Chem. Phys., 58, 3558 (1973). D. V. Matyushov and R. Schmid, J. Chem. Phys., 104, 8627 (1996). D. Ben-Amotz and D. R. Herschbach, J. Phys. Chem., 94, 1038 (1990). R. Schmid and D. V. Matyushov, J. Phys. Chem., 99, 2393 (1995). D. Ben-Amotz and K. G. Willis, J. Phys. Chem., 97, 7736 (1993). T. Boublik, J. Chem. Phys., 53, 471 (1970). G. A. Mansoori, N. F. Carnahan, K. E. Starling, and T. W. Leland, Jr., J. Chem. Phys., 54, 1523 (1971). D. V. Matyushov and B. M. Ladanyi, J. Chem. Phys., 107, 5851 (1997). R. Ravi, L. E. S. Souza, and D. Ben-Amotz, J. Phys. Chem., 97, 11835 (1993). D. Ben-Amotz, J. Phys. Chem., 97, 2314 (1993). D. V. Matyushov and R. Schmid, J. Chem. Phys., 103, 2034 (1995). J. D. Weeks, D. Chandler, and H. C. Andersen, J. Chem. Phys., 54, 5237 (1971). J. A. Barker and D. Henderson, Rev. Mod. Phys., 48, 587 (1976). F. Cuadros, A. Mulero, and P. Rubio, J. Chem. Educ., 71, 956 (1994). D. V. Matyushov and R. Schmid, J. Chem. Phys., 105, 4729 (1996). D. V. Matyushov, R. Schmid, and B. M. Ladanyi, J. Phys. Chem. B, 101, 1035 (1997). L. Reynolds, J. A. Gardecki, S. J. V. Frankland, M. L. Horng, and M. Maroncelli, J. Phys. Chem., 100, 10337 (1996). D. V. Matyushov and G. A. Voth, J. Chem. Phys., 111, 3630 (1999). S. R. Mente and M. Maroncelli, J. Phys. Chem. B, 103, 7704 (1999). J. Catalán, J. Org. Chem., 62, 8231 (1997). B. C. Perng, M. D. Newton, F. O. Raineri, and H. L. Friedman, J. Chem. Phys., 104, 7177 (1996). O. M. Carbacos, C. J. Weinheimer, and J. M. Lisy, J. Chem. Phys., 108, 5151 (1998). O. M. Carbacos, C. J. Weinheimer, and J. M. Lisy, J. Chem. Phys., 110, 8429 (1999). S. G. Bratsch, J. Chem. Educ., 65, 34 (1988). W. C. Röntgen, Ann. Phys. u. Chem., 45, 91 (1892). W. Kauzmann, L’Eau Syst. Biol., Colloq. Int. C.N.R.S., 246, 63 (1975). See, for example: H. Endo, J. Chem. Phys., 72, 4324 (1980). M. Vedamuthu, S. Singh, and G. W. Robinson, J. Phys. Chem., 98, 2222 (1994). I. Ohmine, J. Chem. Phys., 99, 6767 (1995). S. W. Benson and E. D. Siebert, J. Am. Chem. Soc., 114, 4269 (1992). F. O. Libnau, J. Toft, A. A. Christy, and O. M. Kvalheim, J. Am. Chem. Soc., 116, 8311 (1994). H. Kistenmacher, G. C. Lie, H. Pople, and E. Clementi, J. Chem. Phys., 61, 546 (1974). C. H. Cho, S. Singh, and G. W. Robinson, J. Chem. Phys., 107, 7979 (1997). L. Bosio, S. H. Chen, and J. Teixeira, Phys. Rev., A27, 1468 (1983). G. D’Arrigo, G. Maisano, F. Mallamace, P. Migliardo, and F. Wanderlingh, J. Chem. Phys., 75, 4264 (1981). A. K. Soper and M. G. Phillips, Chem. Phys., 107, 47 (1986).
13.1 Solvent effects on chemical reactivity
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
775
S. W. Benson, J. Am. Chem. Soc., 100, 5640 (1978). K. A. T. Silverstein, A. D. J. Haymet, and K. A. Dill, J. Am. Chem. Soc., 120, 3166 (1998). E. A. Jagla, J. Chem. Phys., 111, 8980 (1999). E. D. Isaacs, A. Shukla, P. M. Platzman, D. R. Hamann, B. Barbiellini, and C. A. Tulk, Phys. Rev. Lett., 82, 600 (1999). L. Pauling, J. Am. Chem. Soc., 57, 2680 (1935). G. Nadig, L. C. Van Zant, S. L. Dixon, and K. M. Merz, Jr., J. Am. Chem. Soc., 120, 5593 (1998). R. Schmid, A. M. Miah, and V. N. Sapunov, Phys. Chem. Chem. Phys., 2, 97 (2000). A. Ben-Naim, Hydrophobic Interactions, Plenum, New York, 1980. D. F. Evans and B. W. Ninham, J. Phys. Chem., 90, 226 (1986). C. Tanford, The Hydrophobic Effect: Formation of Micelles and Biological Membranes, Wiley, New York, 1973. K. A. Dill, Biochemistry, 29, 7133 (1990). O. K. Forrisdahl, B. Kvamme, and A. D. J. Haymet, Mol. Phys., 89, 819 (1996). H. Wang and A. Ben-Naim, J. Med. Chem., 39, 1531 (1996). H. S. Ashbaugh, E. W. Kaler, and M. E. Paulaitis, J. Am. Chem. Soc., 121, 9243 (1999). H. S. Frank and M. W. Evans, J. Chem. Phys., 13, 507 (1945). K. Mizuno, Y. Miyashita, Y. Shindo, and H. Ogawa, J. Phys. Chem., 99, 3225 (1995). E. V. Goldammer and H. G. Hertz, J. Phys. Chem., 74, 3734 (1970). R. Pottel and V. Kaatze, Ber. Bunsenges. Phys. Chem., 73, 437 (1969). R. Silveston and B. Kronberg, J. Phys. Chem., 93, 6241 (1989). M. Prévost, I. T. Oliveira, J. P. Kocher, and S. J. Wodak, J. Phys. Chem., 100, 2738 (1996). A. Pohorille and L. R. Pratt, J. Am. Chem. Soc., 112, 5066 (1990). E. Grunwald and C. Steel, J. Am. Chem. Soc., 117, 5687 (1995). H. A. Yu, and M. Karplus, J. Chem. Phys., 89, 2366 (1988). H. Qian and J. J. Hopfield, J. Chem. Phys., 105, 9292 (1996). B. Lee, J. Chem. Phys., 83, 2421 (1985). I. I. Vaisman, F. K. Brown, and A. Tropsha, J. Phys. Chem., 98, 5559 (1994). M. Re, D. Laria, and R. Fernández-Prini, Chem. Phys. Lett., 250, 25 (1996). A. K. Soper and J. L. Finney, Phys. Rev. Lett., 71, 4346 (1993). J. Turner and A. K. Soper, J. Chem. Phys., 101, 6116 (1994). A. K. Soper and A. Luzar, J. Phys. Chem., 100, 1357 (1996). E. C. Meng and P. A. Kollman, J. Phys. Chem., 100, 11460 (1996). L. E. S. de Souza and D. Ben-Amotz, J. Chem. Phys., 101, 9858 (1994). G. Graziano, J. Chem. Soc., Faraday Trans., 94, 3345 (1998). A. Ben-Naim and Y. Marcus, J. Chem. Phys., 81, 2016 (1984). B. Madan and K. Sharp, J. Phys. Chem., 100, 7713 (1996). K. Sharp and B. Madan, J. Phys. Chem. B, 101, 4343 (1997). B. Madan and K. Sharp, J. Phys. Chem. B, 101, 11237 (1997). K. A. T. Silverstein, A. D. J. Haymet, and K. A. Dill, , J. Chem. Phys., 111, 8000 (1999). E. Fois, A. Gamba, and C. Redaelli, J. Chem. Phys., 110, 1025 (1999). J. H. Hildebrand, Proc. Natl. Acad, Sci. USA, 89, 2995 (1979). T. Lazaridis and M. Karplus, J. Chem. Phys., 105, 4294 (1996). T. S. Collett, Chem. Eng. News, 75, 60 (1997). T. P. Silverstein, J. Chem. Educ., 75, 116 (1998). S. Furutaka and S. Ikawa, J. Chem. Phys., 108, 5159 (1998). M. Ikeguchi, S. Seishi, S. Nakamura, and K. Shimizu, J. Phys. Chem. B, 102, 5891 (1998). Y. Marcus, J. Solution Chem., 21, 1217 (1992). D. V. Matyushov and R. Schmid, Ber. Bunsenges. Phys. Chem., 98, 1590 (1994). V. T. Lam, P. Picker, D. Patterson, and P. Tancrede, J. Chem. Soc., Faraday Trans. 2, 70, 1465, 1479 (1974). D. Patterson, J. Solution Chem., 23, 105 (1994). P. Borothorel, J. Colloid Sci., 27, 529 (1968). H. Quinones and P. Borothorel, Compt. Rend., 277, 133 (1973). R. G. Snyder, J. Chem. Phys., 47, 1316 (1967). L. Blum and W. R. Fawcett, J. Phys. Chem., 100, 10423 (1996). Y. Marcus, Ion Solvation, Wiley, New York, 1985; Chapter 6. S. I. Pekar, Investigations in the Electronic Theory of Crystals, GIFML, Moscow (in Russian). R. A. Marcus, J. Chem. Phys., 24, 966 (1956).
776
Roland Schmid
132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
R. R. Dogonadze and A. M. Kuznetsov, Physical Chemistry. Kinetics, VINITI, Moscow, 1973. R. A. Marcus, Disc. Farad. Soc., 29, 21 (1960). N. R. Kestner, J. Logan, and J. Jortner, J. Phys. Chem., 78, 2148 (1974). T. Holstein, Phil. Mag., B37, 499 (1978). P. Siders and R. A. Marcus, J. Am. Chem. Soc., 103, 748 (1981). A. A. Ovchinnimov and M. Ya. Ovchinnikova, Sov. Phys. JETF, 56, 1278 (1969). G. E. McManis, A. Gochev, and M. J. Weaver, Chem. Phys., 152, 107 (1991). G. M. Tom, C. Creutz, and H. Taube, J. Am. Chem. Soc., 96, 7828 (1974). H. Heitele, F. Pollinger, S. Weeren, and M. E. Michel-Beyerle, Chem. Phys., 143, 325 (1990). M. J. Powers and T. J. Meyer, J. Am. Chem. Soc., 98, 6731 (1976). P. P. Levin and P. K. N. Raghavan, Chem. Phys. Lett., 171, 309 (1990). I. R. Gould and S. Farid, J. Am. Chem. Soc., 110, 7833 (1988). G. L. Closs, L. T. Calcaterra, N. J. Green, K. W. Penfield, and J. R. Miller, J. Phys. Chem., 90, 3673 (1986). J. R. Miller, J. V. Beitz, and R. K. Henderson, J. Am. Chem. Soc., 106, 5057 (1984). D. Rehm and A. Weller, Isr. J. Chem., 8, 259 (1970). P. Suppan, Topics in Current Chemistry, Vol. 163, Springer, Berlin, 1992. S. Efrima and M. Bixon, Chem. Phys. Lett., 25, 34 (1974). N. R. Kestner, J. Logan, and J. Jortner, J. Phys. Chem., 78, 2148 (1974). T. Kakitani and N. Mataga, Chem. Phys. Lett., 124, 437 (1986); J. Phys. Chem., 91, 6277 (1987). M. Tachiya, Chem. Phys. Lett., 159, 505 (1989); J. Phys. Chem., 93, 7050 (1989). F. Markel, N. S. Ferris, I. R. Gould, and A. B. Myers, J. Am. Chem. Soc., 114, 6208 (1992). K. Kulinowski, I. R. Gould, and A. B. Myers, J. Phys. Chem., 99, 9017 (1995). B. M. Britt, J. L. McHale, and D. M. Friedrich, J. Phys. Chem., 99, 6347 (1995). L. Perera and M. L. Berkowitz, J. Chem. Phys., 96, 3092 (1992). D. V. Matyushov, Mol. Phys., 79, 795 (1993). H. Heitele, P. Finckh, S. Weeren, F. Pöllinger, and M. E. Michel-Beyerle, J. Phys. Chem., 93, 5173 (1989). E. Vauthey and D. Phillips, Chem. Phys., 147, 421 (1990). H. B. Kim, N. Kitamura, Y. Kawanishi, and S. Tazuke, J. Am. Chem. Soc., 109, 2506 (1987). F. D. Lewis and B. E. Cohen, J. Phys. Chem., 98, 10591 (1994). A. M. Swinnen, M. Van der Auweraer, F. C. De Schryver, K. Nakatani, T. Okada, and N. Mataga, J. Am. Chem. Soc., 109, 321 (1987). D. V. Matyushov and R. Schmid, Chem. Phys. Lett., 220, 359 (1994). D. V. Matyushov and R. Schmid, Mol. Phys., 84, 533 (1995). D. V. Matyushov and R. Schmid, J. Phys. Chem., 98, 5152 (1994). C. M. Elliott, D. L. Derr, D. V. Matyushov, and M. D. Newton, J. Am. Chem. Soc., 120, 11714 (1998). D. L. Derr and C. M. Elliott, J. Phys. Chem. A, 103, 7888 (1999). P. Vath, M. B. Zimmt, D. V. Matyushov, and G. A. Voth, J. Phys. Chem. B, 103, 9130 (1999). M. Born, Z. Phys., 1, 45 (1920). B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Elsevier, Amsterdam, 1981. J. F. Böttcher, Theory of Electric Polarization, Elsevier, Amsterdam, 1973, Vol. 1. W. M. Latimer, K. S. Pitzer, and C. M. Slansky, J. Chem. Phys., 7, 108 (1939). A. Voet, Trans. Faraday Soc., 32, 1301 (1936). A. A. Rashin and B. Honig, J. Phys. Chem., 89, 5588 (1985). R. H. Stokes, J. Am. Chem. Soc., 86, 979 (1964); J. Am. Chem. Soc., 86, 982 (1964). K. J. Laidler and C. Pegis, Proc. R. Soc., A241, 80 (1957). B. E. Conway and E. Ayranci, J. Solution Chem., 28, 163 (1999). B. Roux, H. A. Yu, and M. Karplus, J. Phys. Chem., 94, 4683 (1990). M. Mezei and D. L. Beveridge, J. Chem. Phys., 74, 6902 (1981). Y. Marcus, J. Solution Chem., 12, 271 (1983). Y. Marcus, Chem. Rev., 88, 1475 (1988). D. Y. C. Chan, D. J. Mitchell and B. W. Ninham, J. Chem. Phys., 70, 2946 (1979). L. Blum and W. R. Fawcett, J. Phys. Chem., 96, 408 (1992). W. R. Fawcett, J. Phys. Chem., 97, 9540 (1993). A. M. Pendás, A. Costales, and V. Luana, J. Phys. Chem. B, 102, 6937 (1998). M. O’Keeffe and B. G. Hyde, Structure and Bonding in Crystals, Academic Press, New York, 1981, vol. 1, p. 227. W. H. Baur, Cryst. Rev., 1, 59 (1987). B. S. Gourary and F. J. Adrian, Solid State Phys., 10, 127 (1960).
162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
13.2 Solvent effects on free radical polymerization
777
S. W. Rick and B. J. Berne, J. Am. Chem. Soc., 116, 3949 (1994). J. K. Hyun and T. Ichiye, J. Chem. Phys., 109, 1074 (1998). J. K. Hyun and T. Ichiye, J. Phys. Chem. B, 101, 3596 (1997). D. V. Matyushov and B. M. Ladanyi, J. Chem. Phys., 110, 994 (1999). A. Papazyan and A. Warshel, J. Chem. Phys., 107, 7975 (1997). D. W. Smith, J. Chem. Educ., 54, 540 (1977). G. E. McManis, A. Gochev, R. M. Nielson, and M. J. Weaver, J. Phys. Chem., 93, 7733 (1989). I. Rips, J. Klafter, and J. Jortner, J. Chem. Phys., 88, 3246 (1988). W. R. Fawcett and L. Blum, Chem. Phys. Lett., 187, 173 (1991). H. Heitele, P. Finckh, S. Weeren, F. Pollinger, and M. E. Michel-Beyerle, J. Phys. Chem., 93, 325 (1990). Quantitative Treatments of Solute/Solvent Interactions, P. Politzer and J. S. Murray, Eds., Elsevier, Amsterdam, 1994. 199 A. A. Gorman, M. G. Hutchings, and P. D. Wood, J. Am. Chem. Soc., 118, 8497 (1996).
188 189 190 191 192 193 194 195 196 197 198
13.2 SOLVENT EFFECTS ON FREE RADICAL POLYMERIZATION Michelle L. Coote and Thomas P. Davis Centre for Advanced Macromolecular Design, School of Chemical Engineering & Industrial Chemistry, The University of New South Wales, Sydney, Australia
13.2.1 INTRODUCTION Free radical polymerization is one of the most useful and lucrative fields of chemistry ever discovered - recent years have seen a tremendous increase in research into this area once considered a mature technological field. Free radical synthetic polymer chemistry is tolerant of diverse functionality and can be performed in a wide range of media. Emulsion and suspension polymerizations have been established as important industrial processes for many years. More recently, the ‘green’ synthesis of polymers has diversified from aqueous media to supercritical fluids and the fluorous biphase. An enduring feature of the research literature on free radical polymerization has been studies into specific solvent effects. In many cases the influence of solvent is small, however, it is becoming increasingly evident that solvent effects can be used to assist in controlling the polymerization reaction, both at the macroscopic and at the molecular levels. The purpose of this chapter is to give a brief introduction to the types of specific solvent effect that can be achieved in both free radical homo- and co-polymerizations. 13.2.2 HOMOPOLYMERIZATION Free radical polymerization can be conveniently codified according to the classical chain reaction steps of initiation, propagation, transfer and termination. In cases where a significant solvent effect is operative then the effect is normally exerted in all of these steps. However, for the purpose of facilitating discussion this chapter is broken down into these specific reaction steps. 13.2.2.1 Initiation Solvent effects on the initiation reaction are primarily on the rate of decomposition of initiator molecules into radicals and in the efficiency factor, f, for polymerization. However, in some instances the solvent plays a significant role in the initiation process, for example, in
778
Michelle L. Coote and Thomas P. Davis
initiation reactions with t-butoxy radical where the primary radical rarely initiates a chain but instead abstracts a hydrogen atom from the solvent medium, which subsequently initiates the chain.1 The consequence of this is that the polymer chains contain fragments of solvent. As the stability of the chains to thermal and photochemical degradation is governed, in part, by the nature of the chain ends then the solvent moieties within the chain can have a substantial impact on the material performance of the polymer. The efficiency factor, f, decreases as the viscosity of the reaction medium increases.2 This is caused by an increase in the radical lifetime within the solvent cage, leading to an increased possibility of radical-radical termination. In this regard the diffusion rates of the small radicals becomes an important consideration and Terazima and co-workers3,4 have published results indicating that many small radicals diffuse slower than expected. They have attributed this to specific interactions between radical and solvent molecules. 13.2.2.2 Propagation The ability of solvents to affect the homopropagation rate of many common monomers has been widely documented. For example, Bamford and Brumby5 showed that the propagation rate (kp) of methyl methacrylate (MMA) at 25°C was sensitive to a range of aromatic solvents. Burnett et al.6 found that the kp of styrene (STY) was depressed by increasing concentrations of benzonitrile, bromobenzene, diethyl phthalate, dinonyl phthalate and diethyl malonate, while in other studies7,8 they found that the kp for MMA was enhanced by halobenzenes and naphthalene. More recent work by Zammit et al.9 has shown that solvents capable of hydrogen-bonding, such as, benzyl alcohol and N-methyl pyrrolidone have a small influence on both the activation energy (Ea) and pre-exponential factor (A) in STY and MMA homopropagation reactions. These are but a few of the many instances of solvent effects in the homopolymerization reactions of two typical monomers, STY and MMA. For these monomers, solvent effects are relatively small, and this is indicative of the majority of homopropagation reactions. However, in some instances much larger effects are observed, especially in cases where specific interactions such as H-bonding or ionization occur. Examples of this type include the polymerization of N-vinyl-2-pyrrolidone (where water has been found to dramatically increase kp)10 and the polymerization of acrylamide (where pH plays a strong role).11 There is only limited data on the Arrhenius parameters for homopropagation reactions in different solvents and this indicates that both the activation energy and pre-exponential factor are affected.9,12 In some cases the solvent effect is not on the elementary rate constant kp but on the local monomer concentration (sometimes referred to as the ‘Bootstrap" effect). This effect can originate in the preferential solvation of either the monomer (which is always present as a solvent) or the added solvent. It has also been suggested that in some instances the growing polymer coil can ‘shield’ the radical chain-end resulting in a low monomer concentration. This shielding effect would be expected to be greatest in poor solvents (hence a tighter coil).13 For methyl methacrylate and styrene the largest solvent effects on propagation seem to be in the order of a 40% change in kp.14,15 In some solvents there seems to be reasonably strong evidence that the solvent does cause changes to the geometry of the transition state (e.g., dimethyl formamide and acetonitrile in styrene polymerization)14 and in liquid carbon dioxide it appears that the 40% change in kp for methyl methacrylate can be ascribed to the poor solvent medium.16 Recent work has found that some fluoro-alcohols17 can influence the tacticity of free radical polymerization lending further credence to the concept of solvent-induced changes to the transition state of
13.2 Solvent effects on free radical polymerization
779
the radical addition reaction. The largest solvent effects observed on kp for homopropagations have been for vinyl acetate18 and for α-(hydroxymethyl) ethyl acrylate.12 In the former case the radical is highly unstable and some form of π-complexation between the vinyl acetate radical and aromatic solvents seems plausible. However, the large solvent effect cannot be explained by a simple radical stabilization argument (because of the early transition state for free radical propagation reactions)19 and again the evidence points towards a change in the geometry of the transition state. The solvent effects on α-(hydroxymethyl) ethyl acrylate are in the order of 300% on kp and there are large changes in both the Arrhenius parameters as the solvent medium is changed.12 In monomers exhibiting a strong solvent effect on propagation it is plausible that some control of the stereochemistry of the chains is possible by manipulating the solvent and possibly utilizing Lewis acids as additives. This approach is already being successfully applied to the control of radical reactions in conventional organic chemistry.20 13.2.2.3 Transfer Solvent effects on transfer reactions have not received too much attention. It would be expected (owing to the similarities between the transition states for radical addition and abstraction reactions) that these solvent effects should emulate those found in propagation reactions. However, there is potential for significant polar interactions in transfer reactions. Odian21 has suggested that polar interactions play a significant role in the transfer reactions between styrene and carbon tetrachloride. More recent work supports this idea.22 Significant solvent effects have been observed in catalytic chain transfer reactions using cobaloximes where the transfer reaction appears (in some cases at least) to be diffusion controlled and therefore the speed of the reaction is governed, in part, by the viscosity of the polymerizing medium.23 In transfer reactions involving organometallic reagents then solvent effects may become important where ligand displacement may occur. This is thought to happen in catalytic chain transfer when pyridine is utilized as a solvent.24 13.2.2.4 Termination The solvent effects on the termination reaction have been extensively studied. In early work, it was established that the radical-radical termination reaction is diffusion controlled and the efficacy of termination was found to have a strong relationship with the solvent viscosity.25 Subsequently, more complex models have been developed accounting for the quality of the solvent (hence the size of the polymer coil).26 The current debate centers on the relative roles played by segmental and translational diffusion at different stages of conversion for a variety monomers. Clearly in both cases the nature of the solvent becomes important. Solvent effects are known to play a significant role in determining the strength and onset conversion of the gel effect. This work originated in the classical paper by Norrish and Smith27 who reported that poor solvents cause an earlier gel effect in methyl methacrylate polymerization. Recent careful studies of the gel effect by Torkelson and co-workers28 have reinforced observations made by Cameron and Cameron29 over two decades ago concluding that termination is hindered in poor solvents due to formation of more tightly coiled polymer radicals. 13.2.3 COPOLYMERIZATION When solvent effects on the propagation step occur in free-radical copolymerization reactions, they result not only in deviations from the expected overall propagation rate, but also in deviations from the expected copolymer composition and microstructure. This may be
780
Michelle L. Coote and Thomas P. Davis
true even in bulk copolymerization, if either of the monomers exerts a direct effect or if strong cosolvency behavior causes preferential solvation. A number of models have been proposed to describe the effect of solvents on the composition, microstructure and propagation rate of copolymerization. In deriving each of these models, an appropriate base model for copolymerization kinetics is selected (such as the terminal model or the implicit or explicit penultimate models), and a mechanism by which the solvent influences the propagation step is assumed. The main mechanisms by which the solvent (which may be one or both of the comonomers) can affect the propagation kinetics of free-radical copolymerization reactions are as follows: (1) Polarity effect (2) Radical-solvent complexes (3) Monomer-solvent complexes (4) Bootstrap effect In this chapter we explain the origin of these effects, show how copolymerization models for these different effects may be derived, and review the main experimental evidence for and against these models. Throughout this review the baseline model for copolymerization is taken as the terminal or Mayo-Lewis model.30 This model can be used to derive well-known expressions for copolymer composition and copolymerization propagation kinetics. Deviations from this model have often been interpreted in terms of either solvent effects or penultimate unit effects, although the two are by no means mutually exclusive. Deviations which affect both the copolymer composition and propagation kinetics have been termed explicit effects by Fukuda31 in deriving penultimate unit models, whereas deviations from the kinetics without influencing the copolymer composition have been termed implicit effects. In this review we use the same terminology with respect to solvent effects: that is, a solvent effect on kp only is termed an implicit solvent effect, while a solvent effect on composition, microstructure and kp is termed explicit. The relatively recent discovery by Fukuda and co-workers32 of the seemingly general failure of the terminal model to predict kp, even for bulk copolymerizations that follow the terminal model composition equation, led them to propose an implicit penultimate unit effect as a general phenomenon in free-radical copolymerization kinetics. We conclude this review with a brief examination of the possibility that a implicit solvent effect, and not an implicit penultimate unit effect, may instead be responsible for this failure of the terminal model kp equation. 13.2.3.1 Polarity effect 13.2.3.1.1 Basic mechanism One type of solvent effect on free-radical addition reactions such as the propagation step of free-radical polymerization is the so-called ‘polarity effect’. This type of solvent effect is distinguished from other solvent effects, such as complexation, in that the solvent affects the reactivity of the different types of propagation steps without directly participating in the reaction. The mechanism by which this could occur may be explained as follows. The transition states of the different types of propagation steps in a free-radical copolymerization may be stabilized by charge transfer between the reacting species. The amount of charge transfer, and hence the amount of stabilization, is inversely proportional to the energy difference between the charge transfer configuration, and the product and reactant configurations that combine to make up the wave function at the transition state.33 Clearly, the stability of the charge transfer configuration would differ between the cross- and
13.2 Solvent effects on free radical polymerization
781
homopropagation reactions, especially in copolymerization of highly electrophilic and nucleophilic monomer pairs. Hence, when it is significant, charge transfer stabilization of the transition state occurs to different extent in the cross- and homopropagation reactions, and thus exerts some net effect on the monomer reactivity ratios. Now, it is known that polar solvents can stabilize charged species, as seen in the favorable effect of polar solvents on both the thermodynamics and kinetics of reactions in which charge is generated.34 Therefore, when charge transfer in the transition state is significant, the stability of the charge transfer species and thus the transition state would be affected by the polarity of the solvent, and thus a solvent effect on reactivity ratios would result. 13.2.3.1.2 Copolymerization model There are two cases to consider when predicting the effect of solvent polarity on copolymerization propagation kinetics: (1) the solvent polarity is dominated by an added solvent and polarity is thus independent of the comonomer feed ratio, or (2) the solvent polarity does depend on the comonomer feed ratio, as it would in a bulk copolymerization. In the first case, the effect on copolymerization kinetics is simple. The monomer reactivity ratios (and additional reactivity ratios, depending on which copolymerization model is appropriate for that system) would vary from solvent to solvent, but, for a given copolymerization system they would be constant as a function of the monomer feed ratios. Assuming of course that there were no additional types of solvent effect present, these copolymerization systems could be described by their appropriate base model (such as the terminal model or the explicit or implicit penultimate models), depending on the chemical structure of the monomers. In the second case, the effect of the solvent on copolymerization kinetics is much more complicated. Since the polarity of the reacting medium would vary as a function of the comonomer feed ratios, the monomer reactivity ratios would no longer be constant for a given copolymerization system. To model such behavior, it would be first necessary to select an appropriate base model for the copolymerization, depending on the chemical structure of the monomers. It would then be necessary to replace the constant reactivity ratios in this model by functions of the composition of the comonomer mixture. These functions would need to relate the reactivity ratios to the solvent polarity, and then the solvent polarity to the comonomer feed composition. The overall copolymerization kinetics would therefore be very complicated, and it is difficult to suggest a general kinetic model to describe these systems. However, it is obvious that such solvent effects would cause deviations from the behavior predicted by their appropriate base model and might therefore account for the deviation of some copolymerization systems from the terminal model composition equation. 13.2.3.1.3 Evidence for polarity effects in propagation reactions The idea of charge separation in the transition state of the propagation step of free radical polymerization reactions, as suggested by Price,35 was discounted by Mayo and Walling36 and many subsequent workers.37 Their rejection of this idea was based upon the absence of any unambiguous correlation between the reactivity ratios of a system and the dielectric constant of the solvent. For instance, in the copolymerization of STY with MMA, it was reported that the reactivity ratios were independent of small quantities of water, ethyl benzene, dodecylmercaptans or hydroquinone, or the presence or absence of air30,38,39 and were thus unaffected by the dielectric constant of the system. In contrast, other studies have found a relationship between dielectric constant and the reactivity ratios in specific systems.40,41
782
Michelle L. Coote and Thomas P. Davis
The apparent lack of a general relationship between the dielectric constant of the system and the monomer reactivity ratios does not necessarily discount a polarity effect on reactivity ratios. A polarity effect is only expected to occur if the charge transfer configurations of the transition state are sufficiently low in energy to contribute to the ground state wave function. Since this is not likely to occur generally, a comprehensive correlation between reactivity ratios and the solvent dielectric constant is unlikely. Furthermore, even in systems for which a polarity effect is operating, a correlation between solvent dielectric constant and monomer reactivity ratios may be obscured by any of the following causes. • The operation of additional types of solvent effect, such as a Bootstrap effect, that would complicate the relationship between solvent polarity and reactivity ratios. • Errors in the experimental data base from which the correlation was sought. • The recognized inadequacy of simple reactivity - dielectric constant correlations, that take no account of specific interactions between the solvent and solute molecules.34 In fact, recent theoretical33 and experimental studies42 of small radical addition reactions indicate that charge separation does occur in the transition state when highly electrophilic and nucleophilic species are involved. It is also known that copolymerization of electron donor-acceptor monomer pairs are solvent sensitive, although this solvent effect has in the past been attributed to other causes, such as a Bootstrap effect (see Section 13.2.3.4). Examples of this type include the copolymerization of styrene with maleic anhydride43 and with acrylonitrile.44 Hence, in these systems, the variation in reactivity ratios with the solvent may (at least in part) be caused by the variation of the polarity of the solvent. In any case, this type of solvent effect cannot be discounted, and should thus be considered when analyzing the copolymerization data of systems involving strongly electrophilic and nucleophilic monomer pairs. 13.2.3.2 Radical-solvent complexes 13.2.3.2.1 Basic mechanism Solvents can also interfere in the propagation step via the formation of radical-solvent complexes. When complexation occurs, the complexed radicals are more stable than their corresponding uncomplexed-radicals, as it is this stabilization that drives the complexation reaction. Thus, in general, one might expect complexed radicals to propagate more slowly than their corresponding free-radicals, if indeed they propagate at all. However, in the special case that one of the comonomers is the complexing agent, the propagation rate of the complexed radical may instead be enhanced, if propagation through the complex offers an alternative less-energetic reaction pathway. In any case, the complexed radicals would be expected to propagate at a rate different to their corresponding free-radicals, and thus the formation of radical-solvent complexes would affect the copolymerization propagation kinetics. 13.2.3.2.2 Copolymerization model A terminal radical-complex model for copolymerization was formulated by Kamachi.37 He proposed that a complex is formed between the propagating radical chain and the solvent (which may be the monomer) and that this complexed radical has a different propagation rate constant to the equivalent uncomplexed radical. Under these conditions there are eight different propagation reactions in a binary copolymerization, assuming that the terminal unit is the only unit of the chain affecting the radical reactivity. These are as follows.
13.2 Solvent effects on free radical polymerization
783
k
ij RMi ⋅ +Mj → RMiMj ⋅ or RMiMj ⋅ S where: i, j = 1or 2
k
cij → RMiMj ⋅ or RMiMj ⋅ S where: i, j = 1or 2 RMi ⋅ S + Mj
There are also two equilibrium reactions for the formation of the complex: Ki RMi ⋅ +S ← → RMi ⋅ S where: i, j = 1or 2
Applying the quasi-steady-state and long-chain assumptions to the above reactions, Kamachi derived expressions for ri and kii, which are used in place of ri and kii in the terminal model equations for composition and kp: k ii = k ii
1 + s ci K i [Ci ] 1 + K i [Ci ]
and
ri = ri
1 + s ci K i [Ci ]
1 + (ri / rci )s ci K i [Ci ]
where: ri = kii/kij; ric = kcii/kcij; sci = kcii/kii; i, j = 1 or 2 and i ≠ j
Variants of this model may be derived by assuming an alternative basis model (such as the implicit or explicit penultimate models) or by making further assumptions as to nature of the complexation reaction or the behavior of the complexed radical. For instance, in the special case that the complexed radicals do not propagate (that is, sci = 0 for all i), the reactivity ratios are not affected (that is, ri = ri for all i) and the complex formation serves only removal of radicals (and monomer, if monomer is the complexing agent) from the reaction, resulting in a solvent effect that is analogous to a Bootstrap effect (see Section 13.2.3.4). 13.2.3.2.3 Experimental evidence There is certainly strong experimental evidence for the existence of radical-solvent complexes. For instance, Russell45-47 and co-workers collected experimental evidence for radical-complex formation in studies of the photochlorination of 2,3-dimethylbutane in various solvents. In this work, different products were obtained in aliphatic and aromatic solvents, and this was attributed to formation of a π-complex between the Cl atom and the aromatic solvent. Complex formation was confirmed by flash photolysis.48-51 Complex formation was also proposed to explain experimental results for the addition of trichloromethane radical to 3-phenylpropene and to 4-phenyl-1-butene52 and for hydrogen abstraction of the t-butoxy radical from 2,3-dimethylbutane.53 Furthermore, complexes between nitroxide radicals and a large number of aromatic solvents have been detected.54-57 Evidence for complexes between polymer radicals and solvent molecules was collected by Hatada et al.,58 in an analysis of initiator fragments from the polymerization of MMA-d with AIBN and BPO initiators. They discovered that the ratio of disproportionation to combination depended on the solvent, and interpreted this as evidence for the formation of a polymer radical-solvent complex that suppresses the disproportionation reaction. There is also experimental evidence for the influence of radical-solvent complexes in small radical addition reactions. For instance, Busfield and co-workers59-61 used radical-solvent to explain solvent effects in reactions involving small radicals, such as t-butoxyl radicals towards various electron donor-electron acceptor monomer pairs. The observed solvent effects were interpreted in terms of complex formation between the t-butoxyl radical and the electron-acceptor monomer, possibly via a sharing of the lone pair on the t-butoxyl oxy-
784
Michelle L. Coote and Thomas P. Davis
gen with the π-system of the acceptor monomer. Several workers have invoked frontier orbital theory to rationalize such solvent effects in terms of radical-solvent complex formation, and thus provide a theoretical base.37,62 Many workers have suggested radical-solvent complexes as an explanation for the influence of aromatic compounds on the homopolymerization of vinyl monomers. For instance, Mayo63 found that bromobenzene acts as a chain transfer agent in the polymerization of STY but is not incorporated into the polymer. He concluded that a complex is formed between the solvent molecule and either the propagating polystyryl radical or a proton derived from it. The influence of halobenzenes on the rate of polymerization of MMA was detected by Burnett et al.8,64,65 They proposed that the efficiency of a number of different initiators increased in various halogenated aromatic solvents and, since enhanced initiator or solvent incorporation into the polymer was not observed, they concluded that initiator-solvent-monomer complex participation affected the initiator efficiency. Henrici-Olive and Olive66-71 suggested that this mechanism was inadequate when the degree of polymerization was taken into account and they proposed instead a charge transfer complex between the polymer radical and aromatic solvent. The polymer radical can form a complex with either the monomer or solvent molecule, but only the former can propagate. Bamford and Brumby,5 and later Burnett et al.,72,73 interpreted their solvent-effects data for kp in terms of this donor-acceptor complex formation between aromatic solvents and propagating radicals. Radical-solvent complexes are expected to be favored in systems containing unstable radical intermediates (such as vinyl acetate) where complexation may lead to stabilization. In this regard Kamachi et al18 have noted that solvent effects on vinyl acetate homopolymerization result in a reduced kp. Kamachi et al.74 also measured the absolute rate constants of vinyl benzoate in various aromatic solvents and found that kp increased in the order: benzonitrile < ethyl benzoate < anisole < chlorobenzene < benzene < fluorobenzene < ethyl acetate
They argued that this trend could not be explained by copolymerization through the solvent or transfer to the solvent because there was no correlation with the solvent dielectric constant or polarity, or with the rate constants for transfer to solvent. However, there was a correlation with the calculated delocalization stabilization energy for complexes between the radical and the solvent, which suggested that the propagating radical was stabilized by the solvent or monomer, but the solvent did not actually participate in the reaction. As noted in the introduction to this section, radical-solvent complexes may enhance the propagation rate if propagation through the complex offers an alternative, less-energetic pathway for propagation. An example of this behavior is found in the homopolymerization of acrylamide. The homopropagation rate coefficient for this monomer shows a negative temperature dependence, which has been explained in terms of radical-complex formation. Pascal et al.11,75 suggested that propagation proceeds via a complex that enhances the propagation rate, and this complex dissociates as temperature increases, thus explaining the normal temperature dependence of the propagation rate at high temperatures. This interpretation was supported by the observation that acrylamide behaves normally in the presence of reagents such as propionamide, which would be expected to inhibit complex formation. Given the experimental evidence for the existence of radical-solvent complexes and their influence on free-radical addition reactions such as homopropagation, it is likely that
13.2 Solvent effects on free radical polymerization
785
radical-solvent complexes will affect the copolymerization kinetics for certain copolymerization systems, and indeed many workers have invoked the radical-complex model in order to explain solvent effects in copolymerization. For instance, Heublein and Heublein76 have invoked a radical complex model in combination with a partitioning idea (see Section 13.2.3.4) to explain solvent effects on the copolymerization of vinyl acetate with acrylic acid. More recently, O’Driscoll and Monteiro77 suggested that the effect of benzyl alcohol on the copolymerization of STY-MMA was best described by an RC-type model. This was supported by pulsed-laser studies9 on the homopropagation reactions where Ea values were found to be increased slightly by the presence of benzyl alcohol. Czerwinski (see for example reference78 and references cited therein) has also published a variant of the RC model and has applied his model to a range of copolymerization experimental data. In conclusion, there is a strong experimental evidence for the importance of radical-solvent complexes in a number of specific copolymerization systems, especially when there is a large disparity in the relative stabilities of the different propagating radicals. 13.2.3.3 Monomer-solvent complexes 13.2.3.3.1 Introduction A solvent may also interfere in the propagation step via complexation with the monomer. As was the case with radical-solvent complexes, complexed monomer might be expected to propagate at a different rate to free monomer, since complexation might stabilize the monomer, alter its steric properties and/or provide an alternative pathway for propagation. In examining the effect of such complexation on copolymerization kinetics, there are a number of different mechanisms to consider. In the case that the complex is formed between the comonomers, there are three alternatives: (1) the monomer-monomer complex propagates as a single unit, competing with the propagation of free monomer; (2) the monomer-monomer complex propagates as a single unit, competing with the propagation of free monomer, but the complex dissociates during the propagation step and only one of the monomers is incorporated into the growing polymer radical; (3) the monomer-monomer complex does not propagate, and complexation serves only to alter the free monomer concentrations. In the case that the complex is formed between one of the monomers and an added solvent, there are two further mechanisms to consider: (4) the complexed monomer propagates, but at a different rate to the free monomer; (5) the complexed monomer does not propagate. Models based on mechanisms (1) and (2) are known as the monomer-monomer complex participation (MCP) and dissociation (MCD) models, respectively. Mechanisms (3) and (5) would result in a solvent effect analogous to a Bootstrap effect, and will be discussed in Section 13.2.3.4. In this section, we review the MCP and MCD models, and conclude with a brief discussion of specific monomer-solvent interactions. 13.2.3.3.2 Monomer-monomer complex participation model The use of monomer-monomer charge transfer complexes to explain deviations from the terminal model was first suggested by Bartlet and Nozaki,79 later developed by Seiner and Litt,80 and refined by Cais et al.81 It was proposed that two monomers can form a 1:1 donor complex and add to the propagating chain as a single unit in either direction. The complex would be more reactive because it would have a higher polarizability due to its larger π-electron system that can interact more readily with the incoming radical. The complex would also have a higher pre-exponential factor, as a successful attack may be achieved over a
786
Michelle L. Coote and Thomas P. Davis
wider solid angle.80 The heavier mass of the complex would also serve to increase the pre-exponential factor. In addition to the four terminal model reactions, four complex addition reactions and an equilibrium constant are required to describe the system. k
ij → RMiMj ⋅ where: i, j = 1or 2 RMi ⋅ +Mj o
k
iij → RMiMiMj ⋅ where: i, j = 1or 2 and i ≠ j RMi ⋅ +MiMj
k
iji → RMiMjMi ⋅ where: i, j = 1or 2 and i ≠ j RMi ⋅ +MiMj
K → M1M 2 M1o + M 2 o ←
The composition and propagation rate can be expressed in terms of the following parameters.31 o F1 f1 (A2 B1 )r1f1 + (A1C2 )f 2 = o F2 f 2 (A1B 2 )r2 f 2 o + (A2 C1 )f1 o o
kp =
(A2 B1 )r1 (f 2 o )
2
o
( ) + (A C + A C )f ) + (A r f / k )
+ (A1B 2 )r2 f 2
(A r f
2 11
o
/ k11
o
2
1
2
2
1
1
o
f2
o
o
1 2 2
22
where: A1 = 1 + r1s1cQf1q and A2 = 1 + r2s2cQf2q B1 = 1 + s1c(1 + r1c-1)Qf2q and B2 = 1 + s2c(1 + r2c-1)Qf1q C1 = 1 + r1s1c(1 + r1c-1)Qf1q and C2 = 1+ r2s2c(1 + r2c-1)Qf2q 2Qfiq = {[Q(fj - fi) +1]2 +4Qfi}1/2 - [Q(fi - fj) +1] and Q = k[M] fi is feed composition of Mi fiq = [Miq]/[M] ri = kii/kij; ric = kiij/kiji; sic = kiij/kii where: i,j = 1 or 2 and i ≠ j
The applicability of the MCP model to strongly alternating copolymerization has been a long standing point of contention. In essence, there are two opposing accounts of the strongly alternating behavior observed in copolymers of electron-donor-acceptor (EDA) monomer pairs. In the first account, this behavior has been attributed to the fact that the transition state is stabilized in cross-propagation reaction and destabilized in the homopropagation. Deviations from the terminal model are caused merely by penultimate unit effects. In the second account -the MCP model- the strongly alternating behavior is a result of propagation of a 1:1 comonomer complex which, as seen above, also leads to deviations from the terminal model. An intermediate mechanism, which will be discussed shortly, is the MCD model in which the complex dissociates during the propagation step. The main approach to discriminating between these models has been to compare their ability to describe the copolymerization data of various explicit systems, and to study the effect of added solvents on their behavior. Unfortunately, both approaches have led to inconclusive results.
13.2 Solvent effects on free radical polymerization
787
As an example, the system STY with maleic anhydride (MAH) has been perhaps the most widely studied EDA system and yet there is still uncertainty concerning the role of the EDA complex in its propagation mechanism. Early studies82,83 concluded that its behavior was best modelled by a penultimate model, despite the spectroscopic evidence for EDA complexes in this system. Later Tsuchida et al.84,85 fitted an MCP model, based on the evidence that the rate went through a maximum at 1:1 feed ratio in benzene or CCl4 but in strong donor solvents no such maximum occurred and instead the rate increased with the content of MAH in the feed. They argued that maximum in rate at 1:1 feed ratios was due to the fact that propagation occurred via the complex, which had a maximum concentration at this point. In strong donor solvents, the maximum rate moved to higher concentrations of MAH due to competition between the donor and STY for complexation with the MAH. However, a few years later, Dodgson and Ebdon86,87 conducted an extensive study of STY-MAH in various solvents and discounted the MCP model on the basis of an absence of a dilution effect with the inert solvent MEK. In an MCP model a dilution effect would be expected due to the decrease in the relative concentration of the comonomer complex and the enhanced participation of the free monomer.88 Later, Farmer et al.89 reanalyzed this data and concluded that the composition data was consistent with both models and suggested sequence distribution may provide the answer. They also pointed out that there was a small dilution effect in MEK -greater than that predicted by the penultimate model and less than that predicted by the MCP model. Hill et al.90 has suggested that interpretation of the effect of solvents is complicated by the fact that no solvent is truly inert, hence such results such be treated with caution. More recently Sanayei et al.91 have performed a pulsed-laser polymerization study on STY-MAH copolymerization in butanone and acetonitrile. They concluded that whilst the complex participation model described the copolymer composition it failed to predict the average kp data. Consequently the best description of this copolymerization was given by the penultimate unit model. There have been many other systems for which the MCP model has been proposed as an alternative to the penultimate unit model. For instance, Litt and Seiner used the MCP model to describe the composition of a number of systems, including MAH with 1-diphenylethylene, β-cyanacrolein with styrene,92 and vinyl acetate with dichlorotetrafluoroacetone and with hexafluoroacetone.80 An MCP model has also been suggested for the system STY-SO2.39,83,93-95 In this system, the composition changes with dilution or with solvent changes, strongly alternating behavior is observed across a range of feed ratios, and one of the comonomers (SO2) does not undergo homopolymerization. However, while the MCP model appears to be appropriate for some systems, in other strongly alternating copolymerizations it is clearly not appropriate. For instance, there are many strongly alternating copolymerizations for which there is no evidence of complex formation.36,39,88,96 Even when complex formation is known to occur, results cannot always be explained by the MCP model. For instance, measurements of sequence distribution data revealed that, while both the MCP and penultimate model could provide an adequate description of the composition of STY with acrylonitrile (AN), only the penultimate model could account for the sequence distribution data for this system.97 As will be seen shortly, there is evidence that in some systems the heat of propagation would be sufficient to dissociate the EDA complex and hence it could not add to the monomer as unit. In this case an MCD model would be more appropriate. Thus, it might be concluded that the MCP, MCD and the penultimate models are needed to describe the behavior of strongly alternating systems, and
788
Michelle L. Coote and Thomas P. Davis
the selection of each model should be on a case-by-case basis. There have been many more systems for which the MCP model has been evaluated against the penultimate model on the basis of kinetic behavior. These studies have been extensively reviewed by Hill et al.90 and Cowie98 will therefore not be reviewed here. Instead a few additional sources of evidence for the participation of the EDA complex will be highlighted. UV and NMR evidence for the existence of EDA complexes There is certainly a demonstrable existence of comonomer complexes in solutions of electron donor acceptor monomer pairs. These complexes can be detected, and their equilibrium constants measured, using UV or NMR spectroscopy. Techniques for this are described in detail in reviews of comonomer complexes by Cowie98 and Hill et al.90 The latter review90 also includes a listing of the equilibrium constants for the numerous EDA complexes that have been experimentally detected. The existence of comonomer complexes is not sufficient evidence for their participation in the propagation step of copolymerization, but the fact that they exist in solutions from which strongly alternating copolymers are produced suggests that they play some role in the mechanism. Furthermore, the ability to measure their strengths and quantify the effects of solvents on their observed equilibrium constants without performing kinetic experiments, may provide the key to establishing their role in the propagation mechanism. Since the alternative models for copolymerization include (or in some cases omit) the equilibrium constant for these complexes in different ways; if the equilibrium constant was to be measured separately and not treated as an adjustable parameter in the kinetic analysis, more sensitive model discrimination would be possible. To date, such an analysis does not appear to have been performed but it should be included in subsequent kinetic analyses of these explicit systems. Temperature effects The study of the temperature dependence of copolymerization behavior may also provide evidence for the role of comonomer complexes. As was seen previously in the study of acrylamide, complexes dissociate at high temperatures and hence, if the complex is involved in controlling an aspect of the polymerization behavior, a change in this behavior should be observed at the temperature corresponding to the complete dissociation of the complex. Such evidence has been obtained by Seymour and Garner99,100 for the copolymerization of MAH with a variety of vinyl monomers, including STY, VA, AN, and α-MSTY. They observed that the copolymers undergo a change from strongly alternating to random at high temperatures, and these temperatures are also the temperatures at which the concentration of the EDA complex becomes vanishingly small. It is true that, since reactivity ratios have an enthalpy component, they approach unity as temperature increases. Hence, most models predict that the tendency of copolymers to form a random microstructure increases as temperature increased. Indeed, more recent work by Klumperman101 has shown that for STY-MAH copolymerization, the reactivity ratios do follow an Arrhenius type of temperature dependence. However, further work is required to verify this for the other copolymerization listed above. Based upon the existing copolymerization data, it appears that for many systems there are sudden transition temperatures that correspond to the dissociation of the complex, which does suggests that the complex is in some way responsible for the alternating behavior.99,100 Stereochemical evidence for the participation of the complex Stereochemical data may provide evidence for participation of the EDA complex. The EDA complex will prefer a certain geometry -that conformation in which there is maximum
13.2 Solvent effects on free radical polymerization
789
overlap between the highest occupied molecular orbital (HOMO) of the donor and the lowest unoccupied molecular orbital (LUMO) of the acceptor.102 If the complex adds to the propagating polymer chain as a unit, then this stereochemistry would be preserved in the polymer chain. If, however, only free monomer addition occurs, then the stereochemistry of the chain should be completely random (assuming of course that there are no penultimate unit effects operating). Hence, it is possible to test for the participation of the monomer complexes in the addition reaction by examination of the stereochemistry of the resulting polymer. Such stereochemical evidence has been collected by a number of workers. For instance, Iwatsuki and Yamashia103 observed an unusually high percentage of cis units in MAH/butadiene copolymers. Olson and Butler104 studied the EDA system N-phenylmaleimide (NPM)/2-chloroethyl vinyl ether (CEVE) and found that the stereochemistry at succinimide units in NPM-CEVE copolymers is predominantly cis, and random elsewhere. Furthermore, they noted that the proportion of cis units was correlated with those variables with which the concentration of the EDA complex was also correlated. In these examples, the cis geometry is that which is most stable for the complex. However, Rätzsch and Steinert105 have argued that this preference for cis geometry may also be explained by propagation occurring via a complex between the reacting free monomer and the chain end, as in an RC model. Thus this evidence should be used in conjunction with other evidence for model discrimination. Further stereochemical evidence for the MCP model has been obtained by Butler et al.106 They predicted that the usual preference for head-tail addition in free-radical polymerization would be overcome if propagation occurred via the EDA complex, and its favored geometry was a head-head conformation. They noted that for most EDA pairs head-tail geometry was favored and hence the predominance of head-tail linkages in these copolymers could not discriminate between free monomer addition and complex participation. To solve this problem, they designed and synthesized two monomer pairs for which a head-head conformation would be expected in their EDA complexes. These were the systems dimethyl cyanoethylene dicarboxylate (DMCE) with CEVE, and dimethyl cyanoethylene dicarboxylate (DMCE) with CEVE. They then showed that there were significant head-head linkages in the resulting copolymer and the proportion of these linkages was correlated with same types of variables that had previously affected the cis content of NPM/CEVE copolymers -that is, those variables which affected the concentration of the EDA complex. Thus they concluded that there was strong stereochemical evidence for the participation of the EDA complex in the propagation step. ESR evidence for the participation of the complex ESR studies have also been suggested as a means for providing information about the participation of the EDA complex. Since the addition of the complex is likely to occur more readily in one direction, if propagation occurs as the repeated addition of the complex then the propagating radical should be predominantly of one type. However, if free monomer addition predominantly occurs, both types of radical are likely to be present at any time. Thus ESR can be used to distinguish between the two mechanisms. This approach was used by Smirnov et al.107 to show that, in the system phenyl vinyl ether/MAH, alternating addition of the free monomer predominates, but participation of EDA complexes is important for the system butyl vinyl ether/MAH. They argued that the difference in the behavior of the two EDA systems was a result of the different strengths of their EDA complexes. In another
790
Michelle L. Coote and Thomas P. Davis
study, Golubev et al.,108 used ESR to show that for dimethylbutadiene/MAH the cross-propagation of the free monomers dominated. However, Barton et al.109 has questioned the assignments of ESR signals in the previous studies and suggested that the ESR evidence was inconclusive. Furthermore, the predominance of one type of ESR signal may also be explained without invoking the MCP model. Assuming that cross-propagation is the dominant reaction, and that one of the radicals is much less stable than the other, it might reasonably be expected that the less stable radical would undergo fast cross-propagation into the more stable radical, resulting in an ESR signal dominated by the more stable radical. Hence it appears that ESR is not able to discriminate between this and the MCP mechanism. 13.2.3.3.3 Monomer-monomer complex dissociation model Tsuchida and Tomono84 suggested that the monomer-monomer complex described in the MCP model may dissociate upon addition to the chain, with only one unit adding. The concept was developed by Karad and Schneider110 who argued that the dissociation of the complex is likely since its heat of formation is typically less than the heat of propagation. As an example, they measured the heat of formation for a STY/fumaronitrile complex, and found that it was only 1.6 kcal/mol, significantly less than the heat of propagation (15-20 kcal/mol). Under a complex-dissociation mechanism, the role of the complex is merely to modify the reactivity of the reactant monomers. A model based on the complex-dissociation mechanism was first formulated by Karad and Schneider110 and generalized by Hill et al.111 Again, eight rate constants and two equilibrium constants are required to describe the system. k
ij → RMiMj ⋅ where: i, j = 1or 2 RMi ⋅ +Mj
k
ijc → RMiMj ⋅ where: i, j = 1or 2 RMi ⋅ +MjC
Ki → MiC Mi + C ←
where: i, j = 1or 2
As for the previous models, expressions for kp and composition can be derived in terms of these parameters by first calculating kii and ri and then using them in place of kii and ri in the terminal model equations. The relevant formulae are: k ii = k ii
1 + s ic K i [Ci ] 1 + K i [Ci ]
and ri = ri
1 + s ic K i [Ci ]
1 + (ri / rci )s ic K i [Ci ]
where: ri = kii/kij; ric = kiic/kijc; sic = kiic/kii; i,j = 1 or 2 and i ≠ j
Efforts to compare this model with the MCP model have been hindered by the fact that similar composition curves for a given system are predicted by both models. Hill et al.111 showed that the composition data of Dodgson and Ebdon87 for STY/MAH at 60°C could be equally well described by the MCP, MCD or penultimate unit models. They suggested that sequence distribution would be a more sensitive tool for discriminating between these models. One study which lends some support to this model over the MCP model for describing this system was published by Rätzsch and Steinert.105 Using Giese’s112 ‘mercury method’ to study the addition of monomers to primary radicals, they found that in mixtures of MAH and STY, only reaction products from the addition of free monomers, and not the EDA
13.2 Solvent effects on free radical polymerization
791
complex, to primary cyclohexyl radicals were found. Thus they concluded that the STY/MAH complex in the monomer solution is disrupted during the propagation step. It is likely that both the MCP and MCD mechanisms are valid and their validity in a specific system will depend on the relative strength of the EDA complex concerned. The MCD model may be useful for accounting for those systems, in which EDA complexes are known to be present but the MCP model has been shown not to hold. 13.2.3.3.4 Specific solvent effects Several monomers are particularly susceptible to strong solvent effects via specific interactions such as hydrogen bonding, ionic strength and pH. The kinetic consequences of these specific interactions will vary from system to system. In some cases the radical and/or monomer reactivity will be altered and in other cases a Bootstrap effect will be evident. It is worth noting that monomers which are susceptible to strong medium effects will not have reliable Q-e values, a good example of this is 2-hydroxyethyl methacrylate (HEMA) where there is a large variation in reported values. The reactivity ratios of HEMA with STY have been reported to be strongly dependent on the medium,113 similarly the copolymerization of HEMA with lauryl methacrylate is solvent sensitive;114 behavior which has been attributed to non-ideal solution thermodynamics (cf Semchikov’s work in Section 13.2.3.4). Chapiro115 has published extensively on the formation of molecular associates in copolymerization involving polar monomers. Other common monomers which show strong solvent effects are N-vinyl-2-pyrrolidone, (meth)acrylic acids and vinyl pyridines. 13.2.3.4 Bootstrap model 13.2.3.4.1 Basic mechanism In the Bootstrap model, solvent effects on kp are attributed to solvent partitioning and the resulting difference between bulk and local monomer concentrations. In this way, a solvent could affect the measured kp without changing the reactivity of the propagation step. Bootstrap effects may arise from a number of different causes. As noted previously, when radical-solvent and monomer-solvent complexes form and the complexes do not propagate, the effect of complexation is to alter the effective radical or monomer concentrations, thereby causing a Bootstrap effect. Alternatively, a Bootstrap effect may arise from some bulk preferential sorption of one of the comonomers around the growing (and dead) polymer chains. This might be expected to occur if one of monomers is a poor solvent for its resulting polymer. A Bootstrap effect may also arise from a more localized from of preferential sorption, in which one of the monomers preferentially solvates the active chain end, rather than the entire polymer chain. In all cases, the result is the same: the effective free monomer and/or radical concentrations differ from those calculated from the monomer feed ratios, leading to a discrepancy between the predicted and actual propagation rates. 13.2.3.4.2 Copolymerization model Copolymerization models based upon a Bootstrap effect were first proposed by Harwood116 and Semchikov117 (see references cited therein). Harwood suggested that the terminal model could be extended by the incorporation of an additional equilibrium constant relating the effective and ‘bulk’ monomer feed ratios. Different versions of this so-called Bootstrap model may be derived depending upon the baseline model assumed (such as the terminal model or the implicit or explicit penultimate models) and the form of equilibrium expression used to represent the Bootstrap effect. In the simplest case, it is assumed that the magni-
792
Michelle L. Coote and Thomas P. Davis
tude of the Bootstrap effect is independent of the comonomer feed ratios. Hence in a bulk copolymerization, the monomer partitioning may be represented by the following equilibrium expression: f f1 = K 1bulk f2 f 2 bulk
The equilibrium constant K may be considered as a measure of the Bootstrap effect. Using this expression to eliminate the effective monomer fractions (f1 and f2) from the terminal model equations, replacing them with the measurable ‘bulk’ fractions (f1bulk and f2bulk), the following equations for composition43 and kp118 may be derived. F1 Kf1bulk r1Kf1bulk + f 2 bulk = F2 f 2 bulk r2 f 2 bulk + Kf1bulk kp =
r1K 2 f12bulk + 2Kf1bulk f 2 bulk + r2 f 22bulk 1 Kf1bulk + f 2 bulk r1Kf1bulk / k11 + r2 f 2 bulk / k 22
Examining the composition and kp equations above, it is seen that the Bootstrap effect K is always aliased with one of the monomer feed ratios (that is, both equations may be expressed in terms of Kf1 and f2). It is also seen that once Kf1 is taken as a single variable, the composition equation has the same functional form as the terminal model composition equation, but the kp equation does not. Hence it may seen that, for this version of the Bootstrap effect, the effect is an implicit effect - causing deviation from the terminal model kp equation only. It may also be noted that, if K is allowed to vary as a function of the monomer feed ratios, the composition equation also will deviate from terminal model behavior - and an explicit effect will result. Hence it may be seen that it is possible to formulate an implicit Bootstrap model (that mimics the implicit penultimate model) but in order to do this, it must be assumed that the Bootstrap effect K is constant as a function of monomer feed ratios. It should be noted that the above equations are applicable to a bulk copolymerization. When modelling solution copolymerization under the same conditions, the equations may be used for predicting copolymer composition since it is only the relationship between bulk and local monomer feed ratios that determines the effect on the composition and microstructure of the resulting polymer. However, some additional information about the net partitioning of monomer and solvent between the bulk and local phases is required before kp can be modelled. It should be observed that in a low-conversion bulk copolymerization, knowledge of the monomer feed ratios automatically implies knowledge of the individual monomer concentrations since, as there are no other components in the system, the sum of the monomer fractions is unity. However, in a solution copolymerization there is a third component - the solvent - and the monomer concentrations depend not only upon their feed ratio but also upon the solvent concentration. Modelling kp in a solution copolymerization could be achieved by re-writing the above equilibrium expression in terms of molar concentrations (rather than comonomer feed ratios), and including the solvent concentration in this expression.
13.2 Solvent effects on free radical polymerization
793
The Bootstrap model may also be extended by assuming an alternative model (such as the explicit penultimate model) as the baseline model, and also by allowing the Bootstrap effect to vary as a function of monomer feed ratios. Closed expressions for composition and sequence distribution under some of these extended Bootstrap models may be found in papers by Klumperman and co-workers.43,44 13.2.3.4.3 Experimental evidence The Bootstrap model was introduced by Harwood,116 who studied three solvent sensitive copolymerizations (styrene/methacrylic acid, styrene/acrylic acid and styrene/acrylamide) and found that the copolymers of the same composition had the same sequence distribution irrespective of the solvent used. This meant that the conditional probabilities governing radical propagation were independent of the solvent. On this basis, he argued that composition and sequence distribution were deviating from their expected values because there was a difference between the monomer feed ratios in the vicinity of the active chain end, and those calculated on the basis of the bulk feed. In other words, the solvent was altering the rates of the individual propagation steps by affecting the reactant concentrations and not, as in the other solvent effects models, their reactivities. However, Fukuda et al.31 have argued that the NMR evidence provided by Harwood is not conclusive evidence for the Bootstrap model, since Harwood’s observations could also be described by variation of the reactivity ratios in such a way that their product (r1r2) remains constant. This has also been raised as an issue by Klumperman and O’Driscoll.43 They showed mathematically that a variation in the local comonomer ratio is not reflected in the monomer sequence distribution versus copolymer composition -this relationship being governed by the r1r2 product only. An alternative explanation for Harwood’s experimental data may be the stabilization or destabilization of the radicals by the solvent, an interpretation that would be analogous to the MCD model. Simple energy stabilization considerations, as used by Fukuda et al.119 to derive the penultimate unit effect, also suggest the constancy of r1r2. Prior to Harwood’s work, the existence of a Bootstrap effect in copolymerization was considered but rejected after the failure of efforts to correlate polymer-solvent interaction parameters with observed solvent effects. Kamachi,37 for instance, estimated the interaction between polymer and solvent by calculating the difference between their solubility parameters. He found that while there was some correlation between polymer-solvent interaction parameters and observed solvent effects for methyl methacrylate, for vinyl acetate there was none. However, it should be noted that evidence for radical-solvent complexes in vinyl acetate systems is fairly strong (see Section 3), so a rejection of a generalized Bootstrap model on the basis of evidence from vinyl acetate polymerization is perhaps unwise. Kratochvil et al.120 investigated the possible influence of preferential solvation in copolymerizations and concluded that, for systems with weak non-specific interactions, such as STY-MMA, the effect of preferential solvation on kinetics was probably comparable to the experimental error in determining the rate of polymerization (±5%). Later, Maxwell et al.121 also concluded that the origin of the Bootstrap effect was not likely to be bulk monomer-polymer thermodynamics since, for a variety of monomers, Flory-Huggins theory predicts that the monomer ratios in the monomer-polymer phase would be equal to that in the bulk phase.122 Nevertheless, there are many copolymerization systems for which there is strong evidence for preferential solvation, in particular, polymer solutions exhibiting cosolvency or where one of the solvents is a non-solvent for the polymer. Preferential adsorption and
794
Michelle L. Coote and Thomas P. Davis
desorption are manifest where the polymer adjusts its environment towards maximum solvation. With this knowledge it may be expected that Bootstrap effects based on preferential solvation will be strongest where one of the monomers (or solvents) is a poor or non-solvent for the polymer (such as copolymerization of acrylonitrile or N-vinyl carbazole). Indeed, early experimental evidence for a partitioning mechanism in copolymerization was provided by Ledwith et al.123 for the copolymerization of N-vinyl carbazole with MMA in the presence of a range of solvents. Direct evidence for preferential solvation was obtained by Semchikov et al.,124 who suggested that it could be detected by calculating, from measurements of the solution thermodynamics, the total and excess thermodynamic functions of mixing. Six monomer pairs were selected -Vac-NVP, AN-STY, STY-MA, Vac-STY, STY-BMA and MMA-STY. The first four of these monomer pairs were known to deviate from the terminal model composition equation, while the latter two were not. They found that these first four copolymerizations had positive ∆GE values over the temperature range measured, and thus also formed non-ideal polymer solutions (that is, they deviated from Raoult’s law). Furthermore, the extent of deviation from the terminal model composition equation could be correlated with the size of the ∆GE value, as calculated from the area between the two most different composition curves obtained for the same monomer pair under differing reaction conditions (for example, initiator concentration; or type and concentration of transfer agent). For STY-MMA they obtained negative ∆GE values over the temperature range considered, but for STY-BMA negative values were obtained only at 318 and 343K, and not 298K. They argued that the negative ∆GE values for STY-MMA confirmed the absence of preferential solvation in this system, and hence its adherence to the terminal model composition equation. For STY-BMA they suggested that non-classical behavior might be expected at low temperatures. This they confirmed by polymerizing STY-BMA at 303K and demonstrating a change in reactivity ratios of STY-BMA with the addition of a transfer agent. Based upon the above studies, it may be concluded that there is strong evidence to suggest that Bootstrap effects arising from preferential solvation of the polymer chain operate in many copolymerization systems, although the effect is by no means general and is not likely to be significant in systems such as STY-MMA. However, this does not necessarily discount a Bootstrap effect in such systems. As noted above, a Bootstrap effect may arise from a number of different phenomena, of which preferential solvation is but one example. Other causes of a Bootstrap effect include preferential solvation of the chain end, rather than the entire polymer chain,121,125 or the formation of non-reactive radical-solvent or monomer-solvent complexes. In fact, the Bootstrap model has been successfully adopted in systems, such as solution copolymerization of STY-MMA, for which bulk preferential solvation of the polymer chain is unlikely. For instance, both Davis125 and Klumperman and O’Driscoll43 adopted the terminal Bootstrap model in a reanalysis of the microstructure data of San Roman et al.126 for the effects of benzene, chlorobenzene and benzonitrile on the copolymerization of MMA-STY. Versions of the Bootstrap model have also been fitted to systems in which monomer-monomer complexes are known to be present, demonstrating that the Bootstrap model may provide an alternative to the MCP and MCD models in these systems. For instance, Klumperman and co-workers have successfully fitted versions of the penultimate Bootstrap model to the systems styrene with maleic anhydride in butanone and toluene,43 and styrene
13.2 Solvent effects on free radical polymerization
795
with acrylonitrile in various solvents.44 This latter work confirmed the earlier observations of Hill et al.127 for the behavior of styrene with acrylonitrile in bulk, acetonitrile and toluene. They had concluded that, based on sequence distribution data, penultimate unit effects were operating but, in addition, a Bootstrap effect was evident in the coexistent curves obtained when triad distribution was plotted against copolymer composition for each system. In the copolymerization of styrene with acrylonitrile Klumperman et al.44 a variable Bootstrap effect was required to model the data. Given the strong polarity effects expected in this system (see Section 13.2.2), part of this variation may in fact be caused by the variation of the solvent polarity and its affect on the reactivity ratios. In any case, as this work indicates, it may be necessary to simultaneously consider a number of different influences (such as, for instance, penultimate unit effects, Bootstrap effects, and polarity effects) in order to model some copolymerization systems. 13.2.4 CONCLUDING REMARKS Solvents affect free-radical polymerization reactions in a number of different ways. Solvent can influence any of the elementary steps in the chain reaction process either chemically or physically. Some of these solvent effects are substantial, for instance, the influence of solvents on the gel effect and on the polymerization of acidic or basic monomers. In the specific case of copolymerization then solvents can influence transfer and propagation reactions via a number of different mechanisms. For some systems, such as styrene-acrylonitrile or styrene-maleic anhydride, the selection of an appropriate copolymerization model is still a matter of contention and it is likely that complicated copolymerization models, incorporating a number of different phenomena, are required to explain all experimental data. In any case, it does not appear that a single solvent effects model is capable of explaining the effect of solvents in all copolymerization systems, and model discrimination should thus be performed on a case-by-case basis. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Bednarek, D., Moad, G., Rizzardo, E. and Solomon, D. H., Macromolecules, 1988, 21, 1522. Kiefer, H. and Traylor, T., J. Am. Chem. Soc., 1967, 89, 6667. Terazima, M., Tenma, S., Watanabe, H. and Tominaga, T., J. Chem. Soc.-Farad. Trans., 1996, 92, 3057. Terazima, M. and Hamaguchi, H., J. Phys. Chem., 1995, 99, 7891. Bamford, C. H. and Brumby, S., Makromol. Chem., 1967, 105, 122. Burnett, G. M., Cameron, G. G. and Joiner, S. N., J. Chem. Soc., Faraday Trans., 1972, 69, 322. Burnett, G. M., Dailey, W. S. and Pearson, J. M., Trans. Faraday Soc., 1972, 61, 1216. Anderson, D. B., Burnett, G. M. and Gowan, A. C., J. Polym. Sci. Part A: Polym. Chem., 1963, 1, 1456. Zammit, M. D., Davis, T. P., Willett, G. D. and O’Driscoll, K. F., J. Polym. Sci: Part A: Polym. Chem., 1997, 35, 2311. Shtamm, E. V., Skurlatov, Y. I., Karaputadse, I. M., Kirsh, Y. E. and Purmal, A. P., Vysokomol. Soedin.B, 1980, 22, 420. Pascal, P., Napper, D. H., Gilbert, R. G., Piton, M. C. and Winnik, M. A., Macromolecules, 1990, 23, 5161. Morrison, D. A. and Davis, T. P., Macromol. Chem. Phys., 2000 - submitted. Davis, T. P., O’Driscoll, K. F., Piton, M. C. and Winnik, M. A., Macromolecules, 1989, 22, 2785. Coote, M. L. and Davis, T. P., Eur. Polym. J., 2000-in press. Olaj, O. F. and Schnoll-Bitai, I., Monatsh. Chem., 1999, 130, 731. Beuermann, S., Buback, M., Schmaltz, C. and Kuchta, F.-D., Macromol. Chem. Phys., 1998, 199, 1209. Yamada, K., Nakano, T. and Okamoto, Y., Macromolecules, 1998, 31, 7598. Kamachi, M., Liaw, D. J. and Nozakura, S., Polym. J., 1979, 12, 921. Coote, M. L. and Davis, T. P., Progr. Polym. Sci., 2000, 24, 1217. Renaud, P. and Gerster, M., Angew. Chem. Int. Ed., 1998, 37, 2562. Odian, G., Principles of Polymerization; 2nd edn. Wiley, New York, 1981. Harrisson, S., Kapfenstein, H. M. and Davis, T. P., Macromolecules, 2000- submitted.
796
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
Michelle L. Coote and Thomas P. Davis Heuts, J. P. A., Forster, D. J. and Davis, T. P., Macromolecules, 1999, 32, 3907. Davis, T. P., Haddleton, D. M. and Richards, S. N., J. Macromol. Sci., Rev. Macromol. Chem. Phys., 1994, C34, 234. Benson, S. W. and North, A. M., J. Am. Chem. Soc., 1962, 84, 935. Mahabadi, H. K. and O’Driscoll, K. F., J. Polym. Sci., Polym. Chem. Ed., 1977, 15, 283. Norrish, R. G. W. and Smith, R. R., Nature (London), 1942, 150, 336. O’Neil, G. A., Wisnudel, M. B. and Torkelson, J. M., Macromolecules, 1998, 31, 4537. Cameron, G. G. and Cameron, J., Polymer, 1973, 14, 107. Mayo, F. R. and Lewis, F. M., J. Am. Chem. Soc., 1944, 66, 1594. Fukuda, T., Kubo, K. and Ma, Y.-D., Prog. Polym. Sci., 1992, 17, 875. Fukuda, T., Ma, Y.-D. and Inagaki, H., Macromolecules, 1985, 18, 17. see for example(a) Wong, M. W., Pross, A., Radom, L., J. Am. Chem. Soc., 1994, 116, 6284; (b) Wong, M., W., Pross, A.; Radom, L., J. Am. Chem. Soc., 1994, 116, 11938. For an account of solvent effects in chemical reactions see for example: Pross, A. Theoretical and Physical Principles of Organic Reactivity; John Wiley & Sons, Inc.: New York, 1995. Price, C. C., J. Polym. Sci., 1946, 1, 83. Mayo, F. R. and Walling, C., Chem. Rev., 1950, 46, 191. Kamachi, M., Adv. Polym. Sci., 1981, 38, 56. Lewis, F., Walling, C., Cummings, W., Briggs, E. R. and Mayo, F. R., J. Am. Chem. Soc., 1948, 70, 1519. Nozaki, K., J. Polym. Sci., 1946, 1, 455. Sandner, B. and Loth, E., Faserforsch Textiltechnology, 1976, 27, 571. Sandner, B. and Loth, E., Faserforsch. Textiltechnology, 1976, 27, 633. See for example: (a) Zytowski, T., Fischer, H., J. Am. Chem. Soc., 1996, 118, 437; (b) Heberger, K., Fischer, H., Int. J. Chem. Kinetics, 1993, 25, 249. Klumperman, B. and O’Driscoll, K. F., Polymer, 1993, 34, 1032. Klumperman, B. and Kraeger, I. R., Macromolecules, 1994, 27, 1529. Russell, G. A., J. Am. Chem. Soc., 1958, 80, 4897. Russell, G. A., Tetrahedron, 1960, 8, 101. Russell, G. A., Ito, K. and Hendry, D. G., J. Am. Chem. Soc., 1961, 83, 2843. Sadhir, R. K., Smith, J. D. B. and Castle, P. M., J. Polym. Sci: Part A: Polym. Chem., 1985, 23, 411. Strong, R. L., Rand, S. J. and Britl, A. J., J. Am. Chem. Soc., 1960, 82, 5053. Strong, R. L. and Perano, J., J. Am. Chem. Soc., 1961, 83, 2843. Strong, R. L., J. Phys. Chem., 1962, 66, 2423. Martin, M. M. and Gleicher, G. J., J. Am. Chem. Soc., 1964, 86, 238. Russell, G. A., J. Org. Chem., 1959, 24, 300. Burnett, G. M., Cameron, G. G. and Cameron, J., Trans. Faraday Soc., 1973, 69, 864. Buchachenko, A. L., Sukhanova, O. P., Kalashnikova, L. A. and Neiman, M. B., Kinetika i Kataliz, 1965, 6, 601. Kalashnikova, L. A., Neiman, M. B. and Buchachenko, A. L., Zh. Fiz. Khim., 1968, 42, 598. Kalashnikova, L. A., Buchachenko, A. L., Neiman, M. B. and Romantsev, E. G., Zh. Fiz. Khim., 1969, 43, 31. Hatada, K., Kitayama, T. and Yuki, H., Makromol. Chem., Rapid Commun., 1980, 1, 51. Busfield, W. K., Jenkins, I. D. and Monteiro, M. J., Aust. J. Chem., 1997, 50, 1. Busfield, W. K., Jenkins, I. D. and Monteiro, M. J., J. Polym. Sci.,;Part A, Polym. Sci., 1997, 35, 263. Busfield, W. K., Jenkins, I. D. and Monteiro, M. J., Polymer, 1996, 38, 165. Ratzsch, M. and Vogl, O., Progr. Polym. Sci., 1991, 16, 279. Mayo, F., J. Am. Chem. Soc., 1958, 80, 4987. Burnett, G. M., Dailey, W. S. and Pearson, J. M., Trans. Faraday Soc., 1965, 61, 1216. Burnett, G. M., Dailey, W. S. and Pearson, J. M., Eur. Polym. J., 1969, 5, 231. Hall Jr., H. K. and Daly, R. C., Macromolecules, 1975, 8, 23. Henrici-Olive, G. and Olive, S., Makromol. Chem., 1963, 68, 219. Henrici-Olive, G. and Olive, S., Z. Phys. Chem., 1965, 47, 286. Henrici-Olive, G. and Olive, S., Z. Phys. Chem., 1966, 48, 35. Henrici-Olive, G. and Olive, S., Z. Phys. Chem., 1966, 48, 51. Henrici-Olive, G. and Olive, S., Makromol. Chem., 1966, 96, 221. Burnett, G. M., Cameron, G. G. and Zafar, M. M., Eur. Polym. J., 1970, 6, 823. Burnett, G. M., Cameron, G. G. and Joiner, S. N., J. Chem. Soc., Faraday Trans., 1973, 69, 322. Kamachi, M., Satoh, J. and Nozakura, S.-I., J. Polym. Sci. Polym. Chem. Ed., 1978, 16, 1789.
13.2 Solvent effects on free radical polymerization
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
797
Pascal, P., Napper, D. H., Gilbert, R. G., Piton, M. C. and Winnik, M. A., Macromolecules, 1993, 26, 4572. Heublein, B. and Heublein, G., Acta Polym., 1988, 39, 324. O’Driscoll, K. F. and Monteiro, M. J., 1996. Czerwinski, W. K., Macromolecules, 1995, 28, 5411. Bartlett, P. D. and Nozaki, K., J. Am. Chem. Soc., 1946, 68, 1495. Seiner, J. A. and Litt, M., Macromolecules, 1971, 4, 308. Cais, R. E., Farmer, R. G., Hill, D. J. T. and O’Donnell, J. H., Macromolecules, 1979, 12, 835. Bamford, C. H. and Barb, W. G., Discuss. Faraday Soc., 1953, 208. Barb, W. G., J. Polym. Sci., 1953, 11, 117. Tsuchida, E. and Tomono, H., Makromol. Chem., 1971, 141, 265. Tsuchida, E., Tomono, T. and Sano, H., Makromol. Chem., 1972, 151, 245. Dodgson, K. and Ebdon, J. R., Makromol. Chem., 1979, 180, 1251. Dodgson, K. and Ebdon, J. R., Eur. Polym. J., 1977, 13, 791. Walling, C., Briggs, E. R., Wolfstern, K. B. and Mayo, F. R., J. Am. Chem. Soc., 1948, 70, 1537. Farmer, R. G., Hill, D. J. T. and O’Donnell, J. H., J. Macromol. Sci., Chem., 1980, A14, 51. Hill, D. J. T., O’Donnell, J. J. and O’Sullivan, P. W., Prog. Polym. Sci., 1982, 8, 215. Sanayei, R. A., O’Driscoll, K. F. and Klumperman, B., Macromolecules, 1994, 27, 5577. Litt, M., Macromolecules, 1971, 4, 312. Barb, W. G., Proc. Roy. Soc., Ser. A, 1952, 212, 66. Barb, W. G., J. Polym. Sci., 1952, 10, 49. Booth, D., Dainton, F. S. and Ivin, K. J., Trans. Faraday Soc., 1959, 55, 1293. Lewis, F. M., Walling, C., Cummings, W., Briggs, E. R. and Wenisch, W. J., J. Am. Chem. Soc., 1948, 70, 1527. Hill, D. J. T., O’Donnell, J. H. and O’Sullivan, P. W., Macromolecules, 1982, 15, 960. Cowie, J. M. G., in Alternating Copolymers; Vol. , ed. Cowie, J. M. G., Plenum, New York, 1985, . Seymour, R. B. and Garner, D. P., Polymer, 1976, 17, 21. Seymour, R. B. and Garner, D. P., Polym. News, 1978, 4, 209. Klumperman, B. “Free Radical Copolymerization of Styrene and Maleic Anhydride,” PhD Thesis, Technische Universiteit Eindhoven, 1994. Arnaud, R., Caze, C. and Fossey, J., J. Macromol. Sci. - Chem., 1980, A14, 1269. Iwatsuki, S. and Yamashita, Y., Makromol. Chem., 1967, 104, 263. Olson, K. G. and Butler, G. B., Macromolecules, 1983, 16, 710. Rätzsch, M. and Seinert, V., Makromol. Chem., 1984, 185, 2411. Butler, G. B., Olson, K. G. and Tu, C.-L., Macromolecules, 1984, 17, 1884. Smirnov, A. I., Deryabina, G. L., Kalabina, A. L., Petrova, T. L., Stoyachenko, I. L., Golubev, V. B. and Zubov, V. P., Polym. Sci. USSR (Engl. Transl.), 1978, 20, 2014. Golubev, V. B., Zubov, V. P., Georgiev, G. S., Stoyachenko, I. L. and Kabanov, V. A., J. Polym. Sci., Polym. Chem. Ed., 1973, 11, 2463. Barton, J., Capek, I. and Tino, J., Makromol. Chem., 1980, 181, 255. Karad, P. and Schneider, C., J. Polym. Sci. Part A: Polym. Chem., 1983, 16, 1295. Hill, D. J. T., O’Donnell, J. H. and O’Sullivan, P. W., Macromolecules, 1983, 16, 1295. Giese, B. and Meister, J., Chem. Ber., 1977, 110, 2558. Lebduska, J., Snuparek, J., Kaspar, K. and Cermak, V., J. Polym. Sci: Part A: Polym. Chem., 1986, 24, 777. Ito, K., Uchida, K., Kitano, T., Yamada, E. and Matsumoto, T., Polym. J., 1985, 17, 761. Chapiro, A. and Perec-Spitzer, L., Eur. Polym. J., 1975, 25, 713. Harwood, H. J., Makromol. Chem. Makromol. Symp., 1987, 10/11, 331. Semchikov, Y. D., Macromol. Symp., 1996, 111, 317. Coote, M. L., Johnston, L. P. M. and Davis, T. P., Macromolecules, 30, 8191 (1997). Fukuda, T., Ma, Y.-D. and Inagaki, H., Makromol. Chem., Rapid Commun., 1987, 8, 495. Kratochvil, P., Strakova, D., Stejskal, J. and Tuzar, Z., Macromolecules, 1983, 16, 1136. Maxwell, I. A., Aerdts, A. M. and German, A. L., Macromolecules, 1993, 26, 1956. Maxwell, I. A., Kurja, J., Doremaele, G. H. J. v. and German, A. L., Makromol Chem., 1992, 193, 2065. Ledwith, A., Galli, G., Chiellini, E. and Solaro, R., Polym. Bull., 1979, 1, 491. Egorochkin, G. A., Semchikov, Y., D., Smirnova, L. A., Karayakin, N. V. and Kut’in, A. M., Eur. Polym. J., 1992, 28, 681. Davis, T. P., Polym. Commun., 1990, 31, 442. San Roman, J., Madruga, E. L. and Puerto, M. A., Angew Makromol. Chem., 1980, 86, 1. Hill, D. J. T., Lang, A. P., Munro, P. D. and O’Donnell, J. H., Eur. Polym. J., 1992, 28, 391. Coote, M. L., Davis, T. P., Klumperman, B. and Monteiro, M. J., J.M.S.-Rev. Macromol. Chem. Phys., 1998, C38, 567.
798
Maw-Ling Wang
13.3 EFFECTS OF ORGANIC SOLVENTS ON PHASE-TRANSFER CATALYSIS Maw-Ling Wang Department of Chemical Engineering National Chung Cheng University, Taiwan, ROC
The reaction of two immiscible reactants is slow due to their low solubilities and limited contact surface area. The conventional way to improve the reaction rate or to elevate the conversion of reactants is to increase the agitation speed, temperature, or use the protic or aprotic solvent to dissolve the reactants. The increase in agitation speed can increase the contact surface area between two phases only to a certain value. Thus, the reaction rate or the conversion is limited by the increase in the agitation speed. Usually, the rate of reaction is increased by raising the temperature. However, byproducts are accompanied by elevating the solution temperature. The separation of product from byproducts or catalyst makes the cost to increase. Although protic solvent (CH3OH, or CH3COOH) can dissolve reactants, solvation and hydrogen bonding make the activity of the nucleophilic anion decrease significantly. Thus, the reaction rate using protic solvent is retarded. For the other case, the reaction rate is largely increased using aprotic solvent. The application of aprotic solvent is also limited because of cost and recovery difficulty. For this, the problem of two-phase reaction is not overcome until the development of phase-transfer catalysis (PTC). Phase-transfer catalytic reactions provide an effective method in organic synthesis from two immiscible reactants in recent development.93,103,111,113 In 1951, Jarrouse47 found that the reaction of aqueous-soluble sodium cyanide (NaCN) and organic-soluble 1-chlorooctane (1-C8H17Cl) is dramatically enhanced by adding a small amount of quaternary ammonium salt (R4N+X-, or Q+X-, Q+: R4N+). The reaction is almost complete and a 95% conversion is obtained within two hours when a catalytic amount of tetra-n-butylammonium chloride ((C4H9)4N+Cl-, or Q+Cl-, Q+: (C4H9)4N+) is added. The mechanism of the reaction of sodium cyanide and 1-chlorooctane in organic solvent/water two-phase medium is expressed as
[13.3.1]
As shown in Equation [13.3.1], sodium cyanide (NaCN) and 1-chlorooctane (1-C8H17Cl) are soluble in aqueous phase and organic phase, respectively. In the aqueous phase, NaCN first reacts with tetra-n-butylammonium chloride ((C4H9)4N+Cl-, Q+Cl-) to produce organic-soluble tetra-n-butylammonium cyanide ((C4H9)4N+CN-, Q+CN-). Then, this tetra-n-butylammonium cyanide (QCN) further reacts with 1-chlorooctane (1-C8H17Cl) to produce 1-cyanooctane (C8H17CN) in the organic phase. Tetra-n-butylammonium chloride ((C4H9)4N+Cl-), which is also produced from the organic-phase reaction, transfers to the aqueous phase, prepared for further regeneration. It is obvious that PTC reaction107 involves
13.3 Effects of organic solvents on phase-transfer catalysis
799
a two-phase reaction (aqueous-phase and organic-phase reaction), transfer of QCN from aqueous phase to organic phase and transfer of QCl from organic phase to aqueous phase, and equilibrium partition of QCN and QCl between organic and aqueous phases, respectively. The overall reaction rate highly depends on the intrinsic rate constants in aqueous phase and organic phase, the mass transfer rate of QCN and QBr, and the equilibrium partition coefficient of QCN and QBr, which are all affected by the interaction of components and their environments. The organic solvent provides the environment for the interaction of reactants. Therefore, the organic solvent plays an important role in influencing the reaction rate and the conversion of reactant. Since then, Makosza used an interfacial mechanism65-67 to describe the behavior in the two-phase reaction. Later, Starks107 used the extraction mechanism to explain the behavior in the two-phase reaction and selected phase-transfer catalysis (PTC) to describe this special chemical process.14,161 The most important advantage of using PTC technique is in synthesizing specialty chemicals with almost no byproducts and moderate reaction conditions. Today, PTC is widely applied to various reactions via substitution, displacement, condensation, oxidation and reduction, polymer modification and polymerization to synthesize specialty chemicals. Based on the reaction mechanism, phase-transfer catalysis can be classified as: (1) normal phase-transfer catalysis (NPTC), (2) reverse phase-transfer catalysis (RPTC), and (3) inverse phase-transfer catalysis (IPTC). Equation [13.3.1] illustrates the typical reaction for NPTC. The phase-transfer catalyst (Q+) brings the nucleophilic reagent (CN-) from aqueous phase to organic phase. Quaternary ammonium salts, quaternary phosphonium salts, crown ethers, polyethylene glycols (PEGs) and tertiary amines are the common normal phase-transfer catalyst (NPTC).17,29,94,108,109,110,128,130,152 In general, the cation transfers from aqueous phase to organic phase in the RPTC. The principle of reverse phase-transfer catalysis (RPTC)24,42-44,50 is that an ion pair is formed from catalyst and cation in the aqueous phase. This ion-paired compound then transfers to the organic phase reacting with an organic-phase reactant. Alkyl-aryl sulfonate (RSO3Na), such as sodium 4-dodecylbenzene sulfonate (NaDBS) and tetraarylboronate such as sodium tetra(diperfluoromethyl)phenyl-boronate (TFPB) are the common reverse phase-transfer catalysts. However, few results were reported using reverse phase-transfer catalysis in synthesizing specialty chemicals.24,42-44,50 A typical reaction mechanism in a liquid-liquid two-phase solution is given by Equation [13.3.2] In the NPTC and RPTC, the function of the catalyst is that it first reacts with aqueous-phase reactant to produce an organic-soluble ion-pair compound. Mathias and Vaidya69 found that an aqueous-soluble ion pair was produced in the organic phase from the reaction of alanine and benzoyl chloride catalyzed by 4-dimethylaminopyridine (DMAP). This discovery initiated the research of the field in inverse phase-transfer catalysis (IPTC), in which the catalyst first reacts with organic-phase reactant in the organic phase to produce an aqueous-soluble ion-paired intermediate. Then, this aqueous-soluble ion-paired intermediate transfers to the aqueous phase, prepared for reacting with aqueous-phase reactant to produce the desired product. Catalyst is released in the aqueous phase and transferred to the organic phase for further regeneration. A typical IPTC mechanism of the reaction of benzoyl chloride and sodium acetate to synthesize ester compound in the liquid-liquid two-phase reaction is expressed by Equation [13.3.3]
800
Maw-Ling Wang
[13.3.2]
Inverse phase-transfer catalysis (IPTC) can be applied to synthesize symmetric and antisymmetric acid anhydride in organic synthesis.26,54,69,102,148-150,153 Pyridine 1-oxide (PNO), 4-dimethylaminopyridine (DMAP), 4-pyrrolidinopyridine (PPY) and 1-methyl-2(1H)-pyridothione are usually used as the inverse phase-transfer catalysts (IPTC).159
[13.3.3]
The characteristics of two-phase phase-transfer catalytic (PTC) systems are: the presence of at least two phases and at least one interfacial region separating the phases.24,26,42,43,50,69,107,109,110,152 The reactions involve: (1) transfer of an ion or compound from its normal phase into the reaction phase or interfacial region, (2) reaction of the transferred ion or compound with the non transferred reactant located in the reaction phase or interfacial region, and (3) transfer of the product from the reaction phase or interfacial region into its normal phase. For example, a successful NPTC process involves (1) the maximization of the rate of transfer of reactant anions from the aqueous or solid phase to the organic phase, (2) the maximization of the rate of transfer of product anions from the organic phase to the aqueous phase or solid phase, and (3) the related equilibrium partitioning of the reactant and product anions between organic and aqueous or solid phases. The anion must not only transfer to the organic phase, but once there the anion must be in a highly reactive form. Some organic-phase reactions are so fast that the transferred anion requires little or no activation beyond just being delivered to the organic phase. Other reactions require substantial anion activation before useful and practical reaction rates can be achieved. It is obvious that the polarity of the organic solvent affects the activation of the anion as well as the difference
13.3 Effects of organic solvents on phase-transfer catalysis
801
in the cation-anion interionic distance for the two ion pairs. In principle, anions do not have a great affinity for nonpolar solvent and prefer to reside in an aqueous phase. Ease or difficulty of transfer of most anions into organic-phase solution is also highly affected by the organic solvent, i.e., interaction of the organic solvent and the reactant. In general, a polar solvent may be necessary to obtain an appropriate rate of the anion transfer to the organic phase for a NPTC process. Solvent may be necessary to increase the rate of the organic-phase reaction. The most common solvent, dichloromethane (CH2Cl2), has been extensively used as a polar solvent in the PTC work because it readily dissolves most quaternary salts and other phase-transfer agents, and because it is polar to speed both the transfer step and the organic-phase reaction step. Although the hydrocarbons suffer from lack of polarity, they have also been extensively used as solvents for PTC systems. The main reasons are that they are reasonably safe, inexpensive and easy to recover in a high purity. One strategy for selecting organic solvent is that a high boiling point solvent is selected for a reaction in which the product has a low-boiling point. In other cases, a solvent might be chosen to minimize solubility of phase-transfer agent in the organic phase to force formation of third phase (catalyst) from which the phase-transfer catalyst may be more easily separated or extracted. The phase-transfer catalytic reactions (NPTC, RPTC or IPTC) are usually carried out in a liquid-liquid two-phase medium. They have been extensively applied to liquid-gas, liquid-solid two-phase media.18,21,63,128-130 However, purification of product from catalyst in the liquid phase of a final solution is difficult to produce a product of high purity. In 1975, Regen and coworkers88-92 proposed triphase catalysis (TC) in which the catalyst is immobilized on a porous solid support (usually polymer). The solid catalyst is easily separated from the final products after reaction by mechanical separation processes, such as centrifugation or filtration. The organophilicity and the hydrophilicity of the solid polymer support greatly influence the content and the imbibed composition of the organic phase and the aqueous phase within the solid porous polymer support. Hence, the reaction rates are determined by the concentrations of reactants in both the organic phase and aqueous phase, they are controlled by the organic solvents. Therefore, it is important to understand the characteristics of the organic phase in the triphase catalyst as well as the characteristics of the organic solvent in the liquid-liquid two-phase PTC reaction. 13.3.1 TWO-PHASE PHASE-TRANSFER CATALYTIC REACTIONS 13.3.1.1 Theoretical analysis of the polarity of the organic solvents and the reactions The transfer of anions from an aqueous phase to an organic phase may be achieved by choosing a phase-transfer cation that is not strongly solvated by water, that has organic-like characteristics, and is compatible with the organic phase for NPTC. The factors that affect the mass transfer and the distribution of the phase-transfer catalyst cation-anion pair between the organic and aqueous phases include: (1) the charge-to-volume ratio of the anion, the polarizability, and the organic structure of the cation associated with the anion, (2) the hydrophilic-organophilic balance of the associated cation; (3) the polarity of the organic phase; (4) the hydration of the anion; (5) the presence of aqueous salts and/or aqueous hydroxide ions.
802
Maw-Ling Wang
Both cation and anion of phase-transfer catalyst can affect the distribution of the PTC between two phases, and hence the reaction rate. The partitioning equilibrium of the anion between organic and aqueous phases can be qualitatively estimated from the free energies of the anion transfer from water to organic phase. A large positive free energy of transfer from the aqueous phase to the organic phase clearly indicates that the anion prefers to reside in the aqueous phase. For example, the free energies of transfer of Cl-, Br-, and I- from water to acetonitrile are +11.6, +8.1 and +4.8 Kcal/mol, respectively.110 Thus, the transfer from aqueous to organic phase becomes less unfavorable as one proceeds from chloride to bromide to iodide. These trends may be understood in terms of the change in charge-to-volume ratios of the halide ions. Because chloride has the largest charge-to-volume ratio, it is the least polarizable and the most strongly hydrated. In contrast, iodide has a relatively diffuse charge and is less strongly hydrated. A successful phase-transfer catalytic reaction occurs when the process is able to transfer the anions from the aqueous phase to the organic phase or vice versa for the reaction to proceed, and the transferred anions are active and prepared for reaction. An active catalyst needs to be sufficiently distributed in the organic phase for the reaction to occur. The distribution of catalysts and the associated anion in the organic phase strongly depends on the structure of the quaternary cation and the hydration of anion being transferred into the aqueous phase. Therefore, the following results are used for the reference in selective NPTC catalyst. (1) Tetramethylammonium cation with a simple anion (CH3)4N+Y- (Y=Cl-, Br-, CN-, etc.) is not easily distributed in most organic solutions. Therefore, (CH3)4N+Y- are usually not good PTCs. The only ways to increase the distribution of (CH3)4N+Y- is to couple the cation with a large organic anion13,49,78,95,155,156 or to use an organic solvent of high purity.17,107,109,110,152 (2) Tetraethylammonium ((C2H5)4N+Y-) and tetrapropylammonium ((C3H7)4N+Y-) salts are also poor catalysts for transferring small anions into most organic solutions.51,105 (3) Tetrabutylammonium salts show high efficiencies as phase-transfer catalysts. They are readily available in high purity on a commercial scale. (4) Quaternary ammonium cations, R4N+, R=C5H11 to C10H21 easily extract anions into organic phase and exhibit higher catalytic activities. (5) Higher tetraalkylammonium salts, R4N+X, R: (C12H25)4N+ and higher groups, can easily extract anion into an organic phase. However, the interchange of anions between organic and aqueous phases is slow and the reaction rate decreases compared with quaternary salts where R = C5H11 to C10H21. Table 13.3.1 shows the effect of catalyst structure on the rate of PTC reaction of thiophenoxide with 1-bromooctane. In addition to the preference of anion to reside in the aqueous or organic phase, a distribution ratio (or partition coefficient), α, of phase-transfer catalyst (QX) cation between aqueous and organic phase is defined as α = [QX]org/[QX]aq
[13.3.4]
Use of solvents having higher polarity facilitates distribution of quaternary salts into organic solvents. Hence, it also allows use of smaller quaternary salts as catalysts. With di-
13.3 Effects of organic solvents on phase-transfer catalysis
803
Table 13.3.1. Effect of catalyst chloromethane or solvent of a similar polarity, it structure on the rate of the reaction is possible to use tetramethylammonium cation, of thiophenoxide and 1-bromooctane (CH ) N+, or tetrapropylammonium cation, 3 4 in benzene/water solution + k×103 , Lmol-1s-1
Catalyst (CH3)4N+Br+
-
+
-
<0.0016
(C3H7)4N Br
0.0056
(C4H9)4N Br
5.2 +
(C8H17)3MeN Cl
-
31
+
-
37
+
-
0.015
+
-
0.16
(C8H17)3EtP Br
(C6H13)Et3N Br (C8H17)Et3N Br +
-
0.24
+
-
0.28
+
-
0.15
+
-
0.48
C10H21Et3N Br C12H25Et3N Br C16H33Et3N Br C16H33Et3N Br
(C3H7)4N , or cation salts as catalysts. Wu et al.158 measured the concentration distribution of the quaternary salt between dichloromethane (or chlorobenzene) and alkaline solution and determined the thermodynamic characteristics (the true extraction constant, the distribution coefficient, and the dissociation constant). The distribution coefficient, highly dependent on the organic solvent, increased with increasing NaOH concentration. However, the real dissociation constant decreased with increasing NaOH concentration. Konstantinova and Bojinov52 synthesized several unsaturated 9-phenylxanthene dyes under phase transfer catalysis conditions. They determined the most favorable solvent. The extraction constant of QX between two phases is given by equation:
[Q ] [ X ] +
Data obtained from the work of Herriott and Picker36
E QX =
−
aq
[QX ]org
aq
[13.3.5]
In addition to synthesizing specialty chemicals, the PTC technique can be used to analyze many lipid-rich samples.106 Hydrolyzed samples were treated with phenylisothiocyanate and the phenylthiocarbamyl (PTC). Derivatives obtained were separated by reverse phase HPLC. The PTC/reverse HPLC method was used for analysis of chloroform/methanol extracts of spinal cord, lung and bile after chromatography on Lipidex 5000 in methanol/ethylene chloride, 4:1 (v/v).106 The extraction constant of hydroxide is about 104 times smaller than that of chlo19,20 Table 13.3.2 shows the effect of solvents on the distribution of (C4H9)4N+Br- beride. tween aqueous and organic phases. In addition to improved transfer of anion to an organic phase, more polar solvents are commonly recognized in physical organic chemistry for enhancing the rate of organic-phase reactions by providing a more ion-compatible reaction mechanism. The polarity of the organic phase in conjunction with the structure of the anion and the catalyst cation affects the selectivity of the phase-transfer catalyst partitioning into the organic phase. Increasing the polarity and the hydrogen-bonding ability of the organic phase has a strong favorable effect on the extraction of small ions (large charge-to-volume ratios) and on anion with substantial organic structure. The polarity of the organic phase is thus an important factor influencing the reaction rate. The polarity of the organic phase depends on the polarity of solvent and organic-phase reactant. However, there are many PTC reactions without organic phase (near organic reactant is used as the organic phase). Thus, a substantial change in the polarity of the organic
804
Maw-Ling Wang
Table 13.3.2. Effect of organic solvents on phase may occur as the reaction proceeds. the distribution of (C4H9)4N+Br- between This may have some effect on the reaction organic and aqueous phases rate of the organic-phase (may raise or
lower the value of the rate constant). The changes in the polarity of the organic phase may increase or decrease, causing almost >10.0 C6H6 all catalyst cation-anion pairs to be parti>10.0 C6H5Cl tioned into the organic phase. This behavior >10.0 o-C6H4Cl2 is evidenced in the cyanide displacement on 1-bromooctane catalyzed by tetra-n>10.0 n-C4H9Cl butylphosphonium bromide ((C4H9)4P+Br-, 3.33 Cl(CH2)4Cl or Q+Br-).107,109,110,111,114 The catalyst is only sparingly soluble in 1-bromooctane, but is 0.34 Cl(CH2)3Cl substantially more soluble in aqueous so0.16 Cl(CH2)2Cl dium cyanide solution, so that initially little 0.028 ClCH2Cl Q+CN- is in the organic phase and the displacement reaction is slow. However, 0.021 CHCl3 tetra-n-butylphosphonium salts ((C4H9)4 0.91 C2H5COC2H5 P+CN-, Q+CN-) are more soluble in the 0.071 CH3COC2H5 product 1-cyanooctane; therefore, as the conversion of 1-bromooctane to 0.014 n-C4H9OH 1-cyanooctane continues, increasing quantities of the catalyst are taken into the or*EQBr= [Q+]aq[Br-]aq/[QBr]org ganic phase, and the reaction rate Data adopted from Brandstrom4 accelerates. This behavior signals the autocatalytic character of reaction. Not only does the solvent affect the reaction rate, but it also determines the reaction mechanism. In Starks’ extraction mechanism of PTC, most reacting compound transfers to the bulk phase. However, reaction may occur at the interface of the two phases. For example: hexachlorocyclotriphosphazene has been reported to react very slowly with 2,2,2-trifluoroethanol in an alkaline solution of NaOH/C6H5Cl two-phase system in the absence of phase-transfer catalyst.136-140 Since sodium 2,2,2-trifluoroethanoxide is not soluble in chlorobenzene, the process probably proceeds at the interface region of the system. Similar is the reaction of benzylation of isobutyraldehyde in the presence of tetra-n-butylammonium iodide in an alkaline solution of NaOH/toluene, which is a two-phase system.37 Makosza interfacial mechanism65-67 was employed to rationalize the experimental results. The main reason is that the ammonium salt of the nucleophilic reagent is not soluble in toluene. Usually, the nucleophilic substitutions under NPTC condition are described by an SN2-type reaction both in solid-liquid and liquid-liquid systems in which they can proceed at the interface through the formation of cyclic adsorption complexes.160 The activity of the nucleophilic reagent in the organic phase is determined by the polarity of the organic solution and the hydration in liquid-liquid system. In the solid-liquid system, the reaction is highly affected by the organic solvent. Solvent
Extraction constant, EQBr*
13.3 Effects of organic solvents on phase-transfer catalysis
805
13.3.1.2 Effect of organic solvent on the reaction in various reaction systems (A) Synthesis of ether compound catalyzed by quaternary ammonium salts (NPTC) One of the most useful synthesis applications of phase transfer catalysis (PTC) is in the preparation of ether according to the following general equation R′X + ROX + OH − PTC → ROR′ + H2 O + X −
[13.3.6]
where R and R' are the primary or secondary alkyl or aryl groups, X is a halide and the caustic base is usually sodium or potassium hydroxide in the aqueous solution. The generally accepted reaction mechanism is
[13.3.7]
It is important to consider that the alkoxide ion (RO-) is a reactive nucleophilic but also a strong base. It was shown that 10-3 M C6H5(CH3)3N+OC4H9 is 1000 times more basic than KOC4H9 (both in C4H9OH). Extracted alkoxide bases can be applied in principle to numerous base-catalyzed reactions, e.g., oxidations, eliminations and isomerization. Better quantitative understanding of the extraction of alkoxide into organic phase is important. Dehmlow et al.20 investigated the extraction of aqueous sodium hydroxide solution with organic solvent containing various quaternary ammonium salts by mixing sodium hydroxide with organic solvent containing R4NX. After phase separation, titration of the organic phase showed only traces of base presence if concentrated NaOH solution was employed and if Clwas the counter ion. The Cl-OH exchange was found to be of the order 1-2% for all quaternary ammonium chloride with chlorobenzene as solvent; i.e., 98% of the salts remained in the R4NCl form. However, upon addition of trace amounts of various alcohols, a dramatic change in the behavior of the system was observed and significant amounts of base could be detected in the organic phase. Table 13.3.3 shows the experimental results where 50% aqueous caustic solutions were extracted by equal volumes of 0.1 M (C8H17)4NBr in chlorobenzene containing 0.1 M of various alcohols.
806
Maw-Ling Wang
Table 13.3.3. Extraction of base by chlorobenzene solution of tetra-n-octyl-ammonium bromide and alcohols (0.1 M) from an equal volume of 50% NaOH (percent of the maximum possible basicity) % Primary alcohols
% 2-tert-Butylcyclohexanol
2.0
3- Methylcyclohexanol
2.0
Ethanol
4.5
1-Propanol
5.0
2-Methyl-1-propanol
4.4
tert-Butanol
0.3
1-Pentanol
4.3
2-Methyl-2-butanol
0.2
1- Hexanol
4.3
1-Heptanol
4.8
1,5-Pentanediol
< 0.02
1-Octanol
2.0
2,5-Hexanediol
5.2
1-Dodecanol
0.8
2,2-dimethyl-1,3-propanediol
18.4
2-Methyl-2,4-propanediol
28.0
Secondary alcohols
tert-Alcohols
Diols
2- Propanol
1.9
2,3-Dimethyl-2,3-butanediol
25.8
2-Pentanol
1.2
2,5-Dimethyl-2,5-hexanediol
32.0
2-Hexanol
1.1
2-Octanol
0.7
Ethylene glycol monoethylether
8.9
Cyclohexanol
0.5
Diethylene glycol monobutylether
8.7
4-tert-Butylcyclohexanol
1.5
Glycerol isopropylideneacetal
13.0
Diol monoethers
Data obtained from Dehmlow et al.20
It is apparent that the order of decreasing alkoxide extraction with monohydric alcohols is primary > secondary > tertiary. The better extractivity of diol anions can be attributed to the relatively high acidity of these alcohols in part, but it seems that the main factors are the distance between the two hydroxyl groups and the skeletal structure. In general, the concentration of the extracted base depends on the amount of alcohol added.20 The concentration of aqueous sodium hydroxide is also an important factor in the extraction processes. Herriott and Picker36 carried out the reaction of thiophenoxide ion with 1-bromooctane in a two-phase system. They found that an increase in the ionic strength of the aqueous phase or change to a more polar organic solvent increased the reaction rate. The effect of organic solvent on the reaction rate under NPTC is given in Table 13.3.4. Correlations between the rate constants and the partition coefficients indicate that the major function of the catalyst is simply the solubilization of the nucleophilic in the organic phase. Conventional methods of synthesizing ethers, i.e., Williamson synthesis and alkoxymercuration have been well developed in organic chemistry.76,96 The synthesis of formaldehyde acetal were carried out from the reaction of alcohol and dichloromethane in a 50% sodium hydroxide solution applying Tixoget VP clay as a catalyst. However, completing the reaction for such a low reaction rate takes long time. Dehmlow and Schmidt15 first
13.3 Effects of organic solvents on phase-transfer catalysis
807
Table 13.3.4 Effect of solvent on the rate of used PTC technique to synthesize formreaction of thiophenoxide and bromooctane aldehyde acetals from alcohol and di-
chloromethane in the aqueous phase. Wang and Chang144-146 employed the PTC technique to synthesize formalde(C4H9)4N+IC7H16 0.02 hyde acetals from the reaction of alcohol C6H4Cl2 88 (C4H9)4N+Iand dibromomethane in an alkaline solution of KOH/organic solvent. Alcohol C6H4Cl2 0.45 (C3H7)4N+Br(ROH) first reacted with KOH so as to + 0.04 C6H5CH2(C2H5)3N Br C6H4Cl2 form potassium alkoxide (ROK) in the C6H4Cl2 28 C8H17(C2H5)3N+Braqueous phase. ROK further reacted with quaternary ammonium salt (QBr) C7H16 0.0093 (C6H5)4P+Brin the aqueous phase to produce quater+ C6H4Cl2 47 (C6H5)4P Br nary ammonium alkoxide (QOR) which C6H4Cl2 180 (C6H4)4P+Clis more soluble in the organic solvent. Dibromomethane reacted with QOR in the next step to form the desired product Data obtained from Herriott and Picker36 dialkoxymethane CH2(OR)2 in the organic phase, as shown in Equation [13.3.7]. Dibromomethane, which possesses weak dipole moment, may form a weak dipole-dipole bond with the organic solvent. However, this bond does not significantly affect the reaction rate. QOR solvates in a polar organic solvent. This solvation results in less energy in the nucleophilic agent than that in the transition state compound. The activation energy therefore becomes high due to the solvation of QOR with a highly polar Figure 13.3.1 Effects of the organic solvents on the conver- solvent which is unfavorable in the pression of CH2Br2; 6.88×10-2 mol of 1-butanol, 1.5 molar ratio ent reaction system. The low polarity of 1-butanol/1-octanol, 30 g of KOH, 2.76×10-2 mol of solvent neither solvates QOR, nor sepaCH2Br2, 3.11×10-3 mol of TBAB catalyst, 10 mL of water, 50 rates tetra-n-butylammonium ion (Q+) mL of organic solvent, 1020 rpm, 50oC (Adapted from Ref. from the alkoxide ion (-OR). Thus, the [145], by permission.) reactivity in low polar solvent is low. Organic solvents of appropriate polarity, such as chlorobenzene or dibutyl ether, are the best solvents to obtain a high yields of various alcohols, as shown in Table 13.3.5. As shown in Figure 13.3.1, similar results were obtained in synthesis of unsymmetric acetals under PTC conditions.144-146 The values of kapp, in which the reaction follows a pseudo-first-order kinetic rate law, are 4.59×10-3, 4.58×10-3, 8.17×10-3, and 1.47×10-2 min-1 for reaction of CH2Br2 with butanol and octanol in cyclohexane, n-decane, dibutyl ether and chlorobenzene, respectively. Catalyst
Solvent
k×103 M-1s-1
808
Maw-Ling Wang
Table 13.3.5. Effects of the organic solvents on the conversion of alcohols Organic solvent Reactant
chlorobenzene
dibutyl ether
xylene
benzene
Conversion, X, % 1-Butanol
83.22
58.94
49.65
55.48
1-Heptanol
81.44
50.16
41.94
44.90
1-Octanol
82.17
52.00
46.49
49.35
Cyclohexanol (2 h)
81.37
53.60
48.94
47.89
2-Ethoxyethanol (0.5 h)
92.82
87.86
79.61
73.86
2-(2-Ethoxyethoxy)-ethanol (0.5 h)
98.58
98.87
96.21
92.21
Dielectric constant
5.62
3.08
2.27
2.28
Data obtained from Wang and Chang;145 9.17×10-2 mol of alcohols, 10 mL of H2O, 0.028 mol of CH2Br2, 30 g of KOH, 1 g of TBAB catalyst, 50 mL of organic solvent, 1020 rpm, 50oC
(B) Synthesis of ether compound catalyzed by crown ether (NPTC) The other type of phase transfer catalyst is crown ether, cryptands, polyethylene glycol (PEG) and their derivatives, and other nonionic phase-transfer agents. The phase transfer agent complexes with inorganic cation, along with the anion, can be transferred to the organic phase, preparing for reaction with organic-phase reactant. For example: with 18-crown-6 ether as a phase transfer agent for sodium cyanide:
[13.3.8]
The function of crown ether is that it can chelate with metal ion, such as: lithium, sodium or potassium. Czech et al.12 noted that crown ethers are a better phase-transfer catalyst for solid-liquid reactions, whereas quaternary salts are better for a liquid-liquid system. Table 13.3.6 shows the solubilities of potassium salts in acetonitrile at 25oC in the presence and absence of 18-crown-6 ether. The solubility of potassium salts in CH3CN highly depends on the addition of 18-crown-6 ether.
13.3 Effects of organic solvents on phase-transfer catalysis
809
Table 13.3.6. Solubilities of potassium salts in CH3CN at 25oC in the presence and absence of 18-crown-6 ether Potassium salt KF KCl KBr KI KCN KOAc KSCN
Solubility of potassium salt in 0.15 M crown in CH3CN (A) -3
4.30×10
-2
5.55×10
-1
1.35×10
-1
2.02×10
-1
1.29×10
-1
1.02×10
-1
8.50×10
in CH3OH (B)
Enhancement factor (A/B)
-4
13.52
-4
228.40
-1
64.90
-1
1.92
-3
108.40
-4
204
-1
1.13
3.18×10 2.43×10 2.08×10 1.05×10 1.19×10 5.00×10 7.55×10
Data adopted from the work of Liotta60
The rates of reaction of benzyl bro-
Table 13.3.7. Rates of reaction of benzyl mide and benzyl chloride with potassium chloride with potassium cyanide at 85oC in the presence and absence of 18-crown-6 cyanide were studied as a function of added water in the presence and absence of crown ether as a function of added water
ether in toluene at 85oC,61 as shown in Table 13.3.7. The reaction is highly affected by k×105 s-1 k×105 s-1 Water the addition of 18-crown-6 ether. In addi(crown) (no crown) tion to enhancing the reaction rate, it is im0.0 3.2 0.0 portant to note that in the absence of added 0.36 9.2 0.0 water, the rates followed zero-order kinetics, while in the presence of added water, 0.50 9.4 0.0 the rates followed first order kinetics. 1.00 11.6 0.0 (C) Synthesis of ether compound catalyzed by polyethylene glycols (NPTC) 2.00 14.7 0.0 Similar to quaternary ammonium 10.0 10.2 0.0 salts, polyethylene glycols (PEGs) act as 20.0 5.8 1.3 the phase transfer catalyst. There are two majors effects of PEG on the two-phase re40.0 3.9 1.9 actions. First, part of the PEG, existing in 75.0 4.8 3.2 the organic phase, forms a complex with metal cation. The formation of a complex Data obtained from the work of Liotta;61 0.05 mol of leads to an increase in the solubility of sobenzyl chloride, 0.01 mol of 18-crown-6 ether, 0.15 mol dium alkoxide (RONa) or sodium of KBr, and 0.015 mol of KCN phenoxide (PhONa) for the synthesis of ether in the organic phase. Hence, the reaction rate in the organic phase is promoted. Second, PEG acts as an excellent organic solvent, but it can also dissolve in water. Thus, part of the alkyl halide that is dissolved by PEG is brought into the aqueous phase from the organic phase. The dissolved alkyl halide directly
810
Maw-Ling Wang
reacts with phenoxide (PhO-) or alkoxide (RO-) ion in the aqueous phase, as shown in Figure 13.3.2.143 The reaction rate in the aqueous phase is also enhanced. The mechanism of the reaction rate of alkyl halide (or allyl halide, RX) and phenoxide (PhO-), both existing in the aqueous phase with PEG help, is different from that in presence of quaternary ammonium salt.
[13.3.9]
The reaction catalyzed by PEG can be carried out either in a homogeneous phase or in a two-phase solution. The alkyl halide usually serves as the reactant as well as the solvent. The reaction proceeds because organic-phase reactant dissolves in an organic solvent in the presence of PEGs. Dichloromethane, chlorobenzene, ethyl ether, cyclohexane and n-decane are frequently used solvents. The reaction mechanism of two-phase catalytic reaction by PEG includes formation of a com141,142 This Figure 13.3.2 Dependence of the concentration of allyl chloride in plex of PEG with cation. the aqueous phase on the amount of PEG-1500 added, Vorg=Vaq=50 is different than the reaction catamL, 30oC (Adapted from Ref. [143], by permission.) lyzed by quaternary salts. Table 13.3.8 shows the initial reaction rate using PEG, ((-r)i,PEG) and the initial reaction rate without using PEG, ((-r)i,B) in various organic solvents. Both (-r)i,B and (-r)i,PEG decrease when the polarity of the organic solvent increases. The maximum reaction rate is obtained with n-decane, which has the lowest polarity, as the protic solvent. Same results were obtained from the work of Landini et al.55 on the reaction of n-octylmethylene sulfonate and bromide ion in a homogeneous phase with C16H33P+(C4H9)3Y- as PTC. Wang and Chang142 made a reasonable explanation for this peculiar phenomena, i.e., the transition state possesses a higher degree of dispersity of electric charge than does the ground state. Increasing the polarity of the solvent increases the relative activation energy between the transition state and the reactants. Hence the reaction rate is decreased.
13.3 Effects of organic solvents on phase-transfer catalysis
811
Table 13.3.8. Effect of the aprotic solvent on the initial reaction rate of the allylation of phenoxide
Solvents
Dielectric constant (20oC)
Initial reaction rate×103 Mh-1 (-r)i,B
(-r)i,PEG
(-r)i,PEG/(-r)i,B
n-decane
1.991
171.77
355.62
2.07
cyclohexane
2.023
127.5
277.68
2.16
ethyl ether
4.335
98
205.83
2.10
cyclobenzene
5.708
75.25
174.56
2.32
dichloromethane
9.080
45.61
86.31
1.89
Data obtained from the work of Wang and Chang142
13.3.1.3 Effects of the organic solvents on the reactions in other catalysts (A) Quaternary ammonium salts as NPTC The effects of the organic solvents on the reaction rate are given in Table 13.3.9.121 The rates of the reactions of the tetra-n-butylammonium and potassium salts of phenoxide with 1-chlorobutane and 1-bromobutane in pure solvents and solvent mixtures varying in dielectric constant from 2.2 to 39 were obtained by Uglestad et al.121 Table 13.3.9. Effect of organic solvent on the reaction of tetra-n-butylammonium salts of phenoxide and potassium salts of phenoxide with halobutane Rate constant×105, Lmol-1s-1 Reactants Solvents
1-C4H9Cl
1-C4H9Br
1-C4H9Br
K+-OC6H5
Bu4N+-OC6H5
K+-OC6H5
Bu4N+-OC6H5
0.01
330
2.2 6
Dioxane 10% CH3CN 50%
0.0025
2.8
0.22
400
CH3CN
0.084
4.0
12
600
0.33
2.2
40
300
Acetonitrile
Dielectric constant ( ε)
1-C4H9Cl
39
Data obtained from Uglestad et al.,121 0.2 M C6H5O-, 0.05 or 0.1 M C4H9X, 25oC
The rates of reaction of potassium phenoxide vary over three orders of magnitude with the changes in the dielectric constant of solvent, whereas the corresponding rates with the tetra-n-butylammonium salt vary by approximately a factor of 6. (B) Tertiary amines as NPTC Quaternary ammonium salts, PEGs and crown ethers are the common compounds, employed as PTC. The inexpensive tertiary amines have also been used as the phase transfer catalysts (PTC) in recent years. The synthetic process for producing 2mercaptobenzimidazole (MBI) is a reaction of o-phenylene diamine (C6H4(NH2)2) and carbon disulfide (CS2) in a two-phase medium affected by appropriate choice of solvent.128-130
812
Figure 13.3.3. Effect of solvents on the conversion of o-phenylene diamine in the two-phase catalyzed reaction; 3.17x10-3 mol of C6H4(NH2)2, 2.50×10-2 mol of CS2, 1.679×10-3 mol of tributylamine (TBA), 1000 rpm, 30oC. (Adapted from Ref. [128], by permission.)
Maw-Ling Wang
Figure 13.3.4. Effect of solvents on the conversion of o-phenylene diamine; 0.4 g of C6H4(NH2)2, 4.003 g of CS2, 0.4 mL of tributylamine (TBA), 50 mL of organic solvent, 600 rpm, 30oC. (Adapted from Ref. [129], by permission.)
Several solvents, such as: n-decane, n-hexane, benzene, chlorobenzene and dichloromethane, which are immiscible with water, were used. The effect of solvent on conversion is shown in Figure 13.3.3.128 The order of the conversion of o-phenylene diamine C6H4(NH2)2 in various organic solvents is: dichloromethane > chlorobenzene > benzene > n-hexane > n-decane, which is consistent with the order of solvent polarity, i.e., the greater the polarity of solvent the higher the conversion of o-phenylene Figure 13.3.5. Effect of solvents on the conversion of diamine. o-phenylene diamine; 3.18x10-3 mol of o-phenylene Results for the reaction of o-phenyldiamine, 8 molar ratio of CS2/C6H4(NH2)2, 0.01 M of triethylamine (TEA), and 50 mL of CH3CN, 600 rpm, ene diamine and carbon disulfide in a ho40oC. (Adapted from Ref. [130], by permission.) mogeneous phase (organic solvent) are given in Figure 13.3.4.129 The order of the reactivities is: dichloromethane (8.91) > chlorobenzene (5.6) > chloroform (4.8) > toluene (2.4) > benzene (2.3). The reaction rate is related to the dielectric constant of the organic solvent. A larger conversion of o-phenylene diamine was obtained using solvent with a higher dielectric constant. In choosing a polar organic solvent, such as: MeCN, MeOH, EtOH, DMSO, DMF and THF, a homogeneous solution was used for the reaction.130 Figure 13.3.5 shows the effects of organic solvents (protic or aprotic) on the conversion of o-phenylene diamine. The order of the reactivities for these six organic solvents is: DMF > DMSO > > MeCN > MeOH > EtOH >THF. The corresponding dielectric constants of solvents are: DMF (37.71), DMSO (46.45), MeCN (35.94), MeOH (32.66), EtOH (24.55) and THF (7.58), respectively. The protic solvents, such as MeOH and EtOH, containing hydroxyl group possess acidic proper-
13.3 Effects of organic solvents on phase-transfer catalysis
813
ties. The unpaired electrons on the oxygen atom associate with the anions. A relatively lower conversion is obtained in MeOH or EtOH solvent. This result indicates that the acidic hydrogen bond does not have a strong catalytic capability. The aprotic solvents, which do not possess hydrogen bond, are highly polar. Therefore, the aprotic solvents possess high alkalinity and nucleophilicity required to obtain a high conversion of o-phenylene diamine in the synthesis of mercaptobenzimidazole (MBI). A larger conversion is obtained when using a protic solvent or aprotic solvent of high polarity. However, the structure of DMF, which is an amide, is similar to that the tertiary amine. It possesses similar catalytic property to dimethylaminopyridine (DMAP). The effect of DMF on the conversion of o-phenylene diamine is more pronounced than that of DMSO. The Arrhenius rate equations in various solvents for the reaction of o-phenylene diamine and carbon disulfide catalyzed by tributylamine are as follows: DMF:
kapp = 1.06x1015exp(-1.20x104/T)
DMSO:
kapp = 7.82x108exp(-7.78x103/T)
MeCN:
kapp = 1.39x1013exp(-1.09x104/T)
MeOH:
kapp = 9.62x1014exp(-1.24x104/T)
EtOH:
kapp = 3.84x1010exp(-9.29x103/T)
THF:
kapp = 3.25x1038exp(-2.99x104/T)
[13.3.10]
kapp is the apparent rate constant in which the reaction follows pseudo-first-order rate law. In two-phase phase-transfer catalytic reactions, the solvents significantly affect the reaction rate. The main reason is that the distribution of regenerating catalyst QX and the active catalyst QY between two-phases is highly dependent upon the polarity of the solvent. It is desirable for most of the intermediate products to stay in the organic phase and react with the organic-phase reactant. Therefore, a solvent with high polarity will be preferred for the reaction. (C) Pyridine 1-oxide (PNO) as IPTC The substitution reaction of benzoyl chloride (PhCOCl) and sodium acetate (CH3COONa) using pyridine 1-oxide (PNO) as the inverse phase-transfer catalyst (IPTC) in a two-phase system of organic solvent and water was investigated by Wang, Ou and Jwo.148-150 They found that the polarity of the organic solvent strongly affected conversion of benzoyl chloride, the yield of the main product (acetic benzoic anhydride (PhCOOCOCH3)), and the reaction rate. The reaction follows a pseudo-first-order kinetic rate law. Dichloromethane, chloroform, tetrachloromethane and cyclohexanone (C6H10O) were used as the organic modifier in the two-phase reaction system. The results are given in Table 13.3.10. A linear reaction rate was observed for a more polar organic solvent. The order of relative reactivities in these solvents is cyclohexanone > dichloromethane > chloroform > tetrachloromethane, consistent with their polarities. Kuo and Jwo54 obtained similar results. Wang, Ou and Jwo148 also found that the conversion was substantially increased with initial concentration of PNO increasing in the aqueous phase with CH2Cl2 present as
814
Maw-Ling Wang
the organic solvent. The reason is that the concentration of carboxylate ion54 influences the concentration of PNO in the organic phase. Table 13.3.10. Effect of the composition of organic solvent on the PNO-catalyzed PhCOCl-CH3COONa reaction in a two-phase medium
Organic phase
kapp×10-3 min-1 at T= 5oC
10oC
18oC
25oC
24.7
32.3
48.5(5.73)b
65.8(12.0)b
18.8
26.6
51.5 26.8 18.1
72.8
5.13
38.3 17.9 9.06
36.4(6.42)b
62.7(14.6)b
83.0
124
CH2Cl2+CHCl3 [CHCl3]=5.00 M
26.9
37.6
49.0
CH2Cl2+C6H10O [C6H10O]=3.00 M
35.0(4.49)b
57.2(9.38)b
91.4(20.4)b
14.6
19.5
26.7
CH2Cl2 CH2Cl2+CCl4 [CCl4]=1.00 M [CCl4]=3.00 M [CCl4]=5.00 M CH2Cl2+C6H5NO2 [C6H5NO2]=1.00 M
CHCl3
28.0(2.93)b
16.2
CCl4 C6H10O
33oC
68.7
106
170
Data obtained from Wang et al.148 2.00×10-4 M of PNO, 1.00×10-2 M of PhCOCl, 0.500 M of CH3COONa, 18oC, 1200 rpm, 50 mL of H2O, 50 mL of organic solvent, bNo PNO added; C6H10O, cyclohexanone
Figure 13.3.6. Effect of the mole fraction of CCl4 in the mixed organic solvent on the kapp value in the two-phase H2O/(CH2Cl2+CCl4) medium; 1.00×10-2 M of PhCOCl 0.500 M of CH3COONa, 2.00×10-4 M of PNO, 50 mL of H2O, 50 mL of organic solvent (CH2Cl2+CCl4), 18oC. (Adapted from Ref. [148], by permission.)
In the studies on the inverse phase-transfer catalytic reaction, Wang, Ou and Jwo148 conducted two independent experiments in order to evaluate the effect of polarity of the organic phase on the reaction. In the first experiment, a relatively inert organic substances such as C6H5CH2CN, C6H5N(Et)2, C6H5NO2, C6H5CN, CH3COOC2H5 or C3H7COOC2H5 were individually added to the organic phase (CH2Cl2) as the mixed organic solvent in the two-phase reaction system. The reactions of these compounds with PhCOCl or PNO were negligibly slow compared to the reaction of PhCOCl and PNO. The results are given in Tables 13.3.11 and 13.3.12, respectively. It is shown that kapp increased with added inert substance of high polarity,
13.3 Effects of organic solvents on phase-transfer catalysis
Table 13.3.11. Effect of the inert organic substance on the PNO-catalyzed CH3COONa-PhCOCl reaction in a twophase H2O/CH2Cl2 medium Organic substance, R
kapp×103 min-1
Dipole moment, D
C6H5CH2CN
61.2
C6H5CN
60.5
C6H5N(Et)2
56.8
C6H5NO2
57.0
4.22
CH3COOC2H5
53.1
1.78
C3H7COOC2H5
48.5
CH2Cl2
48.5
1.60
CCl4
42.5
0
4.18
Data obtained from Wang et al.148 1.00×10-2 M of PhCOCl, 0.500 M of CH3COONa, 0.500 M of R, 2.00×10-4 M of PNO, 50 mL of H2O, 50 mL of CH2Cl2, 18oC
815
such as nitrobenzene and ethyl acetate or basic organic substance, such as diethylaniline. The kapp-value increased to a greater extent with added highly polar and basic organic substance, such as benzyl cyanide and benzonitrile. In the second set of experiments, reactions were carried out with nonpolar CCl4 added to the CH2Cl2 as the mixed organic solvent. As shown in Figure 13.3.6, that due to decreased polarity, the value of kapp also decreased with increased amount of added CCl4 to a minimum.148 Then, it increased slightly on further addition of CCl4 due to the increased rate of PNO-catalyzed hydrolysis of PhCOCl. Since the distribution of PhCOCl in the CH2Cl2 decreases with increased amount of CCl4, the reaction rate of PhCOCl with PNO in the aqueous phase leads to the hydrolysis of PhCOCl. Wang, Ou and Jwo148 observed that the yields of PhCOOCOCH3 decrease with increased content of CCl4.
Table 13.3.12. Effect of the amount of inert organic substance on the PNO-catalyzed CH3COONa-PhCOCl reaction in a two-phase H2O/CH2Cl2 medium kapp×103 min-1 with [R]org, M
Inert organic Substance [R]org
0.100
C6H5CH2CN
57.3
C6H5NO2
49.0
C6H5N(Et)2
55.1
CCl4
47.6
0.300
61.1
0.500
1.00
1.50
61.3
62.4
71.8
57.0
62.7
60.7
60.9
38.3
28.5
24.7
56.8 42.1
0.800
2.00
54.9
Data obtained from Wang et al.148 1.00×10-2 M of PhCOCl, 0.500 M of CH3COONa, 2.00×10-4 M of PNO, 50 mL of H2O, 50 mL of CH2Cl2, 18oC
In Table 13.3.12, kapp approached a constant value when nitrobenzene (1.0 M) was added. This result indicates that solvation of the transition structure for the reaction of benzoyl chloride with sodium nitrate reached an upper limit. Benzyl cyanide is a polar solvent. The value of kapp increased with increased content of benzyl cyanide. Further, highly basic diethylaniline (C6H5N(Et)2) could increase the concentration of free PNO and also the reaction rate. However, this compound is less polar than dichloromethane and the polarity decreased with increased proportion of diethylaniline, which caused the value of kapp to de-
816
Maw-Ling Wang
crease. Therefore, the value of kapp reached a maximum, as shown in Table 13.3.12. In general, the value of kapp increased with an increased proportion of highly polar inert organic substance, such as C6H5CH2CN and C6H5NO2, and decreased with increased proportion of slightly polar inert organic substance, such as C6H5N(Et)2 and CCl4. In the case of CCl4, the greater the proportion of nonpolar or less polar compound, the smaller the reaction rate. The results are due to a combination of the hydrolysis of benzoyl chloride, the distribution of PNO between two phases, and the mass transfer of PNO from organic phase to aqueous phase. (D) Electrophile reaction by Table 13.3.13. Initial rates of coupling between NaDBS 4-nitrobenzene diazonium chloride and Although the electrophile N-ethylcarbazole at 0oC in various solvent mixtures transferred to the organic phase in the presence and absence of additives from the aqueous phase by phase transfer catalysis (PTC), the role c 9 Rate ×10 , of organic solvent was still imporAdditveb Solventa -1 -3 mol dm tant. Ellwood and Griffiths24 carCH2Cl2-H2O none 1.25 ried out the coupling reactions between 4-nitrobenzendiazonium NaDBS 11.1 CH2Cl2-H2O chloride and N-ethylearbazole or 18-crown-C-6 ether 0.28 CH2Cl2-H2O N,N-diphenylamine in aqueous Lissapol NXd 0.56 CH2Cl2-H2O media. The coupling reactions were accelerated by using a none < 0.1 Toluene-H2O two-phase water-dichloroNaDBS < 0.1 Toluene-H2O methane containing sodium 4-dodecylbenzene sulfonate none 0.89 EtOAc-H2O (NaDBS) as a transfer catalyst for NaDBS 1.03 EtOAc-H2O the diazonium ion. Effects of solnone 1.39 PhNO2-H2O vents and catalyst (NaDBS) on the rate constants are given in TaNaDBS 10.5 PhNO2-H2O ble 13.3.13. The NaDBS (0.05 none 0.47f AcOH-H2Oe molar proportions) increases the DMF-H2Og none 0.53f rate coupling in dichloroh methane-water by a factor of at 1,4-Dioxane-H2O none 1.33 least 20 relative to the reaction in water-acetic acid. A part of this 24 a Data obtained form the work of Ellwood and Griffths; v/v=1/1 except where stated, b0.05 mmol except where stated, cinitial rate of increase is attributed to incomformation of azo dye, dcommercial (ICI) non-ionic detergent, 0.05 plete solubility of N-ethyl mmol, e95% H2O v/v, containing NaOAc·3H2O, fsuspension of carbazole in the latter solvent. g h N-ethylcarbazole, 60% H2O v/v, 60% H2O v/v homogeneous soluAlso, the polarity of the organic tion containing 1.0 mmol diazonium ion and 1.0 mmol N-ethyl phase is important (cf. CH2Cl2, carbazole 100 mL solvent C6H5NO2 have much higher dielectric constants than toluene and ethyl acetate). This may be attributed to the covalent character of diazonium arylsulfonates. Its ionization is greater in the former solvents. Crown ethers act as transfer agents for the diazonium ion, but the resultant complexes have low coupling reactivity.
13.3 Effects of organic solvents on phase-transfer catalysis
817
(E) Oxidation by dimethyl polyethylene glycol and oxidant Dimethyl polyethylene glycol solubilizes potassium permanganate in benzene or dichloromethane and can thus be used as a phase-transfer agent for permanganate oxidation. The reaction is highly dependent on the organic solvent. If benzene is used as solvent, dimethyl polyethylene glycol does not efficiently extract KMnO4 from an aqueous solution, but it solubilizes the solid reagent when CH2Cl2 is used as the solvent, KMnO4 may be transferred from either aqueous solution or from the solid phase. The effect of organic solvent on the distribution of products is given in Table 13.3.14.58 Table 13.3.14. Oxidation of cyclododecene Phase transfer agent
Solvent system
Oxidation a ratio
Benzene+17% acetic acid
3.3
Polyetherb
1,2-Cyclododecanedione (16), dodecanedioic acid (59), cyclododecane (23)
Benzene+17% acetic acid
3.3
Crown etherc
1,2-Cyclododecanedione (22), dodecanedioic acid (56), cyclododecane (12)
Benzene+17% acetic acid
3.3
Adogen 464
1,2-Cyclododecanedione (8), dodecanedioic acid (58), cyclododecane (9)
Dichloromethane+17% acetic acid
3.3
Polyether
1,2-Cyclododecanedione (8), dodecanedioic acid (77), cyclododecane (1)
Dichloromethane+17% acetic acid
3.3
Crown ether
1,2-Cyclododecanedione (7), dodecanedioic acid (83), cyclododecene
Dichloromethane+17% acetic acid
3.3
Adogen 464
1,2-Cyclododecanedione (7), dodecanedioic acid (83), cyclododecene (2)
Dichloromethane+17% acetic acid
2.2
Adogen
1,2-Cyclododecanedione (18), dodecanedioic acid (63), 2-hydroxycyclododecanone (6)
Dichloromethane+17% acetic acid
1.6
Agogen 464
1,2-Cyclododecanedione (14), dodecanedioic acid (40), 2-hydroxycyclododecanone (6), cyclododecene (23)
Dichloromethane+17% acetic acid
2.2
Adogen 464
1,2-Cyclododecanedione (19), dodecanedioic acid (27), 2-hydroxycyclododecone (7), cyclododecene (23)
Dichloromethane+10% acetic acid
2.2
Adogen 464
1,2-Cyclododecanedione (69), dodecanedioic acid (13), 2-hydroxycyclododecane (3), cyclododecene (9)
Dichloromethane/water +10% acetic acid
2.2
Polyether
1,2-Cyclododecanedione (16), dodecanedioic acid (82)
Dichloromethane/aqueous NaOH
1.0
Benzyl triethyl ammonium chloride
1,2-Cyclododecanediol (50)
Products (%)
Data obtained from Lee and Chang;58 anumber of moles of potassium permanganate per mole of alkene, b dimethylpolyethylene glycol, cdicyclohexano-18-crown-6 ether
818
Maw-Ling Wang
(F) Polymerization by PTC Poly(ethylene glycol)-block-poly(butylacrylate), synthesized by radical polymerization,56 were obtained by PTC in the Williamson reaction. The morphology and the crystallinity of the cast film of the block polymer were significantly affected by the organic solvent. The technique of phase-transfer catalysis has been extensively applied to the two-phase polycondensation using various phase-transfer catalysts, such as quaternary ammonium and phosphonium salts, crown ethers and poly(ethylene glycol)s.8,11,30,46,53,75,87,119,151 Various types of condensation polymers such as aromatic polysulfonates and polysulfides, aromatic polyethers, aliphatic and aromatic polysulfides, and carbon-carbon chain polymers of high molecular weights by the phase-transfer catalyzed polycondensation from combinations of aromatic disulfonyl chlorides, phosphonic dichlorides, activated aromatic dichlorides, and aliphatic dihalides, with bisphenol, aliphatic and aromatic dithiols, and active ethylene compounds. The two-phase polycondensation was generally carried out in a water-immiscible organic solvent-aqueous alkaline solution system at room temperature. The method of polycondensation offers a highly versatile and convenient synthetic method for a variety of condensation polymers. Aromatic polysulfonates of high molecular weights can be prepared from aromatic disulfonyl chlorides and alkaline salts of bisphenols by interfacial polycondensation technique using onium salt accelerators. In the absence of the catalyst, only low molecular polysulfonate III was obtained, even though the reaction was continued for 254 hours, whereas the addition of these quaternary ammonium salts and crown ethers increased the average molecular weight of the polymer remarkably.
[13.3.11]
Among the catalysts, TBAC and DC-18-C-6 were found to be highly efficient, leading to the formation of the polysulfonate with an inherent viscosity [η] of as high as 1.4 dLg-1. Similarly, aromatic polysulfonates and aromatic polyether were synthesized from the polycondensation of phenylphosphonic dichloride (IV) with bisphenol A (II) leading to a polyphosphonate (V), aromatic dihalides (VI) with alkaline salts of bisphenols (VII), under various phase transfer catalysis in two-phase system, i.e.,
[13.3.12]
13.3 Effects of organic solvents on phase-transfer catalysis
819
[13.3.13]
The effect of organic solvent on the two-phase polycondensation is shown in Table 13.3.15. Chloroform, 1,2-dichloroethane, nitrobenzene, acetophenone and anisole were all effective as the polymerization media to produce moderate molecular weight polymer (VII). Table 13.3.15. Synthesis of aromatic polyether VII in various organic solvent/water system with DC-18-C-6 catalyst Reaction temp., oC
Reaction time, h
Polymer [η], dLg-1*
CH2Cl2
20
24
0.84
CHCl3
20
24
0.53
CH2ClCH2Cl
20
24
0.42
C6H5NO2
20
24
0.47
C6H5NO2
80
2
0.51
C6H5NO2
100
1
0.42
Solvent
Data obtained from the work of Imai;41 Reaction conditions: 2.5 mmol of II, 2.5 mmol of VI, 0.05 mmol of DC-18-C-6 in 3.5 mL of solvent, and 5 mL of KOH (1.01 M) solution, *Measured at a concentration of 0.5 dLg-1 in DMF at 30oC
A convenient method for the preparation of polysulfides by the two-phase polycondensation in a KOH solution is known.41 Polycondensation of 1,4-dibromobutane (VIII) and 1,6-hexanedithiol (IX) leading to polysulfide (X) was carried out in various organic solvent/H2O system with DC-18-C-6 catalyst.
[
]
Br (CH2 ) 4 Br + HS(CH2 ) 6 SH → −(CH2 ) 4 − S − (CH2 ) 6 − S − VIII IX X
n
[13.3.14]
The results of the polycondensation are given in Table 13.3.16. All polymerization media employed produced polysulfide with moderately high inherent viscosities; whereas the polymer with the highest viscosity was produced in the absence of organic solvents. Polycondensation conducted in the presence of any catalyst in this system led to the formation of a polymer with moderately high molecular weight.
820
Maw-Ling Wang
Table 13.3.16. Synthesis of aliphatic polysulfide Several polycarbonates59 were X in various organic solvent-water system with synthesized by two-phase DC-18-C-6 catalyst polycondensation of bisphenols and
brominated with trichloromethyl chloroformate in a system of an organic solvent and aqueous alkaline 20 0.30 CH2Cl2 solution of quaternary ammonium 80 0.31 CHCl3 salts. Chlorinated hydrocarbons, di80 0.30 C6H6 chloromethane (DCM), tetrachloromethane (TCM), tetrachloromethane 80 0.58 C6H5NO2 (TCM) and nitrobenzene (NB) 80 0.58 CH3CN served as organic solvents. The effects of solvents on the reaction None 80 0.73 yields are given in Table 13.3.17.59 41 Although polymers with a high yield Data obtained from Imai; Polymerization conditions: 2.5 mmol of VIII and IX, 0.05 mmol of DC-18-C-6 in 2.5 mL of solvent and were obtained using nitrobenzene 5 mL of 1.01 M KOH solution for 48 h, *measured at a concentra- (NB) as an organic solvent, the intion of 0.5 gdL-1 in chloroform at 30oC herent viscosities were low. Polycarbonates were prepared by a two-phase condensation of TCF with bisphenol S. They precipitate from chlorinated hydrocarbon solvents such as DCM, TCM and DCE. According to both the yield and the inherent viscosity of these polymers, the use of BTEAC as a phase-transfer catalyst, sodium hydroxide as a base and DCE as an organic solvent was suitable to prepare a polycondensate having a large molar mass and a high yield. Reaction temp. oC Polymer [η], dLg-1*
Solvent
Table 13.3.17 Synthesis of bisphenol S-based homopolycarbonate by two-phase polycondensation catalyzed by PTCa Reaction conditions b
Polymer yield
Solvent
Catalyst
%
[η], dLg-1*
State
DCM
TBAB
76.7
0.21
ppt.
DCM
TBAC
76.6
0.10
ppt.
DCM
BTEAC
78.7
0.13
ppt.
DCM
BTEAB
79.3
0.20
ppt.
TCM
TBAB
60.0
0.11
ppt.
TCM
BTEAC
83.4
0.12
ppt.
DCE
TBAB
88.9
0.28
ppt.
DCE
TBAC
76.4
0.12
ppt.
DCE
BTEAC
86.5
0.32
ppt.
DCE
BTEAB
89.3
0.21
ppt.
NB
TBAB
85.0
0.17
solution
NB
TBAC
90.6
0.11
solution
13.3 Effects of organic solvents on phase-transfer catalysis
Reaction conditions b
821
Polymer yield
Solvent
Catalyst
%
[η], dLg-1*
State
NB
BTEAC
92.2
0.19
solution
Data obtained from Liaw and Chang;59 aPolymerization was carried out with bisphenol S (5.00 mmol) and TCF (7.50 mmol) in the organic solvent (37.5 mL) and water (30 mL) in the presence of catalyst (3.15 mmol) and sodium hydroxide (28.5 mmol) at room temperature for 2 h. bAbbreviations: DCM, dichloromethane; TCM, tetrachloromethane; DCE, 1,2-dichloroethane; NB, nitrobenzene. *Measured at a concentration of 0.5 gdL-1 in DMF at 25oC
Another type of polysulfide (XIII) was synthesized by the two-phase polycondensation of bis-(3-chloroacryloy)benzenes (XIa and XIb) with 4,4’-oxybisbenzenethiol (XII). The polycondensation was carried out in a chloroform-water system at room temperature with some phase transfer catalysts.
[13.3.15]
Table 13.3.18 shows the results of polycondensation. The polysulfides having inherent viscosities above 0.5 dLg-1 were readily obtained from two bis(2-chloroacryloyl)benzene with or without use of phase transfer catalysts. These activated dichlorides are highly reactive, almost comparable to ordinary dicarboxylic acid chlorides. The use of catalysts, such as DC-18-C-6, was not essential to this type of polycondensation for producing high molecular weight of polysulfides XIII. Table 13.3.18. Synthesis of polysulfides XIII in organic solvent-water systema Dichloride
Solvent
Catalyst
Reaction time, min
Polymer, [η], dLg-1*
XIa
chloroform
none
15
0.21
XIa
chloroform
none
60
0.61
XIa
chloroform
DC-18-C-6
10
0.62
XIa
chloroform
TBAC
60
0.72
XIa
dichloromethane
DC-18-C-6
60
0.55
XIb
chloroform
none
60
0.55
XIb
dichloromethane
none
15
0.42
XIb
dichloromethane
DC-18-C-6
15
0.51
Data obtained from Imai;41 aReaction conditions: 2.5 mmol of XI, 2.5 mmol of XII, 0.05 mmol of catalyst, 5 mL of solvent, 5 mL of 1.01 M KOH at 15oC under nitrogen, *Measured at a concentration of 0.5 gdL-1 in concentrate sulfuric acid at 30oC
822
Maw-Ling Wang
13.3.1.4 Effect of the volume of organic solvent and water on the reactions in various reaction systems In general, the concentration of reactants in the aqueous phase is decreased by increased amount of water. The addition of water probably also decreases the concentration of the intermediate product in the organic phase. Hence, both the mass transfer rate and the degree of hydration with the anion are decreased, which also decreases the reaction rate. However, this argument is not necessarily correct. Figure 13.3.7 indicates that the conversion in the two-phase reaction is not affected by the amount of water added.147 Wang and Yang135 studied the effects of the volume ratio of water to chloroFigure 13.3.7. Effect of the amount of water on the appar- benzene on the conversion for the reaction ent rate constant (kapp); 4 g of 2,4,6-tribromophenol, 0.9 g of KOH, 0.6 mL of benzyl bromide, 50 mL of H2O, 50 of 4-bromophenol and allyl bromide in an mL of chlorobenzene, 40oC. (Adapted from Ref. [147], alkaline solution of KOH/chlorobenzene at by permission.) 50oC under phase-transfer catalytic conditions. The reaction followed the pseudo-first-order rate law and the corresponding apparent rate constant decreased gradually when the water content was increased, as shown in Figure 13.3.8.135 The reason was that the concentration of the intermediate product tetra-n-butylammonium phenoxide (ArOQ, or the active catalyst) in the aqueous phase decreased with the increase in the amount of water. The mass transfer rate of the intermediate product (or the active catalyst) from the aqueous phase to the organic Figure 13.3.8. Effect of the volume ratio of water to chlorobenzene phase decreased when a large on the conversion; 1.568 g of 4-bromophenol, 1.0 g of KOH, 0.7 g of amount of water was used. In adallyl bromide, 0.2 g of TBAB catalyst, 50 mL of chlorobenzene, dition, the dilution effect led to re50oC. (Adapted from Ref. [135], by permission.) duction of the reaction rate in the aqueous phase. In general, a higher concentration of the intermediate product (tetra-n-butylammonium alkoxide, or the active catalyst, ArOQ) in the aqueous phase enhances the reaction rate. This is due to a large concentration gradient across the interface in transferring the species from the aqueous phase to the organic phase. For the reaction of allyl bromide and 2,4-dibromophenol in synthesizing 2,4-dibromophenyl allyl ether in an
13.3 Effects of organic solvents on phase-transfer catalysis
823
alkaline solution of KOH/chlorobenzene two-phase medium under PTC conditions,132 the conversion increases with the increase in the concentration of ArOQ in the aqueous phase (or decreasing content of water). However, this change is small, reflecting a small mass transfer resistance, as shown in Figure 13.3.9.132 The influence of the amount of water on the conversion in the reaction of carbon disulfide and o-phenylene diamine catalyzed by tertiary amine in a two-phase medium was studied. The conversion decreased with the increase in the Figure 13.3.9. Effect of the water content on the conversion; 2.28 g amount of water. Therefore, the of 2,4-dibromophenol, 0.2 g of TBAB catalyst, 0.7 g of allyl brovalue of apparent rate constant mide, 1.0 g of KOH 50 mL of chlorobenzene, 50oC. (Adapted from (kapp), in which the reaction folRef. [132], by permission.) lows pseudo-first-order-rate law, decreases with the increase in the volume of water.128 Wang and Chang144-146 found that the conversion increases with the increase in the volume ratio of water to organic solution up to 1/5. The conversion is independent of the volume ratio of water to organic phase (chlorobenzene), greater than 1/5. The reason is that the reaction is carried out in a large amount of KOH (solid form). Probably, the omega phase is generated for the volume ratio of water to chlorobenzene less than 1/5. However, this change in the conversion vs. the volume ratio of water to chlorobenzene is not significant. (A) Omega phase reaction It is found that 92% of the 18-crown-6 ether added to a salt (KCN and KCl) and toluene system resided in the organic phase. However, all but approximately 1-2% of the crown ether was translocated onto the surface of the salt upon addition of small quantities of water. The results of Liotta et al.62 are given in Table 13.3.19. The initial water added to the system coats the surface of the salt particles and it was this aqueous salt coating that extracted the crown from the organic phase. Liotta et al.61,62 called this new region of the reaction system the omega phase. The 8% of the crown located on the surface of the salt particles prior to the addition of water was probably due to the presence of water already present in the salt. The distribution of 18-crown-6 ether between the organic phase and the omega phase was determined.62 The amount of crown ether in the organic phase remained low and relatively constant (0.06-0.07 milimoles of 18-crown-6 ether in organic phase). The omega phase adsorbed most of added crown ether. For the accompanying pseudo first-order kinetics reaction of benzyl bromide with potassium cyanide, the results are given in Table 13.3.20.122 There is a slight increase in the rate as the number of millimoles of 18-crown-6 ether increases, but the rate remains essentially constant with the increase of crown ether.
824
Maw-Ling Wang
Table 13.3.19 Effect of added water on the concentration of 18-crown-6 ether in toluene at room temperatures Water, µL
Equiv. of water, mole of H2O/mole of crown
Percent crown in toluene, %
0
0.00
91.5
10
0.14
81.4
15
0.21
77.3
22
0.31
50.0
25
0.35
34.6
30
0.42
17.7
45
1.25
2.5
50
1.39
2.0
80
2.22
1.0
Table 13.3.20 18-Crown-6 catalyzed reactions of benzyl bromide with KCN as a function of added crown ether Millimoles of 18-crown-6
k×105 sec-1
3.0
2.16, 2.47
5.0
3.97, 3.63
7.0
3.86, 3.99
10.0
3.75, 4.00
12.0
3.80, 3.60
Data obtained from Vladea and Simandan;122 1.0 mL of H2O, 0.15 mole KBr, 0.15 mol of KCN, 50 mL of toluene, 25oC
Data obtained from Liotta et al.62 0.0040 mole of 18-crown-6, 0.027 mol of KCN, 10 mL of toluene
(B) Reaction catalyzed by PEGs The structure of polyethylene glycol (HO(CH2CH2O)nH, PEG) is similar to that of crown ether. Polyethylene oxide chains (CH2CH2O) form complexes with cations, much like crown ethers, and these complexes cause the anion to be transferred into the organic phase and to be activated.122 Table 13.3.21 shows the binding constant, K, for PEG complexes with sodium cation depend on both the value of n (i.e., average molecular weight of PEG or number of (CH2CH2O) unit) and on the end-group substituents.112,114,115 Gokel and coworkers31,32 determined the binding strength for Na+ in anhydrous methanol solution with PEGs and obtained the binding constant K=1.4. They concluded that the strength of complexation is a function of the total number of binding sites present and not the number of polymer chains, suggesting that a long PEG chain may be involved in binding more than one cation. PEGs and their derivatives have been extensively investigated as phase transfer catalysts and are used in many commercial processes. In the absence of strong acids, PEGs are nontoxic, inexpensive, and thermally stable. For some reactions such as with hydroxide transfer, PEGs are excellent catalysts, sometimes better than crown ethers, especially when used in liquid-solid PTC reactions with potassium salts, and with little or no added water, and with at least moderately polar organic solutions. PEGs are water soluble and if the organic phase is not sufficiently polar the PEG will reside almost completely in the aqueous phase; or with concentrated aqueous solutions of organic salts, the PEG may form a third catalyst-rich phase, a change that normally leads to a high level of catalytic activity. PEGs are themselves soluble in water. To obtain partitioning of PEG into an organic solution may require use of a mono- or diether derivative. Harris and Case34 found that with
13.3 Effects of organic solvents on phase-transfer catalysis
825
exception of dichloromethane as an organic-phase solvent, most PEGs are themselves partitioned into the aqueous phase, depending also on the concentration of dissolved salts in the aqueous phase.35 To improve organic-phase solubility of PEG, several dialkyl ethers of PEGs as permanganate-PTC catalysts are suggested in Table 13.3.22. Table 13.3.21 Binding constants for complexation of PEGs and some symmetrical derivatives with sodium cation Na+ + PEG = [Na+⋅PEG] complex Log K (binding constant) with various sodium salts
PEG Avg. MW
Avg. n
HO-
200
4.1
1.64
300
6.4
2.02
400
8.7
2.26
600
13.2
2.59
2.09
1.99
1.87
1000
22.3
2.88
2.55
2.48
2.37
1500
33.7
3.09
2.86
2.80
2.68
2000
45.0
3.28
3.08
3.05
2.81
CH3O-
C2H5O-
PhO-
cyC5H10N-
0.5 1.55
1.25
1.05
1.16
1.49
1.51
2.46
3.08
Data obtained from Szabo et al.112,114,115
Aliphatic hydrocarbons are im-
Table 13.3.22 Partition of PEG-dialkyl ethers miscible with PEGs. Therefore, it is between C6H6 and H2O
important to select a good organic-phase solvent such as aromatic Partitioning, % in hydrocarbons, chlorinated hydrocarPEG Ether C6H6/% in H2O bons, or acetonitrile.80 In toluene, 14 C4-PEG1500-C4 PEGs are more effective catalysts than crown ethers for the reaction of benzyl 84 C6-PEG1500-C6 chloride and solid potassium acetate. Emulsion C18-PEG6000-C18 In butanol, the effectiveness of PEGs 108 C18-PEG750-Me and crown ethers as phase-transfer catalysts were the same for the reaction of 39 C18-PEG1900-Me benzyl chloride and solid potassium 37 C18-PEG5000-Me acetate.27 12 C8-PEG5000-Me 13.3.1.5 Effects of organic solvents on other 13 C4-PEG5000-Me phase-transfer catalytic PEG6000 <1 reactions (A) Liquid-liquid-liquid three Data obtained from Harris et al.34,35 C4-PEG1500-C4 represents phase reaction a PEG of MW 1500, capped by a butyl group at both ends Not only will the organic solvent influence the NPTC reaction, but it also affects the character of NPTC. Another type of transfer mechanism takes place when the
826
Maw-Ling Wang
quaternary salt catalyst is not highly soluble in either the aqueous or the organic phase, but instead forms a third phase.154 For example, tetrabutylammonium salts in the presence of highly concentrated aqueous solutions, and with toluene as solvent for the organic phase, form a third (quaternary compound) phase.126 Jin et al.48 investigated the relationship between the properties and the catalytic activity of the liquid-liquid-liquid three-phase phase-transfer catalytic system. The condition of formation of the third phase are investigated by changing the kind of PTC and organic solvent. With tetra-n-butylammonium bromide Bu4N+Br- is used as the phase transfer catalyst, the third phase is formed with both dodecane and toluene as the organic solvents. On the other hand, when tetra-n-hexylammonium bromide (Hex)4N+Br- is used as the phase transfer catalyst, the third phase forms with dodecane but not with toluene. In such situations, most of the reaction actually occurs in the third phase with both aqueous and organic reagent transferring to this phase for conversion. Third-phase reactions of this type may be faster than simple PTC reactions. Because formation of the third phase offers simplified catalyst removal and recovery procedures, third-phase catalyst is highly attractive for commercial operations. In the conversion of benzyl chloride to benzyl bromide using tetra-n-butylammonium bromide as the catalyst, a third phase was also observed.126 More rapid reaction rates were obtained in the presence of this additional phase. The kinetics associated with the base-catalyzed isomerization of p-allylanisole in the presence of a variety of polyethylene glycols (PEGs) has been reported.79 The reaction followed first-order kinetics in p-allylanisole and the reaction system was characterized by three phases consisting of an organic solvent phase, an aqueous base phase, and a complex liquid phase consisting of PEG and potassium hydroxide. It was suggested that the isomerization reaction took place in the complex third phase. PEG forms a third phase between toluene and aqueous KOH when some methanol is added.39 Some methanol must be added to reduce the amount of PEG that otherwise would dissolve in the toluene phase. Thus, dehydrobromination of 2-bromooctane in toluene, using PEG as catalyst, could be accomplished by removal and replacement of the organic and aqueous phases after completion of reaction, and recycle of the catalyst phase. After four cycles no catalyst was lost. Reaction rates for KOH dehydrohalogenation of 2-bromooctane in toluene with PEG catalysts were increased by a maximum factor of 126 by addition of methanol.38,40 The base efficiency (moles base per mole of catalyst) PEGs with molecular weights 3000 and 20,000 exceeded unity and reached a maximum of 12 on addition of methanol. Hydrogenative dehalogenation of polychlorinated aromatic halides can be accomplished by hypophosphite reduction using a quaternary ammonium salt as a phase-transfer catalyst in conjunction with a palladium-on-carbon co-catalyst.68 The quaternary salt, being insoluble in both reactant phases, coats the Pd/C catalyst forming a third phase. The strongly alkaline medium and the phase-transfer agent are synergistic. In principle, formation of a third catalyst could be accomplished either by (1) use of a phase-transfer catalyst that has limited solubility in both the organic and aqueous phase, or (2) by use of a special solvent, perhaps a fluorocarbon, having great affinity for the catalyst but only modest affinity for both of the reactants. Tetra-n-butyl-ammonium salts frequently form third layer when used in conjunction with an organic phase that has little polarity (e.g., neat 1-chlorooctane, toluene, or cyclohexane as solvent, but only dichloromethane) and with a concentrated aqueous solution of inorganic salts.
13.3 Effects of organic solvents on phase-transfer catalysis
827
The catalyst may form a third layer during the PTC reaction, when using high salt concentrations and nonpolar organic solvents. Wang and Weng123-126,154 found that a third layer phase was built when using a low polarity organic solvent in the PTC reaction system. This has been observed for PEGs, quaternary onium salts and crown ether. In such cases, catalyst recovery involves a simple phase separation. In industrial kettles (with limited visibility in the reactor) or when a “rag” layer distorts an otherwise sharp phase boundary, a phase separation operation may not be simple. Nevertheless, choosing conditions in which the catalyst separates as a third layer is usually advantageous. A dehydrohalgenation was performed four times consecutively, with no loss in catalytic activity by simply replacing the organic phase and replenishing the aqueous base phase after each use, leaving the third phase containing PEG in the reactor. (B) Electrochemical and PTC reaction The combination of phase transfer catalyTable 13.3.23 Effect of organic solvent on current efficiency sis (PTC) with other processes, such as supercritical fluid extraction22 and electrochemical 5-7,23,120 have been investigated in detail. Organic solvent Current efficiency, % process 23 carried out the anodic oxidation Do and Chou 80 CH2Cl2 of benzyl alcohol in the two-phase system con75 CH2ClCH2Cl taining both the redox mediator, ClO-/Cl-, and a phase-transfer catalyst (PTC). The reaction 86 CHCl3 mechanism and the factors which affect the ef64 CH3COOC2H5 ficiency of benzaldehyde production were explored. The current efficiency is mainly Data obtained from the work of Do and Chou;23 0.5 governed by the pH value and the nature of the o M of C6H5CH2OH in the organic phase; 25 C, 600 rpm, 20 mAcm-2 of current density, 70 mL of organic organic solvent as well as the types and the phase, 70 mL of aqueous phase, 0.005 M of concentration of phase transfer catalyst (PTC). Bu4NHSO4, 1.0 M of NaCl, pH=6.9, electricity When ethyl acetate and chlorinated hydrocarpassed 2 Fmol-1 of benzyl alcohol bons were used as solvents, the current efficiencies were between 64 and 86%, respectively as shown in Table 13.3.23.23 With ethyl acetate as solvent, the current efficiency was less than for chlorinated hydrocarbons. The results indicate that the chlorinated hydrocarbons have a higher extraction capacity for Bu4N+ClO- from the aqueous phase than has ethyl acetate. Similar results were obtained by Dehmlow and Dehmlow.16 The current efficiencies change slightly when different chlorinated hydrocarbons were used as organic solvents. Tsai and Chou120 carried out the indirect electrooxidation of cyclohexanol by using a double mediator consisting of ruthenium and chlorine redoxes in the multiphase system. Table 13.3.24 shows that the current efficiency had the highest value at 83% using carbon tetrachloride as organic solvent.120 The current density decreased in order, carbon tetrachloride > chloroform > toluene > cyclohexane. The selectivity was 100% except when toluene was used as organic solvent. For this case, the concentration of cyclohexanol in carbon tetrachloride is higher than that of the other solvents.
828
Maw-Ling Wang
Table 13.3.24. Effect of organic solvent on the conversion and selectivity Solvent
Conversion, %
Selectivity, %
C.E., %
Energy consumption, KWH/mole
CCl4
42
100
83
0.226
CHCl3
38
100
70
0.268
Toluene
17
81
34
3.154
Cyclohexane
12
100
23
4.662
Data obtained from Tasi and Chou;120 Reaction conditions: 0.8 M of cyclohexanol, 15 mAcm-2 of current density, graphite cathode and graphite anode, 0.007 M of RuO2, pH=4, 0.99 Fmol-1 electricity passed, NaCl saturated solution as electrolyte, 1275 rpm, 5oC
13.3.1.6 Other effects on the phase-transfer catalytic reactions Simple mechanical separation such as filtration, centrifugation or phase separation can be used to separate the product and the phase-transfer catalyst by use of insoluble catalysts. However, the more frequently encountered technical problems in use of PTC for industrial applications is the need to separate the product and the phase-transfer catalyst by chemical equilibrium separation method in the liquid-liquid two-phase phase transfer catalytic reaction. The most commonly used methods for separation of products and PTC catalysts on an industrial scale are extraction and distillation. Other separation methods include sorption3,33,57 and reaction.45 The principle of extraction method used to separate PTC and product is based on solubility of quaternary ammonium salt in alkaline aqueous solution.2,25,104 For example, tetrabutylammonium bromide is soluble to the extent of 27% in dilute (1% NaOH) aqueous solutions, but when the solution is made more concentrated (15% NaOH), the solubility of Bu4N+Br- decreases to 0.07%. When the products are obtained in PTC system, they can be usually separated from PTC by distillation method. PTC catalyst in the distillation residue may sometimes be reusable. With quaternary ammonium salts as catalysts, temperatures above 100-120oC usually result in partial or total decomposition of the quaternary salts to trialkylamines and other products. Mieczynska et al.70 and Monflier et al.72 investigated the hydrogenation and hydroformylation under phase transfer catalytic conditions. They found that the yield of aldehydes obtained in hydroformylation of 1-hexene strongly depends on solvent: 24% in toluene, 53-86% in toluene-water-ethanol mixture and 77-94% in water-ethanol solution. The mixture of water-ethanol as a solvent was also found to be the best for hydrogenation of 1-hexene (96% of hexane). Conversion of Ph2PCH(CH3)(COOH) phosphine into sodium salt Ph2PCH(CH3)(COONa) yields aldehyde in toluene, 92% in toluene-water and 94% in toluene-water-ethanol mixture. In principle, hydroxide anion is very difficult to transfer from aqueous to organic phases, yet it is one of the most valuable and most commonly used anions in the PTC systems. Addition of small amounts of alcohols to PTC systems requiring hydroxide transfer causes a dramatic increase in rates. Therefore, addition of alcohol enhances the PTC reaction as the cocatalytic effect. For example: formation of alkoxide anions, RO-, which are more readily transferred than the highly hydrated hydroxide anion, and which can serve as a strong base just as well as OH-, and solvation of the hydroxide with alcohol rather than with water, making the hydroxide anion more organophilic and more easily transferred.99,100
13.3 Effects of organic solvents on phase-transfer catalysis
829
Cyanide displacements catalyzed by quaternary ammonium salts usually do not proceed without the presence of water to facilitate exchange and transfer of anions. However, PTC displacement depends on alcohol structure. Benzyl alcohol is about 1.5-2 times as effective as either methanol or ethanol. In the synthesis of BTPPC (benzyltriphenylphosphonium chloride) from benzyl chloride and triphenylphosphine, second-order rate constants and activation parameters for the reaction of benzyl chloride and triphenylphosphine were measured in several protic and aprotic solvents covering a wide range of dielectric constant were obtained by Maccarone et al.64 Wang, Liu and Jwo127 also used eight solvents in studying their effect on the reaction of triphenylphosphine and benzyl chloride. They classified these solvents into two categories depending on the solubility of benzyltriphenylphosphonium chloride (BTPPC). Solvents that dissolve BTPPC are acetic acid, dichloromethane, methanol and water. Solvents that do not dissolve BTPPC are acetone, benzene, toluene and ether. In general, triphenylphosphine (TP) does not dissolve in methanol or water. The effect of solvents on the reaction rate was measured by the apparent rate constant in which the reaction follows pseudo-first-order rate law. The order of relative activities of solvents is methanol (0.34 h-1) > acetic acid (0.176 h-1) > dichloromethane (0.0468 h-1) > acetone (0.0114 h-1) > diethyl ether (0.0043 h-1) > benzene (0.0018 h-1) > toluene (0.0008 h-1). Table 13.3.25. Second-order rate constants and activation parameters for the reaction of benzyl chloride with triphenylphosphine in various solvents k×104, Lmol-1s-1
Dielectric constant (20oC)
60oC
Decalin
2.26
0.00134
Toluene
2.38
0.0169
0.0843
0.181
0.353
Anisole
4.33
0.0569
0.260
0.466
1.15
Bromobenzene
5.40
0.0933
0.276
0.457
0.909
Chlorobenzene
5.62
0.0512
0.243
1.13
Benzyl alcohol
13.1
8.94
38.7
107
1-Butanol
17.1
4.93
19.5
43.7
Acetopnenone
17.39
0.300
1.32
9.19
1-Propanol
20.1
4.41
9.86
Acetone
20.3
0.166
0.417
Ethanol
24.3
3.60
7.73
Benzonitrile
25.2
0.545
1.71
6.38
Nitroethane
28.06
0.470
3.05
12.9
Methanol
32.65
8.86
N,N-dimethylformanide
36.7
0.460
1.88
6.59
Acetonitrile
37.5
1.20
Solvent
70oC
80oC
90oC
100oC 0.0355
2.79
830
Maw-Ling Wang
k×104, Lmol-1s-1
Solvent
Dielectric constant (20oC)
60oC
N,N-Dimethylacetamide
37.8
0.29
1.06
4.29
N-Methylformamide
189.5
6.08
28.7
86.5
70oC
80oC
90oC
100oC
Data obtained from the work of Maccarone et al.64
The BC-TP reaction shows better reactivity in protic or polar solvent since the activated complex is more polar than both reactant molecules. 13.3.2 THREE-PHASE REACTIONS (TRIPHASE CATALYSIS) As stated, the solid PTC is suitable for the industrial processes concerning the removal of the catalyst from the reaction mixture and its economic recycle. The real mechanism of reaction in a triphase catalysis is not completely understood. However, the reaction rate and the conversion of reactant in a triphase catalysis (TC) is highly dependent on the organiphilicity (hydrophilicity or hydroprobicity) of the polymer support of the catalyst and the polarity of the organic solvent. Not only the partition of the organic to the aqueous solutions is affected by the organophilicity of the polymer-supported catalyst, but also the concentration distribution of the catalyst between two phases is influenced by the organophilicity of the polymer-supported catalyst. Ohtani et al.82-86 used polystyrene-supported ammonium fluoride as a phase transfer catalyst (triphase catalysis) for several base-catalyzed reactions, such as cyanoethylation, Knoevenage reaction, Claisen condensation and Michael addition. The catalytic activity of the polystyrene-supported ammonium fluid was comparable to that of tetrabutylammonium fluoride (TBAF). The ionic loading and the ammonium structure of the fluoride polymers hardly affected the catalytic efficiency. The reaction was fast in a non-polar solvent (e.g., octane or toluene) from which the rate-determining step of the base-catalyzed reaction is very similar to that of the SN2 nucleophilic substitution reactions. The solvent may affect the catalytic activity in several ways. The greater its swelling power, the larger the volume fraction of catalytic occupied by the more mobile liquid, and the swollen volume fraction of the more rigid polymer network. The degree of swelling and the viscosity within the polymer matrix affect intraparticle diffusion rates. The solvent may also affect intrinsic reactivity at the active sites. Experimentally, it is difficult to distinguish solvent effect on diffusivity from solvent effects on reactivity. Tomoi and Ford116 found that the triphase catalysis followed pseudo-first-order rate law. The corresponding apparent rate constant kapp decreases with solvent in the order: chlorobenzene > toluene > decane over wide ranges of particle sizes and polymer crosslinking. The ability of the solvent to swell the catalysts decreases in the same order. 13.3.2.1 The interaction between solid polymer (hydrophilicity) and the organic solvents In triphase catalysis, solvated resin supports are important carriers for solid-phase organic synthesis in combinatorial chemistry. The physical properties of resin, resin swelling, dynamic solvation, and solvated supports are important factors in affecting the synthesis.160 However, these factors are also affected by solvent. Selective solvation of resin alters the local reactivity and accessibility of the bound substrate and the mobility of the entrapped re-
13.3 Effects of organic solvents on phase-transfer catalysis
831
agent. Resin solvation changes during the course of the reaction when the attached substrate changes its polarity or other physicochemical properties. The basic steps involved in reactions with resin-supported PTC catalysts differ from ordinary two-phase PTC reactions in one important respect: ordinary PTC reactions require only one reagent to be transferred from their normal phase to the phase of the second reactant. Use of resin-supported catalysts requires that both reagents diffuse to active PTC sites on the catalyst surface, or for reactions with slow intrinsic rates, both reagents must also diffuse to the active sites inside the resin bulk phase. The need for diffusion processes with solid catalysts also means that both reagents are required to diffuse to and penetrate the stagnant outer layer of liquid(s) (the Nernst layer), coating the catalyst particle. Ford and Tomoi28 carried out the reaction of 1-bromooctane with aqueous sodium cyanide, catalyzed by tributylphosphonium groups bound into beads of an insoluble styrene-divinylbenzene resin, C 8 H 5 PBu 3 Br 1− C8H17Br( org ) + NaCN ( aq ) polymer − → 1− C8H17CH( org ) + NaBr( aq ) [13.3.16]
The reaction includes the following steps: (a) Diffusion of aqueous sodium cyanide through the bulk phase and through the resin bulk to active sites (b) Equilibrium exchange of CN- for Br- at the active sites (c) Diffusion of Br- out of the catalyst particle and into the aqueous bulk phase (d) Diffusion of RBr (1-C8H17Br(org)) through the organic bulk phase and through the bulk resin phase to active sites. Some reactions may occur at sites on the catalyst surface, but since the number of surface sites is small compared to the number of sites within the bulk of the catalyst, most of the reaction occurs inside the catalyst bulk. (e) Chemical reaction (intrinsic reaction) between RBr and Resin-PR3+CN- at active sites to produce RCN and Br(f) Diffusion of RCN out of the catalyst particle and into the organic phase. A schematic diagram of the general resin-bound PTC catalyst is given in Figure 13.3.10. As indicated in Figure 13.3.10, spacer chains can increase some reactions by removing the active site away from the polymer chain, and from other active sites. When active sites, particularly quaternary onium salts, are located close to one anFigure 13.3.10. Schematic diagram of the general other, they join to form doublets, triplets, and higher aggregates that are less active resin-bound PTC catalysts. catalyst centers, and that tend to present an “aqueous” face to the reactants. Thus, the use of spacer chains increases the rates of some reactions, such as nucleophilic displacements, two-to-four-fold.1,9,73 When the spacer chain also contains complexable ether oxygen atoms, using 15-crown-5 ether as the PTC functional group, catalyst activity is even greater, as observed in halide exchange of KI with 1-bromooctane.10 Preparation of phase transfer catalyst (PTC) functional groups bound to insoluble resins and their activity for catalyzing two-phase reactions has been extensively stud-
832
Maw-Ling Wang
ied.82,83,97,101 Much work has been done on the preparation and testing of phase-transfer catalysts supported on resins including extensive work by Montanari and co-workers71,74 and by Ford and Tomoi28 and their co-workers. Most published works on resin-bound phase-transfer cataFigure 13.3.11. Crosslink to other polymer chains. lysts use a styrene-divinylbenzene resin (SDV) and related resins, taking advantage of the huge amount of technology available on these resins due to their use as ion-exchange resin supports (Figure 13.3.11). Tomoi and co-workers117 suggest that solvents may affect rates of triphase-catalyzed reaction in three ways: intrinsic chemical reactivity; solvent effect on ion-exchange rate; and overall activity, including diffusion effects due to swelling of polymer-supported phosphonium salts under three-phase conditions. First, the intrinsic activity of the catalysts, as well as of soluble phosphonium salts, depended slightly on organic solvents for cyanide displacement reactions. Second, the exchange rate of chloride ion in the catalysts vs. that of acetate depends on the solvents when the degree of ring substitution is less than 16%. With 30% ring-substituted catalysts, the rate increases and hardly depends on the solvents. Third, the overall catalyst reactivity for the reaction of organic halides with NaCN depends on the substrate and organic solvents. For 1-bromooctane, the catalysts were more reactive in good solvents (e.g., chlorobenzene) than in poor solvent (e.g., octane). Shan and co-workers,98 Wang and Wu136,137 examined the effects of solvents and other resin-bound catalysts parameters (macroporosity, microporosity, crosslink density and size of catalyst pellet). They all show that the swelling in organic solvents is an important factor affecting the conversion of the reactant, as shown in Table 13.3.26.98 Table 13.3.26. Effect of the organic solvent on the yield of ester from benzyl bromide and aqueous KOAc under standardized conditions Catalysta
Solvent (dielectric constant, ε)b
Yield of ester, %
Macro (6%)-400
Cyclohexane (2.02) Toluene (2.38) Chlorobenzene (5.62) Benzyl ethyl ketone (17.4) Nitrobenzene
13.2 16.5 23.3 94.5 100.0
Micro (6%)-400
Cyclohexane (2.02) Toluene (2.38) Chlorobenzene (5.62) Benzyl ethyl ketone (17.4) Nitrobenzene
9.2 13.5 18.2 86.0 100.0
Data obtained from Shan, Kang and Li;98 a6% crosslinking; bPEG-400 used for active sites on catalyst
13.3 Effects of organic solvents on phase-transfer catalysis
833
13.3.2.2 Effect of solvents on the reaction in triphase catalysis The disadvantage of using liquid-liquid phase-transfer catalysis (ll-PTC) is in the separation of catalyst from product after reaction. This problem can be overcome using the immobilized catalyst on a solid support (e.g., porous polymer pellet). Simple mechanical separation processes, such as filtration or centrifugation, can be employed to separate the solid catalyst from the product in liquid form. A detailed investigation of the effect of polymer particle on the reaction rate was conducted by Wang and coworkers.131,133-137 Wang and Wu136 studied the reaction of substitution of hexachlorocyclotriphosphazene and 2,2,2-trifluoroethanol in an organic solvent/alkaline solution by triphase catalysis. The polymer, which was prepared from the polymerization of styrene monomer and chloromethylstyrene monomer, served as the support for the immobilization of the catalyst. The effects of the structure of the polymer support, which can be related to the factors of the degree of crosslinking, ring substitution (RS), lipophilicity of the polymer, the chloride density and solvents on the imbibed compositions84,118 on the reaction rate or conversion were investigated. This imbibed composition, influenced by the internal structures of the triphase catalyst particles, affected the reactivities. The interaction of the polymer support pellet and the organic solvents play an important role in determining the reaction rate and the conversion of the reactant. The reaction could be improved to obtain a high reaction rate by using a polar solvent. For investigating the degree of crosslinking of the polymer, the resistance of mass transfer within the catalyst pellet is small. When a smaller degree of crosslinking of the polymer support is used. This is due to the fact that a larger value of the swell of the polymer was obtained when a small degree of crosslinking of the polymer was used. Wang and Yu131 have similar observations for the reaction of allyl bromide and 2,4-dibromophenol under triphase catalysis. A maximum value exists for the degree of swell and the imbibed composition, as shown in Table 13.3.28 for the degree of crosslinking.136 In Table 13.3.27,136 the degree of swell for the polymer support with a 6% crosslinking is larger than that for the two other degrees of crosslinking. This implies that greater amounts of NaOCH2CF3 were imbibed into the catalyst pellet with a 6% crosslinking. The reaction rate is directly related to the amount of the imbibed composition. Also, in Table 13.3.27, the imbibed compositions are affected by the structure of the polymer support. The reactivity of the triphase catalysis can also be determined from the composition imbibed by the particles. It can be observed that the reactivities were highly affected by the lipophilicity of the catalyst pellet for the substitution reaction in the organic phase and the hydrophilicity of the catalyst pellet of the ion exchange in the aqueous phase. For example, the reaction rate in the organic phase was promoted by using a lipophilic polymer support catalyst when the substitution reaction rate was slow. In Table 13.3.27, the amount of chlorobenzene and water imbibed in the macroporous pellet was greater than that in the microporous pellets for most cases.136 However, the macroporous pellet with 10% degree of crosslinking had the least lipophilicity and degree of swelling; therefore, the reactivity was the lowest for the macroporous pellet with a 10% crosslinking among the polymer-support catalysts. The reactivity environments which were created by the lipophilicity and the hydrophilicity of the polymer support plays an important role in determining the reactivity. It is known that the distribution of organic phase and aqueous phase existing in the porous pellet is affected by a change of the ring substitution (RS) of the polymer support.84 Wang and Wu136 prepared three kinds of polymer supports with different numbers of ring
834
Maw-Ling Wang
substitution, such as 10%RS, 20%RS and 49%RS, to analyze the lipophilicity of the polymer support. The order of the lipophilicity was 10%RS > 20%RS> 49%RS, which is the same as the order of swelling. However, a maximum value of the apparent rate constant was obtained for using a 20%RS pellet catalyst among the three kinds of ring substitution polymer pellet. Therefore, it is concluded that the lipophilicity of the polymer cannot be too large to enhance the reaction rate. This is due to the fact that the ion exchange rate is retarded to lower the reaction rate because of using a high lipophilic polymer support. It is concluded that the lipophilicity and the hydrophilicity highly influence the reactivity in triphase catalysis. For a two-phase PTC, it is recognized that the polarity of the organic solvent affects the reaction rate. In general, the reaction rate increases with the augmentation of the polarity of the solvents. Table 13.3.28 shows the effects of the organic solvents on the apparent rate constant, ko,app and ka,app.136 A higher value of the apparent rate constant was obtained using solvent of high polarity. This result is consistent with the swelling and the imbibed compositions that are given in Table 13.3.29.136 Table 13.3.27. Compositions of the imbibed solvents and swelling volume of the triphase catalyst pellet with various polymer structures Triphase catalyst
microporous 2%
microporous 6%
microporous 10%
macroporous 2%
macroporous 6%
macroporous 10%
Conditions
ClC6H5 g
H2O g
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
1.31 1.23 1.92
0.33 0.67
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
1.19 1.17 1.90
0.62 0.50
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
1.06 0.96 1.40
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
NaOCH2CF3, g (calcd value, g)
Swelling volume ratio
0.40 (0.29)
2.4 2.7 3.6
0.60 (0.22)
2.2 2.8 3.4
0.29 0.50
0.19 (0.22)
2.0 2.1 2.9
1.28 1.33 2.29
0.73 0.50
0.34 (0.22)
3.1 3.8
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
1.54 1.25 2.2
0.82 0.59
0.42 (0.25)
2.5 3.1 3.8
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
0.76 1.05 1.28
0.63 0.38
0.17 (0.16)
2.7 2.6
Data obtained from Wang and Wu;136 50 mL of chlorobenzene, 20 mL of water, 0.059 mole of (NPCl2)3, 0.7 meq of catalyst was used, 20oC
13.3 Effects of organic solvents on phase-transfer catalysis
835
Table 13.3.28. Effects of the concentrations of NaOCH2CF3 and kind of solvent on the apparent intrinsic rate constants, ko,app and ka,app
Solvent
ko,app for [NaOCH2CF3] (M), (min.meq)
-1
ka,app for [NaOCH2CF3] (M), (min.meq)
1.6 M
2.2 M
2.8 M
1.6 M
2.2 M
2.8 M
CH2Cl2
0.25
0.33
0.58
0.036
0.028
0.036
C6H5Cl
0.063
0.12
0.19
0.017
0.015
0.014
C6H5CH3
0.027
0.056
0.15
0.0055
0.006
0.011
n-C6H14
0.015
0.031
0.092
0.0008
0.0008
0.018
-1
Data obtained from Wang and Wu;136 50 mL of solvent, 20 mL of water, 0.0059 mol of (NPCl2)3, 0.18 meq of macroporous catalyst, 20oC, 40-80 mesh of particle
Table 13.3.29. Effects of solvents on the composition of the imbibed solvents and swelling volume of the triphase catalyst pellet Solvent
Conditions
Solvent g
H2O g
NaOCH2CF3, g (calcd value, g)
Volume ratio
CH2Cl2
CH2Cl2 H2O/CH2Cl2
2.75 2.24
0.96
3.2 3.8
ClC6H5
ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5
1.28 1.33 2.29
0.73 0.50
0.34 (0.18)
2.2 3.1 3.8
CH3C6H5 H2O/CH3C6H5 2.8M NaOCH2CF3/CH3C6H5
0.57 0.61 0.37
0.29 0.30
0.01 (0.10)
1.7 2.1 1.8
n-C6H5 H2O/n-C6H5 2.8M NaOCH2CF3/n-C6H5
0 0.10 0.10
0.30 0.15
0.01 (0.05)
1 1.4 1.24
CH3C6H5
n-C6H14
Data obtained from Wang and Wu;136 30 mL of solvent, 0.80 meq of catalyst (1 g), 40-80 mesh of macroporous particle, 20oC
The overall kinetics can be divided into two steps by virtue of the presence of the two practically immiscible liquid phases, i.e., (1) a chemical conversion step in which the active catalyst sites (resin with 2,2,2-trifluoroethanoxide ions) react with hexachlorocyclotriphosphazene in the organic solvent, i.e., yResin+-OCH2CF3(s) + (NPCl2)3(org) → yResin+Cl-(s) + N3P3Cl6-y(OCH2CF3)y(org), y=1-6
[13.3.17]
836
Maw-Ling Wang
(2) the ion exchange step in which the attached catalyst sites are in contact with the aqueous phase, i.e., Resin+Cl-(s) + NaOCH2CF3(aq) → Resin+OCH2CF -3(s) + Na+Cl -(aq)
[13.3.18]
The total moles of the catalyst active sites are S; thus S = [Resin+-OCH2CF3(s) + Resin+Cl -(s)]
[13.3.19]
The reaction rates for (NPCl2)3 in the organic phase and for NaOCH2CF3 in the aqueous phase follow pseudo-first-order kinetics and can be written as − −
d [(NPCl 2 ) 3 ] 0 dt
= k 0, app S[(NPCl 2 ) 3 ] ( org )
d [NaOCH 2 CF3 ]a dt
= k a, app S[NaOCH 2 CF3 ]( org )
[13.3.20]
[13.3.21]
where k0,app and ka,app are the apparent rate constants of (NPCl2)3 per unit amount of catalyst (molar equivalent) in the organic phase for triphase catalysis and the apparent rate constant of NaOCH2CF3 per unit amount of catalyst (molar equivalent) in the aqueous phase for triphase catalysis, respectively. Wang and Yang134 carried out the reaction of 2,4,6-tribromophenol and allyl bromide catalyzed with tributylamine immobilized on the solid styrene-chloromethylstyrene polymer support in an alkaline solution of KOH/chlorobenzene. The experimental results indicate that the swelling power is enhanced in an organic solvent of high polarity. Thus, the reactivity of the reaction is increased with the increase in the polarity of the organic solvents. 13.3.2.3 Effect of volume of organic solvent and water on the reactions in triphase catalysis In investigating the effect of the amount of water, the contents of other components are fixed. Changing the amount of water affects the volume ratio of organic phase to aqueous phase and the concentration of nucleophile in the aqueous phase. For the reaction of hexacyclotriphosphazene and sodium 2,2,2-trifluoroethanoxide catalyzed by tributylamine immobilized on the solid styrene-chloromethylstyrene polymer support catalyst.138-140,157 As shown in Figure 13.3.12,140 the reaction rate is decreased with the increase in the amount of water up to a concentration of sodium 2,2,2-trifluoroethanoxide at 2.8 M. However, the reaction rate is then increased with further increase in the amount of water larger than 2.8 M sodium 2,2,2-trifluoroethanoxide. This result indicates that a high concentration of sodium 2,2,2-trifluoroethanoxide reaction will decrease the reaction rate. The main reason is that the intraparticle diffusion is also affected by the concentration of sodium 2,2,2-trifluoroethanoxide due to changing the amount of water (or the volume ratio of organic phase to aqueous phase).
13.3 Effects of organic solvents on phase-transfer catalysis
837
Figure 13.3.12. The apparent rate constants vs. the molar ratios of NaOH/HOCH2CF3; 0.059 mol of (NPCl2)3, 9.6×10-5 mol of TBAB catalyst, 50 mL of chlorobenzene, 20 mL of H2O, 20oC; and (∆) 0.07 mol of HOCH2CF3, (*) 0.058 mol of HOCH2CF3, (O) 0.075 mol of NaOH, (c) 0.063 mol of NaOH (Adapted from Ref. [138], by permission.)
The other experiments, in which the concentrations of the components in the aqueous phase are fixed, were carried out by Wu157. The volume of organic phase is also fixed at 50 mL, in which the amount of catalyst and phosphazene are also fixed. A generalized apparent rate constant (pseudo-first-order rate law) k'i,app is defined as
[
]
k ′i, app = k i, app / Resin (s ) / Va ; i = r, f
[13.3.21]
where Resin(s) and Va indicate the total molar equivalent active sites and the volume of water. The results are given in Table 13.3.30.157 The apparent rate constants k'r,app and k'f,app are increased with the increase in the volume of aqueous phase. These results are explained by low concentration of sodium 2,2,2-trifluoroethanoxide in the organic phase because HOCH2CF3 and NaOCH2CF3 are all insoluble in chlorobenzene. After 4 hours of reaction, only 8% of hexacyclotriphosphasene is reacted. The increase in reaction rate by increasing the volume of aqueous solution is not due to the increase in the concentration of NaOCH2CF3 in the organic phase. However, the mass transfer rate within the particles is obviously affected by increasing the concentration of NaOCH2CF3 in the aqueous phase. Table 13.3.30. Apparent rate constants in various NaOCH2CF3 concentration under constant amount of water or constant amount of NaOCH2CF3 NaOCH2 CF3 (M)
Variation of NaOCH2CF3 NaOCH2CF3 (mole)
kr,appa
Variation of water k'f,app
kf,appa
b
1.3
Volume of water (mL)
kr,appa
kf,appa
k'f,appb
50
0.033
0.0031
8.61
1.6
0.035
0.025
0.005
5.56
40
0.038
0.0033
7.33
2.2
0.0525
0.041
0.005
5.56
30
0.044
0.0041
6.83
2.8
0.07
0.055
20
0.055
0.0050
5.56
a
0.005 -1 b
5.56 -1
-4
Data obtained from Wang and Wu;157 min , Lmin meq×10 ; Reaction conditions: 50 mL of chlorobenzene, 0.0059 moles of (NPCl2)3, 0.175 meq of catalyst, 20oC
838
Maw-Ling Wang
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
A. Akelah and D. C. Sherrington, Eur. Polym. J., 18, 301 (1982). B. C. Berris, U. S. Patent 5,030,757 (1991). G. E. Boyd and Q. V. Larson, J. Am. Chem. Soc., 89, 6038 (1967). A. Brandstrom, Adv. Phys. Org. Chem., 15, 267 (1977). L. M. Chen, Y. L. Chen and T. C. Chou, Denki Kagaku, 62(12), 1173 (1994). Y. L. Chen and T. C. Chou, Ind. Eng. Chem. Res., 33(3), 676 (1994). Y. L. Chen and T. C. Chou, J. Appl. Electrochem., 24(5), 434 (1994). T. Chiba and M. Okimoto, J. Org. Chem., 56(21), 6163 (1991). M. S. Chiles, D. D. Jackson and P. C. Reeves, J. Org. Chem., 187, 357 (1980). K. B. Chung and M. Tomoi, J. Polym. Sci., Part A: Polym. Chem., 30(6), 1089 (1992). E. Cordoncillo, E., Monros, G., M. A. Tena, P. Escribano and J. Corda, J. Non-Crystalline Solids, 171(2), 105 (1994). B. P. Czech, M. J. Pugia and R. A. Bartsch, Tetrahedron, 41, 5439 (1985). W. H. Daly, J. D. Caldwell, V. P. Kien and R. Tang, Polym. Prepr., Am. Chem. Soc., Div. Polym. Chem., 23, 145 (1982). O. I. Danilova and S. S. Yufit, Mendeleev Commun., 4, 165 (1993). E. V. Dehmlow and J. Schmidt, Tetrahedron Letters, 2, 95 (1976). E. V. Dehmlow and S. S. Dehmlow, Phase Transfer Catalysis, Verlag Chemie GmbH, Weinheim (1980). E. V. Dehmlow and S. S. Dehmlow, Phase Transfer Catalysis, Weinheim, Deerfield Beach, Florida; Basel, Verlag Chemie, (1983). E. V. Dehmlow and T. Remmler, J. Chem. Res.(S), 72 (1977). E. V. Dehmlow, M. Slopianka and J. Heider, Tetrahedron Letters, 27, 2361 (1977). E. V. Dehmlow, R. Tiesen, Y. Sasson and E. Pross, Tetrahedron, 41(14), 2927 (1985). E. V. Dehmlow, Tetrahedron Letters, 9, 1 (1976). A. K. Dillow, S. L.J. Yun, D. Suleiman, D. L. Boatright, C. L. Liotta, and C. A. Eckert, Ind. Eng. Chem. Res., 35(6), 1801 (1996). J. S. Do and T. C. Chou, J. Appl. Electrochem., 19, 922 (1989). M. Ellwood and J. Griffiths, J. Chem. Soc. Chem. Comm., 181 (1980). T. L. Evans, U. S. Patent 4,520,204 (1985). W. K. Fife and X. Yue, Bull. Chem. Soc. Jpn., 109, 1278 (1987). O. E. Filippova, I. N. Topchieva, V. V. Lutsenko and V. P. Zubov, Vysolomol Soedin., Ser. A, 26, 402 (1984). W. T. Ford and M. Tomoi, Adv. Polym. Sci., 55, 49 (1984). H. H. Freedman, Pure & Appl. Chem., 58(6), 857 (1986). A. Gobbi, D. Landini, A. Maia and S. Petricci, J. Org. Chem., 63 (16), 5356 (1998). G. W. Gokel, ACS Ser., 326, 24 (1987). G. W. Gokel, D. M. Goli and R. A. Schultz, J. Org. Chem., 48, 2837 (1983). H. P. Gregor and J. I. Bergaman, J. Colloid. Sci., 6, 323 (1951). J. M. Harris and M. G. Case, J. Org. Chem., 48, 5390 (1983). J. M. Harris, N. H. Hundley, T. G. Shannon and E. C. Struck, J. Org. Chem., 47, 4789 (1982). A. W. Herriott, and D. Picker, J. Am. Chem. Soc., 97(9), 2345 (1975). T. C. Huang and S. C. Lin, J. Chin. Inst.Chem. Eng., 19, 193 (1988). T. Ido, M. Saiki and S. Goto, Kagaku, Kagaku Ronbunshu, 14, 539 (1988). T. Ido, M. Saiki and S. Goto, Kagaku, Kagaku Ronbunshu, 15, 403 (1989). T. Ido, Y. Kitamura and S. Goto, Kagaku, Kogaku Ronbunshu, 16, 388 (1990). Y. Imai, J. Macromol. Sci., Chem., A15(5), 833 (1981). H. Iwamoto, Tetrahedron Letters, 24, 4703 (1983). H. Iwamoto, H. Kobayashi, P. Murer, T. Sonoda and H. Zollinger, Bull. Chem. Soc. Jpn., 66(9), 2590 (1993). H. Iwamoto, M. Yoshimara and T. Sonodo, Bull. Chem. Soc. Jpn., 56, 791 (1983). D. Jaeger, M. D. Ward and A. K. Dutta, J. Org. Chem., 53, 1577 (1988). Y. S. Jane and J. S. Shih, J. Mol. Catal., 89(1-2), 29 (1994). J. Jarrouse, C. R. Hebd. Seances Acad. Sci., Ser. C232, 1424 (1951). G. Jin, T. Ido and S. Goto, J. Chem. Eng. Jpn., 31(5), 741 (1998). T. Kawai, British Patent 2,219,292 (1989). H. Kobayashi, and T. Sonoda, Chem. Letters, 1185 (1982). Kohjin Co., Ltd., Japanese Kokai Tokkyo Koho JP 58/147,402 (1983). T. N. Konstantinova and V. B. Bojinov, Dyes and Pigments, 39(2), 69 (1998).
13.3 Effects of organic solvents on phase-transfer catalysis
839
J. Kuji and T. Okano, J. Syn. Org. Chem. Jpn., 52(4), 276 (1994). C. S. Kuo and J. J. Jwo, J. Org. Chem., 57, 1991 (1992). D. Landini, A. Maia and F. Montanari, J. Am. Chem. Soc., 100, 2796 (1978). M. D. Lang, G. Zhang, X. F. Chen, N. Feng and S. H. Li, J. Appl. Sci., 70(8), 1427 (1998). H. Ledon, Synthesis, 347 (1974). D. G. Lee and V. S. Chang, J. Org. Chem., 43(8), 1532 (1978). D. J. Liaw and P. Chang, J. Appl. Polym. Sci., 63(2), 195 (1997). C. L. Liotta, Application of Macrocyclic Polydentate Ligands to Synthetic Transformations, in Synthetic Multidentate Compounds, R. M. Izatt, J. J. Christensen, eds., Academic Press, New York, 111 (1978). 61 C. L. Liotta, E. M. Burgess, C. C. Ray, E. D. Black and B. E. Fair, ACS Symp. Ser., 326, 15 (1987). 62 C. L. Liotta,, E. M. Burgess and E. D. Black, Polym. Prepr., Am. Chem. Soc., Div. Polym. Chem., 31, 65 (1990). 63 B. L. Liu, A Study of Catalyzed Reaction of Synthesizing of 2-Mercaptobenzimidazole and Its Derivatives", Ph.D. Thesis, Dept. of Chem. Eng., Nat’l Tsing Hua Univ., Hsinchu, Taiwan (1995). 64 E. Maccarone, G. Perrini and M. Torre, Gazzetta Chimica Italiana, 112 (1982). 65 M. Makosza and E. Bialecka, Tetrahedron Letters, 2, 1983 (1977). 66 M. Makosza, Pure Appl. Chem., 43, 439 (1975). 67 M. Makosza, Russ. Chem. Rev., 46, 1151 (1977). 68 C. A. Marques, M. Selva and P. Tundo, J. Chem. Soc., Perkin Trans. I, 529 (1993). 69 K. Mathias and R. A. Vaidya, Bull. Chem. Soc. Jpn., 108, 1093 (1986). 70 E. Mieczynski, A. M. Trzeciak, R. Grzybek and J. J. Ziolkowski, J. Mol. Catal., A: Chemical, 132 (2-3), 203 (1998). 71 H. Molinari, F. Montanari, S. Quici and P. Tundo, J. Am. Chem. Soc., 101, 3920 (1979). 72 E. Monflier, S. Tilloy, G. Fremy, Y. Castanet and A. Mortreux, Tetrahedron Letters, 36(52), 9481 (1995). 73 F. Montanari, S. Quici and P. Tundo, J. Org. Chem., 48, 199 (1983). 74 F. Montanari, D. Landini, A. Maia, S. Quici and P. L. Anelli, ACS Symp. Ser., 326, 54 (1987). 75 P. W. Morgan, Condensation Polymers by Interfacial and Solution Methods, p. 349, Interscience, New York, (1965). 76 R. T. Morrison and R. N. Boyd, Organic Chemistry, p. 702, 5th ed., Allynand Bacon, Inc., Mass, USA (1987). 77 K. Nakamura, S., Nishiyama, S. Tsuruya and M. Masai, J. Mol. Catal., 93(2), 195 (1994). 78 H. Namba, N. Takahashi, K. Abe and M. Saito, Japanese Patent, Jpn. Kokai Tokkyo Koho 63/196547 (1988). 79 R. Neumann and Y. Sasson, J. Org. Chem., 49, 3448 (1984). 80 R. Neumann, S. Dermeik and Y. Sasson, Israel J. Chem., 26, 239 (1985). 81 T. Nishikubo and T. Iziawa, J. Syn. Org. Chem. Jpn., 51(2), 157 (1993). 82 N. Ohtani, Yuki Gosei Kagaku Kyokaishi, 43, 313 (1985). 83 N. Ohtani, Kagaku Kogyo, 39, 331 (1988). 84 N. Ohtani, C. A. Wilkie, A. Nigam and S. C. Regen, Macromolecules, 14, 516 (1981). 85 N. Ohtani, Y. Inoue, A. Nomoto and S. Ohta, React. Polym., 24(1), 73 (1994). 86 N. Ohtani, Y. Inoue, H. Mizuoka and K. Itoh, J. Polymer. Sci., Part A- Polym. Chem., 32(13), 2589 (1994). 87 J. Pielichowski and P. Czub, Angewandte Makromolekulare Chemie, 251, 1 (1997). 88 S. L. Regen, J. Org. Chem., 42, 875 (1977). 89 S. L. Regen, J. Am. Chem. Soc., 97, 5956 (1975). 90 S. L. Regen, J. C. K. Heh and J. McLick, J. Org. Chem., 44, 1961(1979). 91 S. L. Regen, J. J. Bese and J. McLick, J. Am. Chem. Soc., 101, 116 (1979). 92 S. L. Regen and J. J. Besse, J. Am. Chem. Soc., 101, 4059 (1979). 93 G. Rothenberg, L. Feldberg, H. Wiener and Y. Sasson, J. Chem. Soc., Perkin Trans. 2, 11, 2429 (1998). 94 Y. Sasson and R. Neumann, Handbook of Phase Transfer Catalysis, Blackie Academic & Professional, Chapman & Hall, New York (1997). 95 Y. Sasson and S. Zbaida, Ger. Patent DE 3,307,164 (1983). 96 E. G. See, C. A. Buehler and D. E. Pearson, Survey of Organic Synthesis, p. 285, Wiley Interscience, New York (1970). 97 F. Sevc, J. Kahovec and J. Hradil, Chem. Listy, 81, 183 (1987). 98 Y. Shan, R. Kang and W. Li, Ind. Eng. Chem. Res., 28, 1289 (1989). 99 S. S. Shavanon, G. A. Tolstikov and G. A. Viktorov, Zh. Obsch. Khim., 59, 1615 (1989). 100 S. S. Shavanov, G. A. Tolstikov, T. V. Shutenkova, Zh. Obshch. Khim., 57, 1587 (1987). 53 54 55 56 57 58 59 60
840
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154
Maw-Ling Wang D. C. Sherrington, K. J. Kelly, Org. Coat. Appl. Polym. Sci., Proc., 46, 278 (1981). S. Shimizu, Y. Sasaki and C. Hirai, Bull. Chem. Soc. Jpn., 63, 176 (1990). K. Shinoda, and K. Yasuda, J. Org. Chem., 56(12), 4081 (1991). K. Sjoberg, Aldrichimica Acta, 13, 55 (1980). M. B. Smith, Synth. Commun., 16, 85 (1986). M. Stark, Y. Q. Wang, O. Danielsson, H. Jornvall and J. Johansson, Anal. Biochem., 265(1), 97 (1998). C. M. Starks and C. L. Liotta, Phase Transfer Catalysis, Principles and Techniques, Academic Press, New York (1978). C. M. Starks, J. Am. Chem. Soc., 93, 195 (1971). C. M. Starks, ACS Symposium Series, No. 326, 1 (1987). C. M. Starks, C. L. Liotta and M. Halpern, Phase Transfer Catalysis, Fundamentals, Applications and Indutrial Perspectives, Chapman and Hall, New York (1994). C. M. Starks and R. M. Owens, J.Am. Chem. Soc., 93, 3615 (1973). G. T. Szabo, K. Aranyosi and L. Toke, Acta Chim. Acad. Sci., Hung., 110, 215 (1982). J. Y. Tarng and J. S. Shih, J. Chin. Chem. Soc., 41(1), 81 (1994). L. Toke and G. T. Sazbo, Acta chem. Acad. Sci., Hung., 93, 421 (1977), as summarized by G. E. Totten and N. A. Clinton, Rev. Macromol. Chem. Phys., C28, 293 (1988). L. Toke, G. T. Szabo and K. Somogyi-Werner, Acta Chim. Acad. Sci., Hung., 101, 47 (1979). M. Tomoi and W. T. Ford, J. Am. Chem. Soc., 103, 3821 (1981). M. Tomoi, E. Nakamura, Y. Hosokawa and H. Kakuuichi, J. Polym. Sci., Polym. Chem. Ed., 23, 49 (1985). M. Tomoi, E. Ogawa, Y. Hosokama and H. Kakiuchi, J. Polym. Chem., Polym. Chem., 20, 3421 (1982). A. W. Trochimczuk, A. W. and B. N. Kolarz, Eur. Polym. J., 28(12), 1593 (1992). M. L. Tsai and T. C. Chou, J. Chin. Inst. Chem. Eng., 27(6), 411 (1996). J. Uglestad, T. Ellingsen and A. Beige, Acta Chem. Scand., 20, 1593 (1966). R. Vladea and T. Simandan, Rev. Chim. (Bucharest), 41, 421 (1990). D. H. Wang and H. S. Weng, J. Chin. Inst. Chem. Eng., 27(3), 129 (1996). D. H. Wang and H. S. Weng, J. Chin. Inst. Chem. Eng., 26(3), 147 (1995). D. H. Wang and H. S. Weng, Chem. Eng. Sci., 50(21), 3477 (1995). D. H. Wang and H. S. Weng, Chem. Eng. Sci., 43, 2019 (1988). M. L. Wang and A. H. Liu and J. J. Jwo, Ind. Eng. Chem. Res., 27, 555 (1988). M. L. Wang and B. L. Liu, J.Mol. Catal. A: Chemical, 105, 49 (1996). M. L. Wang, M. L. and B. L. Liu, Int. J. Chem. Kinetics, 28, 885 (1996). M. L. Wang and B. L. Liu, Ind. Eng. Chem. Res., 34(11), 3688 (1995). M. L. Wang, M and C. C. Yu, J. Polymer Sc., Part A: Polym. Chem., 31, 1171 (1993). M. L. Wang and C. C. Yu, J. Chin. Inst. Eng., 23(3), 153 (1992). M. L. Wang, M. L. and C. Z. Peng, J. Appl. Polym. Sci., 52, 701 (1994). M. L. Wang and H. M. Yang, Ind. Eng. Chem. Res., 31(8), 1868 (1992). M. L. Wang and H. M. Yang, J. Mol. Catal., 62, 135 (1990). M. L. Wang and H. S. Wu, Ind. Eng. Chem. Res., 31, 490 (1992). M. L. Wang and H. S. Wu, Ind. Eng. Chem. Res., 31(9), 2238 (1992). M. L. Wang and H. S. Wu, Chem. Eng. Sci., 46, 509 (1991). M. L. Wang and H. S. Wu, J. Chem. Soc., Perkin Trans., 2, 841 (1991). M. L. Wang and H. S. Wu, J. Org. Chem., 55, 2344 (1990). M. L. Wang and K. R. Chang, Ind. Eng. Chem. Res., 29, 40 (1990). M. L. Wang and K. R. Chang, J. Chin. Inst. Eng., 16(5), 675 (1993). M. L. Wang and K. R. Chang, J. Mol. Catal., 67, 147 (1991). M. L. Wang, M. L. and S. W. Chang, Chem. Eng. Commum., 146, 85 (1996). M. L. Wang and S. W. Chang, Bull. Chem. Soc. Jpn., 66, 2149 (1993). M. L. Wang, M. L. and S. W. Chang, Ind. Eng. Chem. Res., 34, 3696 (1995). M. L. Wang and Y. M. Hsieh, Develop. In Chem. Eng. and Min. Proc., 1(4), 225 (1993). M. L. Wang, M. L., C. C. Ou and J. J. Jwo, Ind. Eng. Chem. Res., 33, 2034 (1994). M. L. Wang, M. L., C. C. Ou and J. J. Jwo, J. Mol. Catal., A: Chemical, 99, 153 (1995). M. L. Wang, M. L., C. C. Ou and J. J. Jwo, Bull. Chem. Soc. Jpn., 67, 2249 (1994). Y. M. Wang, Z. P. Zhang, Z. Wang, J. B. Meng and P. Hodge, Chin. J. Polym. Sci., 16(4), 356 (1998). W. P. Weber and G. W. Gokel, Phase Transfer Catalysis in Organic Synthesis, Springer Verlag: Berlin, Heideberg, New York (1977). Y. L. Wen, M. Y. Yeh, Y. S. Lee and Y. P. Shih, J. Chin. Inst. of Chem. Eng., 27(6), 427 (1996). H. S. Weng and D. H. Wang, J. Chin. Inst. of Chem. Eng., 27(6), 419 (1996).
13.4 Effect of polymerization solvent
841
155 C. R. White U. S. Patent 4,642,399 (1987). 156 J. Wild and N. Goetz, Ger. Patent DE 3,820,979 (1989). 157 H. S. Wu, Study on the Displacement Reaction of Phosphazene with Trifluoroethanol by Phase-Transfer Catalysis, Ph.D. Thesis, Department of Chemical Engineering, National Tsing Hua University, Hsinchu, Taiwan (1990). 158 H. S. Wu, Fang, T. R., S. S. Meng and K. H. Hu, J. of Mol. Cata., A: Chemical, 136(2), 135 (1998). 159 M. Yamada, Y. Watabe, T. Sakakibara and R. Sudoh, J. Chem. Soc. Chem. Commun., 179 (1979). 160 B. Yan, Combinatorial Chemistry & High Throughput Screening, 1(4), 215 (1998). 161 S. S. Yufit, Russian Chemical Bulletin, 44(11), 1989 (1995).
13.4 EFFECT OF POLYMERIZATION SOLVENT ON THE CHEMICAL STRUCTURE AND CURING OF AROMATIC POLY(AMIDEIMIDE) Norio Tsubokawa Faculty of Engineering, Niigata University, Niigata, Japan
13.4.1 INTRODUCTION Aromatic poly(amide-imide) (PAI) has an outstanding resistance not only to the thermal operations but also mechanical, electrical, and chemical operations. Although the properties of PAI are inferior to those of aromatic polyimide (PI), which is one of the most heat-resistant polymers, PAI has been widely utilized as a high performance heat resistant polymer as well as PI, because PAI is superior to PI in its workability in industry.1,2 In general, PAI is prepared by the following two processes: diamine process and diisocyanate process. Diisocyanate process is achieved by the direct polycondensation of trimellitic anhydride (TMAH) with aromatic diisocyanate, such as 4,4’-diphenylmethane diisocyanate, in a polar solvent such as N-methyl-2-pyrrolidone (NMP) as shown in Eq. [13.4.1].1-3 On the other hand, the diamine process is achieved by a two-step reaction:1 (1) the polymerization (polycondensation and polyaddition) of trimellitic anhydride chloride (TMAH-CI) with aromatic diamine, such as 4,4’diaminodiphenylether (DDE) or 4,4’diaminodiphenylmethane (DDM) in a polar solvent such as N,N-dimethylacetamide (DMAc), to give poly(amic acid-amide) (PAAA) and (2) the imidation of PAAA by heating as shown in Eq. [13.4.2].
[13.4.1]
842
Norio Tsubokawa
[13.4.2]
PAI is supposed to be a linear polymer containing equivalent amounts of amide and imide bonds. Therefore, PAI is soluble in polar solvents such as NMP, DMAc, and N,N-dimethylformamide (DMF). However, it is well known that PAI becomes insoluble and infusible when it is heated over 200°C. We have pointed out that some remaining carboxyl groups of PAI play an important role in the curing of PAI by heating.4,5 Because of such a thermosetting property, PAI solution is used as a heat resistant temperature coating or an insulating enamel of magnet wire. In the following section, the effect of polymerization solvent, DMAc and mixed solvent of methyl ethyl ketone (MEK) with water,6 on the chemical structure and curing of PAIs prepared by diamine process will be summarized.7 13.4.2 EFFECT OF SOLVENT ON THE CHEMICAL STRUCTURE OF PAI 13.4.2.1 Imide and amide bond content of PAI Four kinds of PAAA, as precursor of PAI, were prepared by the polycondensation of TMAH-Cl with DDE or DDM in DMAc and MEK/water mixed solvent (MEK containing 30 vol% of water) at room temperature. The results are shown in Table 13.4.1. It was found that the conversions reached 78-96% within 2 h at room temperature and the rate of the polycondensation of TMAHCl with diamine in MEK/water mixed solvent was much larger than that in DMAc. Table 13.4.1. Polymerization conditions and conversion of samples [Data from reference 7] Sample No.
Solvent
Diamine
Time, h
Conversion, %
PAAA-1
MEK/H2O
DDE
0.5
90.3
PAAA-2
DMAc
DDE
2.0
89.6
PAAA-3
MEK/H2O
DDM
0.5
78.4
PAAA-4
DMAc
DDM
2.0
96.8
TMAH-Cl=diamine=0.06 mol ; solvent, 172 ml; TEA, 5.0 ml; room temp.
These PAAAs, PAAA-1, PAAA-2, PAAA-3, and PAAA-4, were heated at 180°C for 2 h to give PAI-1, PAI-2, PAI-3, and PAI-4, respectively. Figure 13.4.1 shows the infrared spectra of (A) PAAA-1, (B) PAI-1, and (C) PAI-3.
13.4 Effect of polymerization solvent
843
IR spectra of PAAA-1 show absorptions at 1530 and 1660 cm-1, which are characteristic of amide bond, 1510 and 1590 cm-1 which are characteristic of benzene nuclei, and 1230 cm-1, which is characteristic of ether bond, but the absorption at 1380 and 1780 cm-l, which are characteristic of imide bond, are hardly observed. On the other hand, IR spectra of PAI-1 and PAI-3, obtained by heating of PAAA-1 and PAAA-3, respectively, show new absorptions at 1380 and 1780 cm-1. Among the adsorptions of imide bond, the absorption at 1380 cm-1 is assigned to C-N stretching vibrations of all imide bond (cyclic and acyclic imide bond as shown in Eq. [ 13.4.3 ] and 1780 cm-1 is assigned to C=O stretching vibrations of five-member Figure 13.4.1 IR spectra of PAAA-1, PAI-1, and PAI-3 imide rings (cyclic imide bond). The absorption at 1530 cm-1 is assigned to N-H [Data from reference 7] stretching vibrations of amide bond. Furthermore, the absorption at 1510 cm-1 is assigned to benzene nuclei that is stable to heat treatment.
Therefore, the content of cyclic imide, amide and all imide bond of PAI was estimated by the absorbance ratio, D1380/D1510 (the absorbance ratio of absorbance at 1380 cm-1 to that of benzene nuclei), D1780/D1510, and D1530/D1510, respectively. Table 13.4.2 shows the absorbance ratio of cyclic imide, amide, and all imide bond before and after heat treatment of PAAAs. The considerable increase of cyclic and all imide bond content of PAAAs and the decrease of amide bond were observed by heating. But the effect of solvent on the imidation of PAAAs was hardly observed.
844
Norio Tsubokawa
Table 13.4.2. IR absorbance ratio of PAAA and PAI [Data from reference 7] Heat treatment Sample No.
Temperature o C
Time h
PAAA-1 PAI-1
180
2
PAAA-2 PAI-2
180
2
PAAA-3 PAI-3
180
2
PAAA-4 PAI-4
180
2
Absorbance ratio Cyclic imide
Amide
All imide
D1780/D1510
D1530/D1510
D1380/D1510
0.05
0.68
0.37
0.24
0.43
0.57
0.10
0.59
0.36
0.26
0.45
0.58
0.08
0.98
0.66
0.26
0.69
0.82
0.04
0.86
0.29
0.27
0.63
0.79
13.4.2.2 Intrinsic viscosity and carboxyl group content Table 13.4.3 shows the effect of solvent on the intrinsic viscosity, [η], and carboxyl group content of PAIs. The carboxyl content was determined by potentiometric titration. It is interesting to note that intrinsic viscosity of PAI from PAAA prepared in DMAc is larger than that in MEK/water mixed solvent. This suggests that the polymerization degree decreases with decreasing activity of TMAH-Cl in MEK/water mixed solvent because of the hydrolysis of TMAH-Cl by water. Table 13.4.3. Properties of PAI samples [Data from reference 7] Sample No.
Solvent
Diamine
Conversion, %
[η], dl/ga
COOH, eq/gb
PAI-1
MEK/H2O
DDE
90.3
0.36
214
PAI-2
DMAc
DDE
89.6
0.54
44
PAI-3
MEK/H2O
DDM
78.4
0.26
328
PAI-4
DMAc
DDM
96.8
0.32
73
a
Solvent, NMP ; 30.0°C. bDetermined by potentiometric titration
In addition, the content of carboxyl groups in PAIs prepared in MEK/water mixed solvent was considerably larger than that prepared in DMAc. This also suggests the hydrolysis of TMAH-Cl in MEK/water mixed solvent. 13.4.3 EFFECT OF SOLVENT ON THE CURING OF PAI BY HEAT TREATMENT 13.4.3.1 Chemical structure of PAI after heat treatment Table 13.4.4 shows the effect of solvent on the change of chemical structure and formation of insoluble part of PAI-1 and PAI-2 after heat treatment. It became apparent that the con-
13.4 Effect of polymerization solvent
845
tent of cyclic imide bond and all imide bonds of PAI-1 and PAI-2 further increased by heating of the corresponding PAAAs at 180°C for 2 h followed by heat treatment at 260°C for 2 h. The content of amide bond decreased by heating of PAAA at 180°C for 2 h, but increased by post-heating at 260°C for 2 h. Table 13.4.4. IR absorbance ratio and insoluble part of samples after heat treatment in air [Data from reference 7] Heat treatment Sample No.
Temperature o C
Time h
PAAA-1
Absorbance ratio
Insoluble part %
Cyclic imide
Amide
All imide
D1780/D1510
D1530/D1510
D1380/D1510
0.05
0.68
0.37
0
PAI-1
180
2
0.24
0.43
0.57
0
PAI-1
260
2
0.33
0.56
0.71
76.7
0.10
0.59
0.36
0
PAAA-2 PAI-2
180
2
0.26
0.45
0.58
0
PAI-2
260
2
0.31
0.50
0.64
16.8
13.4.3.2 Curing PAI by post-heating The formation of the insoluble part in NMP was observed and the amount of insoluble part formed by heating of PAI-1 (obtained in MEK/water mixed solvent) was larger than that by heating of PAI-2 (obtained in DMAc). The increase of imide and amide bond content by post-heating at 260°C is considered as follows: the imidation of amic acid structure may be proceeded by both intermolecular and intramolecular imidation. The latter produces cyclic imide bond, but the former produces acyclic imide bond to give crosslinking material as shown in Eq. [13.4.3]. Since PAIs obtained by heating of PAAAs at 180°C are completely soluble in NMP, the intramolecular imidation preferentially proceeds at 180°C, but intermolecular imidation scarcely proceeds. On the other hand, by post-heating at 260°C, the crosslinking reaction proceeds by the intermolecular imidation of terminal carboxyl groups of PAI with remaining amic-acid structure (Eq. [13.4.4]) and the amide bond in main chain of PAI (Eq. [13.4.5]) to give the insoluble part in NMP. The reaction induced the increase of imide bond content of PAI after post-heating. The increase of amide bond after post-heating may be due to the formation of crosslinking structure by the amidation of terminal amino groups of PAI with carboxyl groups of PAI in main chain (Eq. [13.4.5]-[13.4.8]). Therefore, PAI obtained from the heating of PAAA obtained in MEK/water mixed solvent, which has many carboxyl groups, produces more insoluble part by post-heating. Figure 13.4.2 shows the effect of heating temperature on the curing of PAI obtained by heating of PAAA at 180°C in MEK/water mixed solvent and DMAc. As shown in Figure 13.4.2, by post-heating at 280°C for 2 h in air, insoluble part in NMP reached 100%, indicating the almost complete curing of PAIs. It is interesting to note that PAIs formed by heating of PAAA obtained in MEK/water mixed solvent (PAI-1 and PAI-3) were found to be cured
846
Norio Tsubokawa
more easily than in DMAc (PAI-2 and PAI-4). This is due to the fact that PAI-1 and PAI-3 contains more carboxyl groups than PAI-2 and PAI-4. 13.4.4 CONCLUSIONS The effects of polymerization solvent on the curing of PAIs were investigated and the following results were obtained: (1) The carboxyl group content of PAIs from PAAAs prepared in MEK/H2O mixed solvent was larger than that in Figure 13.4.2. Relationship between heating temperature and insoluble part of PAI samples after heating (2h) in air DMAc. [Data from reference 7]. (2) The curing of PAIs prepared in MEK/H2O proceeded easier than those in DMAc. (3) The curing of PAIs by heating was due to the intermolecular reaction between functional groups of PAIs, such as carboxyl and amino groups. REFERENCES 1 2 3 4 5 6 7
M.Fujita and T.Fujita, Rubber Digest, 18 (1), 21 (1966). S.Terney, J.Keating, J.Zielinski, J.Hakala, and H.Sheffer, J. Polym. Sci., A-1, 8, 683 (1970). Y.Imai and T.Takahasi, Kobunshi Kagaku, 29, 182 (1972). N.Tsubokawa, I.Yamamoto, and Y.Sone, Kobunshi Ronbunshu, 44, 389 (1987). N.Tsubokawa, I.Yamamoto, and Y.Sone, Kobunshi Ronbunshu, 44, 831 (1987). H.Uchiyama and Y.Imai, Kobunshi Kagaku, 28, 73 (1971). N.Tsubokawa, M.Murata, and Y.Sone, Netsu Kokasei Jushi, 12, 12 (1991).
14
Solvent Use in Various Industries Attempts to reduce solvent use in the production of various materials require background information on the current inventory, the reasons for selecting certain solvents, the effect of various solvents on the properties of final products, future trends and the possibilities for solvent replacement. Information on solvent use investigates these areas. This chapter is possible only because of a thorough evaluation by large groups of scientists and engineers assembled by US Environmental Protection Agency. This has produced Compliance Sector Notebooks which contain invaluable information on solvent use. Full documents can be found on the EPA website at http://es.epa.gov/oeca/sector/index.html. These are useful in the investigation of a particular industry. Similar data from other countries were not available but open literature and patents have been included to help the reader to understand changes occurring in other countries.
14.1 ADHESIVES AND SEALANTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Adhesives and sealants are manufactured from a variety of polymers. Their selection and their combinations used impact solvent selection. Most solvent systems are designed to optimize the solubility of the primary polymer. Adhesives can be divided into ones which bond by chemical reaction and ones which bond due to physical processes.1 Chemically reactive adhesives are further divided into three more categories for those that bond through polymerization, polyaddition, or polycondensation. Physically bonding adhesives include pressure sensitive and contact adhesives, melt, or solution adhesives, and plastisols. Polymerization adhesives are composed of cyanoacrylates (no solvents), anaerobic adhesives (do not contain solvents but require primers for plastics and some metals which are solutions of copper naphthenate),2 UV-curable adhesives (solvent-free compositions of polyurethanes and epoxy), rubber modified adhesives (variety solvents discussed below). Polyaddition adhesives include epoxy and polyurethane polymers which can either be 100% solids, water-based, reactive or non-reactive hot melts or contain solvents mostly to regulate viscosity. Typical solvents include methyl ethyl ketone, acetone, mineral spirits, toluene, and xylene.3 Polycondensation adhesives include phenol-formaldehyde resin, polyamides, polyesters, silicones and polyimides. With the exception of polyesters (which require ethanol and N-methylpyrrolidone as solvents) and polyimides (which require
848
George Wypych
methyl amyl ketone, butyl acetate, methyl ethyl ketone, 2-ethoxyhexyl acetate as solvents), these adhesives can be made without solvents. Pressure sensitive and contact adhesives are made from a variety of polymers including acrylic acid esters, polyisobutylene, polyesters, polychloroprene, polyurethane, silicone, styrene-butadiene copolymer and natural rubber. With the exception of acrylic acid ester adhesives which can be processed as solutions, emulsions, UV curable 100% solids and silicones (which may contain only traces of solvents), all remaining rubbers are primarily formulated with substantial amounts of solvents such as hydrocarbon solvents (mainly heptane, hexane, naphtha), ketones (mainly acetone and methyl ethyl ketone), and aromatic solvents (mainly toluene and xylene). Melt adhesives and plastisols do not contain solvents. The solution adhesives group includes products made from the following polymer-solvent systems: nitrocellulose (typical solvents include solvent combinations usually of a ketone or an ester, an alcohol and a hydrocarbon selected from isopropanol, 2-butylhexanol, amyl acetate, acetone, methyl ethyl ketone), nitrile rubber (main solvent - methyl ethyl ketone), polychloroprene (which is usually dissolved in a mixture of solvents including a ketone or an ester, an aromatic and aliphatic hydrocarbon selected from naphtha, hexane, acetone, methyl ethyl ketone, benzene, toluene), and polyvinyl acetate (water). In addition to the solvents used in adhesives, solvents are needed for surface preparation4 and primers. Their composition may vary and is usually designed for a particular substrate, often using fast evaporating solvents and environmentally unfriendly materials with significant adverse health effects. Detailed data on the total amount of solvents used by adhesive industry could not be found. The adhesive manufacturing industry continues to grow at a very fast pace. Total adhesive production, according to Frost & Sullivan, was $18.25 billion in 1996 and this is expected to grow to $26.2 billion in 2003.5 Solvent-based materials in 1995 constituted 13% of total production in North America, 14% in Europe, 15% in Japan, and 25% in the Far East.6 Many industries which use solvent-based adhesives have moved to South America and Asia where regulations restricting emissions are less severe.5 The shoe industry is now concentrated in South America. There are many initiatives to decrease solvent emissions. For example, World Bank’s assistance program for developing countries focuses on this issue.5 But in spite of the fact that solvent-based adhesives lost some of their markets (3.3% during the period of 1994-1996),5 they still hold 14-15% of the European market.6 It is estimated that the use of solvents contributes 24% of all VOC emissions. According to one source adhesives were responsible for a 6% share in these emissions in 1993.7 Another source8 blames adhesives for 7% of total VOC emissions in Germany in 1995. Sealants are divided into groups according to the generic names of polymer base. The main groups include: polyurethanes, silicones, acrylics, polysulfide and others (PVC, polybutylene, styrene-butadiene-styrene copolymers, polychloroprene, and several others). The amount of solvent used in sealants is controlled by the standards which previously divided sealants into two groups: these below 10% VOC and those above. Recently, a provision was made to include water-based acrylics and the limit of VOC for class A sealants was increased to 20%. Polyurethane sealants and structural adhesives can be made without solvent (the first solvent-free polyurethane sealant was made in 1994).9 Solvents are added to reduce sealant viscosity and to aid in the manufacture of polymer. Typical solvents used are mineral spirits, toluene, and xylene. A small amount of solvent is emitted from curatives
14.1 Adhesives and sealants
849
which contain methyl ethyl ketone. Most formulations of silicone sealants do not contain solvents. In some sealants, traces of benzene and toluene can be found. Acrylic sealants are water-based but they may also contain ethylene and propylene glycols, mineral spirits and mineral oil. There are also solvent-based acrylic sealants which contain substantial amounts of solvents such as mineral spirits, toluene and xylene. Polysulfide sealants usually contain toluene but methyl ethyl ketone is also used. The group of class B sealants contains substantially more solvents (up to 40% by volume) but there are some exceptions. PVC sealants are based on plastisols and they can be made without solvents. Butyl rubber based sealants usually contain hydrocarbons (C6-C12). Styrene-butadiene-styrene based sealants usually have a large amount of solvents selected from a group including toluene, heptane, hexane, methyl ethyl ketone, isobutyl isobutyrate, n-amyl acetate, n-amyl ketone. They are usually processed in solvent mixtures. Polychloroprene is usually dissolved in a mixture of solvents including ketones or esters, and aromatic and aliphatic hydrocarbons. The list includes naphtha, hexane, acetone, methyl ethyl ketone, benzene, and toluene. The world market of sealants was estimated in 1996 at $2 billion and was expected to grow in 2003 to $2.75 billion with an annual growth rate of 4.5% which is slightly lower than that expected for adhesives (5.3%).5 The changing trends are clearly visible when developments in technology are studied but many barriers to reductions in solvent use exist such as the high investment required, longer processing time, frequently higher material cost of adhesives, and the psychological barrier of changing established adhesive practices. In many instances, adhesive performance is predicted by its superficial characteristics such as strong smell which might suggests that the material has superior properties, its initial green strength which for many indicates good bonding properties, and high viscosity often related to good processing characteristic.10 Since the alternative materials may not have much odor, or require of longer time to reach strength and have a low viscosity, users are suspicious that their potential performance may be inferior. The following information reviews some recent findings which may contribute to future changes. In the shoe industry, a major breakthrough occurred in 1928 when polychloroprene was first introduced.1,10 The first, simple formulation is still manufactured and is used worldwide because the glue can be easily prepared by simply making a solution of the polymer. This gives a product with good adhesion to various substrates. Many new products are available today as potential replacements. Hot melt adhesives can be used in some applications but they still require solvents for cleaning, degreasing, and swelling. Also, their bond strength is frequently inadequate. Reactive systems are not yet used in the shoe industry but reactive hot melts are finding applications. Their broader use is hampered by their sensitivity to moisture which requires special equipment and special care.10 Water-based adhesives are the most likely replacement product. They also need special equipment for processing because of the high heat of evaporation of water (although water based adhesives contain 50% polymer compared with 15-20% in solvent based adhesives).10 Two sport shoe manufacturers, Nike and Reebok, already use this technology. Traditional polychloroprene adhesives can be modified in several ways to be useful in water-based systems. Figure 14.1.1 shows peel strength of several adhesives. The solvent based adhesive (A) has excellent properties both in terms of green strength and bond strength. A simple emulsion of polychloroprene (B) has relatively good ultimate strength
850
George Wypych 60
150 Peel strength, N/25 mm
Peel strength, N/25 mm
1 min 7 day
100
50
0
A
B
C
D
Adhesive type Figure 14.1.1. Green strength and adhesion of several adhesives. Symbols are explained in the text. [Data from B Archer, International J. Adhesion & Adhesives, 18, No.1, 15-8 (1998).]
50 40 30 20 10 0
A
B
C
D
Adhesive type Figure 14.1.2. Contactability of various adhesives. Symbols are explained in the text. [Data from B Archer, International J. Adhesion & Adhesives, 18, No.1, 15-8 (1998).
but lacks green strength and therefore does not meet the performance requirements of shoe manufacturers and other industries.11 Adhesive (C) is a blend of polychloroprene with polyurethane in a water-based system. This modification gives both green strength and peel strength but Figure 14.1.2 shows that peel strength of a freshly applied adhesive is lower than that of solvent based polychloroprene which may cause problems in holding both adhering surfaces together. The adhesive (D) was developed in an interesting new process which involves the emulsification of a solvent based adhesive obtained from styrene-isoprene-styrene rubber.11 After emulsification, the solvent is stripped under vacuum to produce a solvent-free adhesive. The reason for emulsification of the complete adhesive as opposed to emulsification of the rubber alone is to produce a homogeneous adhesive system which would otherwise suffer from separation of polymer particles surrounded by a layer of emulsifying agents. Figures 14.1.1 and 14.1.2 show that this system is superior to solvent-based adhesives. Adhesives can be further improved by polymer blending and by adhesive foaming. A foamed adhesive layer requires less material (approximately 4 times less than a conventional adhesive), it requires less drying time (less water and faster evaporation), and provides an improved bond strength. A foamed adhesive has a larger surface area which increases the surface area of contact with the substrate. The compressed rubber foam has a higher tear strength than unfoamed film of the same thickness.11 Also, regulations are helping to reduce the solvent content of adhesives.12,13 The use of chlorinated solvents, frequently used in primers and for cleaning, has been discontinued based on the Montreal Protocol. From June 1998, the production of a pair of shoes in Europe should not involve the use of more than 20 g of solvent. This is only partially successful solution since shoe production is expected to move out of developed countries to less restrictive jurisdictions. Solvent Emission Directive will continue to restrict solvent use in Europe. Many changes have occurred and more are expected in adhesives based on thermoplastic polyurethanes (TPU).3,14,15In the last 30 years, TPU based adhesives were manufac-
14.1 Adhesives and sealants
851
tured with solvents for shoes, food packaging, and textile and plastic film lamination. Current technologies use TPU in the form of a hot melt, as a reactive PUR and as a thermoplastic laminating film.3 Reactive hot-melts were first introduced in the early 1980s and since than have grown very rapidly. After application the adhesive is cured by moisture.14 These adhesives are already in use by the automotive industry (bonding carpet to door panels, tray assembly, lenses to headlamp housing, and lamination of foam to fabric) and in furniture and building products (moldings, picture frames, decorative foil, edgebanding), in bookbinding, and in the footwear industry. Polyurethane water dispersions are expected to grow 8-10%/year from the current 5,000-6,000 tones/year market in Europe.14 Applications are similar to those of hot melts. UV-curable pressure-sensitive adhesives are the most recent application of the advancing radiation curing technology.16 Low viscosity formulations allow the use of standard application techniques with several advantages such as improved production rate, energy efficiency, improved properties of the final products, and new potential applications for pressure-sensitive additives in thicker films with mechanical performance. It is expected that radiation cured materials will expand at a rate of 10%/year.17 Adhesives constitute 16% by value and 13% by volume of radiation cured products (two major applications for radiation cured materials are coatings and inks). Henkel introduced a series of water-based laminating adhesives.18 Hot melt systems, high-solids solvent systems with a 3 times higher solids content, and water based adhesives have been introduced to textile lamination to replace traditional low-solids solvent-based adhesives.19 Odor elimination is the additional benefit which has helped to drive the replacement of solvent-based systems.20 In packaging materials, most odors are related to the solvents used in inks, coatings and adhesives. Also, coalescing solvents from water-based systems caused odors. Elimination of solvent is a priority but solvent replacement may also change the response to the odor because solvents such as toluene and xylene smell like lubricating oils or turpentine whereas isopropanol smells more like a disinfectant. Odors stem not only from solvents but also from products of the thermal and UV degradation of other components and solvents.20 In view of the above efforts, it is surprising that the majority of recent patents on adhesives are for solvent-based systems.21-26 The new inventions include a universal primer,21 an adhesive composition in which solvents have been selected based on Snyder’s polarity (only solvents which belong to group III are useful in adhesive for automotive applications to avoid a deleterious effect on paint),22 a low VOC adhesive for pipes and fittings,23 a solvent-containing heat-resistant adhesive based on siloxane polyimide,24 a water-based polyimide adhesive,25 and two-component solvent-free polyurethane adhesive system for use in automotive door paneling.26
REFERENCES 1 2 3 4 5 6 7
R Vabrik, G Lepenye, I Tury, I Rusznak, A Vig, International Polym. Sci. Technol., 25, No.3, T/1-9 (1998). D Raftery, M R Smyth, R G Leonard, International J. Adhesion Adhesives, 17, No.4, 349-52 (1997). J B Samms, L Johnson, J. Adhesive Sealant Coun., Spring 1998. Conference proceedings, Adhesive & Sealant Council, Orlando, Fl., 22nd-25th March 1998, p. 87. A Stevenson, D Del Vechio, N Heiburg, Simpson R, Eur. Rubber J., 181, 1, 2829, (1999). K Menzefricke, Adhesive Technol., 15, No.1, 6-7 (1998). L White, Eur. Rubber J., 179, No.4, 24-5 (1997). J Baker, Eur. Chem. News, 69, No.1830, 20-2 (1998).
852
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
George Wypych P Enenkel, H Bankowsky, M Lokai, K Menzel, W Reich, Pitture e Vernici, 75, No.2, 23-31 (1999). J. van Heumen, H Khalil, W Majewski, G. Nickel, G Wypych, US Patent 5,288,797, Tremco, Ltd., 1994. S Albus, Adhesive Technol., 16, No.1, 30-1 (1999). B Archer, International J. Adhesion & Adhesives, 18, No.1, 15-8 (1998). J C Cardinal, Pitture e Vernici, 74, No.6, 38-9 (1998). Pitture e Vernici, 73, No.20, 39-40 (1997). M Moss, Pigment Resin Technol., 26, No.5, 296-9 (1997). Hughes F, TAPPI 1997 Hot Melt Symposium. Conference Proceedings. TAPPI. Hilton Head, SC, 15th-18th June 1997, p.15-21. D Skinner, Adhesive Technol., 15, No.3, 22-4 (1998). B Gain, Chem. Week, 160, No.14, 28-30 (1998). G Henke, Eur. Adhesives & Sealants, 14, No.1, 18-9 (1997). G Bolte, J. Coated Fabrics, 27, 282-8 (1998). R M Podhajny, Paper, Film & Foil Converter, 72, No.12, 24 (1998). M Levy, US Patent 5,284,510, Paris Laque Service, 1994. I R Owen, US Patent 5,464,888, 3M, 1995. C D Congelio, A M Olah, US Patent 5,859,103, BFGoodrich, 1999. D Zhao, H Sakuyama, T Tomoko, L-C Chang, J-T Lin, US Patent 5,859,181, Nippon Mektron Ltd., 1999. H Ariga, N Futaesaku, H Baba, US Patent 5,663,265, Maruzen Petrochemical Co. Ltd., 1997. E Konig, U F Gronemeier, D Wegener, US Patent 5,672,229, Bayer AG, 1997.
14.2 AEROSPACE George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Figure 14.2.1 shows a schematic diagram of the aerospace manufacturing process.1 Metal finishing is the process in which most solvents and solvent containing materials are used. The main function of the metal finishing process is corrosion protection which requires proper cleaning, surface preparation, and the selection of suitable coatings. The functions of coatings used in aircrafts are different from those used in ordinary coating applications therefore an extrapolation of the progress made with solvent replacement in other coating types is not justified. The typical flight conditions of operating altitude (about 10,000 m above the earth), speed (most frequently 900 km/h), temperature (very low in space at about -60oC and substantially higher after landing up to 80oC), humidity (low in space and high at earth level are combined with condensation due to the tempera-
Figure 14.2.1. Schematic diagram of aerospace manufacturing process. [Reproduced from Profile of the Aerospace Industry. EPA Office of Compliance Sector Notebook Project. US Environmental Protection Agency. November 1998.]
14.2 Aerospace
853
Table 14.2.1. Total releases of solvents by the aerospace industry. [Data from ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
62,000
methyl ethyl ketone
995,000
bromotrifluoromethane
750
methyl isobutyl ketone
47,000
n-butyl alcohol
7,000
methyl tetr-butyl ether
550
sec-butyl alcohol
10,000
tetrachloroethylene
285,000
cyclohexane
400
1,1,1-trichloroethane
781,000
1,2-dichlorobenzene
600
trichloroethylene
429,000
1,1-dichloro-1-fluoroethane
10,000
trichlorofluoromethane
1,800
dichloromethane
314,000
toluene
414,000
isopropyl alcohol
1,000
xylene
103,000
methanol
21,000
ture difference), UV radiation (substantially higher during flight), mechanical abrasion due to the high speed of travel, exposure to salt in the atmosphere, exposure to higher level of acids and sulfur dioxide, and exposure to de-icing fluids during winter.2,3 These unusual conditions should be considered in conjunction with the mechanical movement of the coating caused by rapid changes in temperature and the flexing of aircraft elements because of changes in pressure and severe load variations on wings.2 In addition, because of their size, aircrafts must often be painted at low temperatures which requires a coating that will cure at these temperatures without leaving entrapped volatiles. These could evaporate in the low pressure conditions at high altitude and cause the formation of voids where corrosion could start. These factors make the design of an effective coating system a severe technological challenge. Coatings are used by the aerospace industry both for OEM and maintenance purposes. In each case surface cleaning and preparation is required. A paint stripping operation is added to the task in maintenance repainting. Coatings are applied by spraying, brushing, rolling, flow coating, and dipping. Depending on the method of application, the rheological properties of coatings must be adjusted with solvents and, in some cases, with water. An alternative method of viscosity adjustment involves heating the coatings to lower its viscosity by increasing its temperature. This reduces solvent usage. Solvents are also used for equipment cleaning. In addition to paints, sealants are also used. Sealants are mostly based on polysulfides, containing solvents as discussed in the previous section. Also, non-structural adhesives containing solvents are used as gaskets around windows and for carpeting. Paint removal is accomplished by either chemical or blast depainting. Dichloromethane is the most common solvent used for this application. Aerospace industry estimates that 15,000 to 30,000 different materials are used for manufacturing some of which are potentially toxic, volatile, flammable, and contain chlorofluorocarbons. Some of these substances may result in air emissions, waste-waters, and solid waste.
854
George Wypych
Table 14.2.2. Total transfers of solvents by the aerospace industry. [Data from ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
n-butyl alcohol
2,600
methyl ethyl ketone
500,000
cyclohexane
18,000
methyl isobutyl ketone
17,000
1,2-dichlorobenzene
4,000
tetrachloroethylene
110,000
1,1-dichloro-1-fluoroethane
230
1,1,1-trichloroethane
133,000
dichloromethane
68,000
trichloroethylene
98,000
N,N-dimethylformamide
500
trichlorofluoromethane
3,800
ethylene glycol
14,000
toluene
87,000
methanol
12,000
xylene
12,000
Air emissions result from sealing, painting, depainting, bonding, as well as from leakage in storage, mixing, drying, and cleaning. The most common solvents involved in coatings are trichloroethylene, 1,1,1-trichloroethane, toluene, xylene, methyl ethyl ketone, and methyl isobutyl ketone. Wastewater is generated through contamination by paints and solvents used for cleaning operations. Solid waste containing solvents comes from paint overspray intercepted by emission control devices, depainting, cleanup, and disposal of unused paint. Solvents used for cleaning are usually a mixture of dimethyl-benzene, acetone, 4-methyl-2-pentanone, butyl ester of acetic acid, naphtha, ethyl benzene, 2-butanone, toluene and 1-butanol. Some solvents used for painting and cleaning are either recycled are burned to recover energy. In 1996, 199 aerospace facilities (out of 1885 analyzed in the report) released and transferred off-site or discharged to sewers about 12,000 kg of 65 toxic chemicals (solvents in these releases are reported in Table 14.2.1 and transfers in Table 14.2.2). Methyl ethyl ketone, 1,1,1-trichloroethane, trichloroethylene, and toluene accounted for 66% of all releases. 70% of all transfers was for recycling purposes. The aerospace industry released 10,804 tons of VOC in 1997 which constituted 0.61% of the total releases from 29 industries which were analyzed. Thirteen other industries release more VOC than the aerospace industry. Recycling and disposal of solvents in the aerospace industry equals the purchase cost of the solvents. Therefore reduction of solvent use is very cost effective. Some chemical stripping operations are now being replaced by cryogenic stripping with liquid nitrogen. Also, supercritical carbon dioxide has been used in Hughes Aircraft Company in some cleaning applications. Solvent emissions can be reduced through control of evaporation (lids, chillers), by dedicating process equipment (reduces cleaning frequency), production scheduling, immediate cleaning of equipment, better operating procedures, reuse of solvent waste, and use of optimized equipment for paint application. There are plans to evaluate powder coatings and water-based paints.1,2 There are trials to use water as the paint thinner and to lower viscosity of paints by application of resins which have lower viscosity.2 Work is under the way to replace dichloromethane/phenol stripper with benzyl alcohol.2 The introduction of an intermediate layer between the primer and the top coat has been proposed.
14.3 Asphalt compounding
855
This will aid the stripping action of the proposed stripping solvent, benzyl alcohol. VOC have already been reduced in several components: bonding primer (from 1030 to 850 g/l), undercoats (from 670 to 350), top coats (from 700-900 to 250-800), clear coats (from 700-800 to 250-520), surface cleaners (from 850 to 250) as well as other materials.3
REFERENCES 1 2 3
Profile of the Aerospace Industry. EPA Office of Compliance Sector Notebook Project. US Environmental Protection Agency. November 1998. R W Blackford, Surface Coatings International, 80, No.12, 564-7 (1997). R Blackford, Polym. Paint Colour J., 186, No.4377, 22-4 (1996).
14.3 ASPHALT COMPOUNDING George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Numerous construction products are formulated from asphalt and coal tar for such applications as driveway sealers, cutback asphalts, flashing cements, concrete primers, concrete cold mixes, roof cements, expansion joint fillers, patch liquids, waterproofing liquid-applied membranes, and pipeline coatings. All these products are likely to contain solvents. The simplest formulations are mixtures of asphalt and (usually) mineral spirits used for sealing , priming, and coating of concrete. These are usually very low performance products which are used in large quantities because of their low price. They release about 40% of their weight to atmosphere during and after application. Since they do not perform well they have to be re-applied at frequent intervals. Driveway sealer is an example of a product which is used every spring, in spite of the fact that, in addition to the pollution it causes, it also produces a gradual degradation and cracking of the driveway. The only solution for elimination of this unnecessary pollution seems to be banning the product by regulation. Some of these products can be replaced by asphalt emulsions which contain water in place of organic solvents. Several products are used for patching and joint filling purposes. These materials (flashing cement, roof cement, patch liquid, and expansion joint filler) also use solvents to regulate viscosity. The solvents are usually mineral spirits, fuel oil, or polycyclic aromatic hydrocarbons. In addition to the base components, inexpensive fillers such as calcium carbonate or limestone but also still asbestos are added. These products harden on evaporation of the solvent and fill the joints, adhere to surfaces, and provide some waterproofing. These are again, low technology materials, traditionally used because of their very low cost. Most of these products can be replaced by modern sealants which will result in higher initial cost but longer service. The most technologically advanced products are used for waterproofing and pipeline coatings. These products are also based on dispersion of asphalt in the above mentioned solvents but reinforced with addition of polymer. The addition of polymer modifies the plastic behavior of asphalt and renders it elastomeric. Additional solvents are usually added to improve the solubility of polymeric components. Reactive polyurethanes are the most frequently used modifiers for waterproofing liquid membranes. Toluene and xylene are the
856
M Matsumoto, S Isken, J A M de Bont
most frequently used additional solvents. These materials partially solidify because of evaporation of the solvent. Their elastomeric properties are derived from chain extension and crosslinking reactions which form an internal polymeric network which reinforces asphalt. There is no data available on the solvent emissions from these materials but their scale of production suggests that their emissions are probably comparable with the entire rubber industry. This is one industry which should be closely monitored not only because of the emission of the above listed solvents but because some of the low grade solvents used contain large quantities of benzene and hexane. It is also cause for concern that asphalt and tar have carcinogenic components. Recent inventions1-5 are driven by product improvement needs and environmental aspects of application of these products. Janoski’s patent1 describes a product which is an anhydrous blend of polymer and asphalt and is substantially solvent-free. This technology shows that it is possible for an ingenious designer to produce low viscosity materials without using solvents but by selecting the appropriate type and concentration of bituminous materials, polyurethane components, and plasticizers. In another invention,2 a modifier is introduced to increase the adhesion of asphalt/water emulsions to aggregates. Emulsified asphalt is not so deleterious to the environment but its performance suffers from aggregate delamination. In yet another recent invention,3 terpene solvent, which is a naturally occurring (but never in this high concentration), biodegradable material, was used to replace the mineral spirits, xylene, trichloroethane, toluene, or methyl ethyl ketone normally used in cutback formulations (cutback asphalt is a dispersion of asphalt in a suitable solvent to reduce viscosity and allow for cold application). The two other patents4,5 discuss inventions leading to an improvement of high and low temperature properties of asphalt with no special impact on reduction of solvents used.
REFERENCES 1 2 3 4 5
R J Janoski, US Patent 5,319,008, Tremco, Inc., 1994. P Schilling, E Crews, US Patent 5,772,749, Westaco Corporation, 1998. R W Paradise, US Patent 5,362,316, Imperbel America Corporation, 1994. M P Doyle, J L Stevens, US Patent 5,496,400, Vinzoyl Petroleum Co., 1996. M P Doyle, US Patent 5,749,953, Vinzoyl Technical Services, LLC, 1998.
14.4 BIOTECHNOLOGY 14.4.1 ORGANIC SOLVENTS IN MICROBIAL PRODUCTION PROCESSES
Michiaki Matsumoto, Sonja Isken, Jan A. M. de Bont Division of Industrial Microbiology Department of Food Technology and Nutritional Sciences Wageningen University, Wageningen, The Netherlands
14.4.1.1 Introduction Solvents are not dominating compounds in the biosphere of our planet. Under natural conditions, their presence in appreciable amounts is restricted to specific areas. Only a very limited number of solvents is of biological origin and some may reach higher concentrations in nature. The best known example is ethanol. However, also butanol and acetone can be
14.4.1 Organic solvents in microbial production
857
formed readily by microbes and locally high concentrations may occur. In fact, in the beginning of the 20th century, very large production facilities were in operation for the microbial production of butanol and acetone. Furthermore, terpenes are natural solvents that are produced mainly by plants and locally they can reach high concentrations. For instance, limonene is present in tiny droplets in the peel of oranges. All these solvents are toxic to microbial cells. Some others as higher hydrocarbons that are present for instance in olive oil, are not toxic to microbes as will be discussed later. With the advent of the chemical industry, this picture has changed dramatically. In polluted locations, microorganisms may be confronted with a large number of solvents at high concentrations. With a few exceptions only, it has turned out that microbes can be found that are able to degrade these compounds if their concentration is low. This degradative potential is not unexpected in view of trace amounts that may be present locally in the natural biosphere. But the exposure of cells to unnatural high concentrations of these solvents usually leads to irreversible inactivation and finally to their death. The chemical industry is largely based on solvent-based processes. But in biotechnological processes, the microbes usually are exploited in a water-based system. This approach is quite understandable in view of the preference of microbes for water and the problems solvents pose to whole cells. Solvents often are used to extract products from the aqueous phase but only after the production process has been completed. At this stage, damage to whole cells is obviously no longer relevant. In both, chemical industry and biotechnology, organic solvents have many advantages over water because of the nature of either product or substrate. Consequently, during the last decades many possibilities have been investigated to use solvents in biocatalytic processes.1,2 The more simple the biocatalytic system, the less complex it is to use solvents. Free or immobilized enzymes have been exploited already in a number of systems. Here, biocatalysis may take place in reversed micelles or in an aqueous phase in contact with an organic solvent.3 In a powdered state some enzymes are able to function in pure organic solvents.4 Furthermore, modified enzymes such as polymer bound enzymes5 or surfactant-coated enzymes6 have been developed so that they can solubilize in organic solvents to overcome diffusion limitation. The advantages of enzymatic reactions using organic solvents can be briefly summarized as follows:1,3,4 1) hydrophobic substances can be used; 2) synthetic reactions can take place; 3) substrate or production inhibition can be diminished and 4) bioproducts and biocatalysts can easily be recovered from the systems containing organic solvents. Although organic solvents have often been used in enzymatic reactions, the application of organic solvents for biotransformation with whole-cell systems is still limited. Cells might be continuously in direct contact with the organic phase in a two-phase water-solvent system during the whole production cycle (Figure 14.4.1.1).7 Alternatively, cells may remain separated from the bulk organic phase by using membrane bioreactors (Figure14.4.1.2).8 In these instances, cells encounter phase toxicity9 or molecular toxicity, respectively. Because whole bacterial cells are more complex than enzymes, they pose by far greater problems in operating bioproduction processes when organic solvents are present. The most critical problem is the inherent toxicity of solvents to living organisms.1,2,10,11 As
858
M Matsumoto, S Isken, J A M de Bont
Fig.14.4.1.1. Schematic diagram of a two-phase bioreactor system for continuous 1-octanol production. [After reference 7]. 1-Octanol is produced from n-octane in hexadecene by Pseudomonas oleovorans or recombinant strains containing the alkane oxidation genes.
Fig.14.4.1.2. Schematic diagram of a two-phase hollow-fiber membrane bioreactor system for hydrolytic epoxide resolution. [After reference 8]. The yeast cells contain an epoxide hydrolase that enantioselectively hydrolyzes racemic epoxide resulting in enantiopure epoxide that partitions to the organic phase. Diol produced partitions to the water phase.
14.4.1 Organic solvents in microbial production
859
some solutions have been already found in this area, further progress is expected in the near future. A key problem is the selection of useful solvents in combination with a suitable microbe. Many solvents may be considered with their specific process properties, and many microorganisms may be considered also with their specific advantages or disadvantages. Only on the basis of a detailed understanding of the functioning of a microbial cell in the presence of solvents will it be possible to make a rational selection for a good combination solvent/organism. In the following section we will discuss these key questions. We will also describe recently found solvent-tolerant bacteria and will provide some examples of biotransformation using solvent-tolerant bacteria. 14.4.1.2 Toxicity of organic solvents The toxicity of an organic solvent is closely related to its hydrophobicity, as expressed by the log PO/W values,12-16 the logarithm of the partition coefficient of the solvent between 1-octanol and water (Table 14.4.1). In general, the Gram-negative bacteria show relatively higher solvent tolerances than the Gram-positive bacteria.12,14 This may be caused by the difference in the composition of the cell envelope. Solvents with a high log PO/W resulted in the highest activities, but the range of log PO/W tolerated by an organism is dependent on the type of microorganism. In the following section we will describe what happens in the cell in the presence of solvents.
+
+
+
decane
6.0
+
+
+
+
+
+
+
+
-
nonane
5.5
+
+
+
+
+
+
+
+
-
n-hexyl ether
5.1
+
+
+
+
+
+
+
+
-
octane
4.9
+
+
+
+
+
+
+
+
-
isooctane
4.8
+
+
+
+
+
+
+
-
-
cyclooctane
4.5
+
+
+
+
+
+
-
-
-
diphenyl ether
4.2
+
+
+
+
+
-
-
-
-
n-hexane
3.9
+
+
+
+
+
-
-
-
-
propylbenzene
3.7
+
+
+
+
-
-
-
-
-
o-dichlorobenzene
3.5
+
+
+
-
-
-
-
-
-
cyclohexane
3.4
+
+
+
-
-
-
-
-
-
ethylbenzene
3.2
+
+
-
-
-
-
-
-
-
ATCC26602
+
Saccharomyces uvarum
+
AHU1219
+
Bacillus subtilis
+
Agrobacterium tumefaciens IFO3058
Achromobacter delicatulus LAM1433
+
JCM1474
Escherichia coli IFO3806
+
Alcaligenes faecalis
Pseudomonas fluorescens IFO3507
7.0
IFO3738
Pseudomonas putida IH-2000
dodecane
Pseudomonas putida
Solvent
log PO/W
Table 14.4.1. Relationship between the growth of a cell exposed to an organic solvent and the value of log PO/W of the solvent. [After reference 14]
-
-
-
styrene
2.9
+
-
-
-
-
-
-
-
-
toluene
2.6
+
-
-
-
-
-
-
-
-
benzene
2.1
-
-
-
-
-
-
-
-
-
ATCC26602
-
AHU1219
-
Bacillus subtilis
-
Agrobacterium tumefaciens IFO3058
-
JCM1474
+
Alcaligenes faecalis
+
Achromobacter delicatulus LAM1433
Pseudomonas putida
3.1
Pseudomonas fluorescens IFO3507
Pseudomonas putida IH-2000
p-xylene
IFO3738
Solvent
log PO/W
Saccharomyces uvarum
M Matsumoto, S Isken, J A M de Bont
Escherichia coli IFO3806
860
+, growth; -, no growth
Bar9 suggested that the toxicity in two-phase systems was caused by both the presence of a second phase (phase toxicity) and solvent molecules which dissolved in the aqueous phase (molecular toxicity). Basically, both mechanisms are governed by the same principle in that the solvent accumulates in the microbial membrane. In case of the direct contact between cells and pure solvent, the rate of entry of solvents in a membrane will be very high. If the solvent has to diffuse via the water phase, then the accumulation in membranes will be slower. This latter mechanism on the molecular toxicity has been investigated in more detail.17 In experiments with liposomes from E. coli, and ten representative organic solvents labeled by 14C under aqueous-saturating levels, it was observed that the solvents accumulate preferentially in the cell membrane. The partition coefficients (log PM/B) of the solvents between the model liposome membrane and buffer correlate with those (log PO/W) in a standard 1-octanol-water system: log PM/B = 0.97 x log PO/W - 0.64
[14.4.1.1]17
The accumulation of an organic solvent in the membrane causes changes in the membrane structure. Organic solvents residing in the hydrophobic part of the membrane disturb the interactions between the acyl chains of the phospholipids. This leads to a modification of membrane fluidity which eventually results in the swelling of the bilayer.10 In addition to this, conformations of the membrane-embedded proteins may be altered.10 These changes in the integrity of the membrane also affect the membrane function. The principal functions of the cytoplasmic membrane involve: 1) barrier function, 2) energy transduction and 3) formation of a matrix for proteins. The disruption of lipid-lipid and lipid-protein interactions by the accumulation of organic solvents has a strong effect on the membrane’s function as a selective barrier for ions and hydrophilic molecules. Permeability is of particular importance for protons because the leakage of protons directly affects the primary energy transducing properties of the membrane. The initial rates of proton influx in the absence and presence of different amounts of hydrocarbon were measured.17 The permeability for protons increases with increasing amounts of hydrocarbon. Hence, leakage of protons occurs in the presence of organic sol-
14.4.1 Organic solvents in microbial production
861
Figure 14.4.1.3 Schematic presentation of possible mechanisms of solvent tolerance. A Changes in the structure of cytoplasmic membrane. B Changes in the structure of outer membrane. C Transformation of the solvent. D Active export of the solvents [After reference 32]
vents.17 Not only the impairment of the barrier function is caused by the alterations that occur in the membrane structure when it interacts with organic solvents. It is well known that the activities of the proteins embedded in the membrane are regulated by the membrane thickness, head group hydration, fluidity and fatty acid composition.18,19 All these parameters are also known to be affected by the accumulation of organic solvents. The effects of solvents on these parameters were reviewed by Sikkema et al.10
862
M Matsumoto, S Isken, J A M de Bont
14.4.1.3 Solvent-tolerant bacteria As described in the previous section, the organic solvents with 1 < log PO/W < 4 are considered to be toxic to microorganisms. In 1989, Inoue and Horikoshi21 found a toluene-tolerant Pseudomonas putida strain that grew in a two-phase toluene-water system (log PO/W = 2.5 for toluene). This finding was surprising and went against the dominant paradigm at that time. Solvent tolerance was confirmed by other strains of P. putida22-26 and by other representatives of the genus Pseudomonas.27-30 Furthermore, solvent tolerance has been found in the strains of Gram-positive bacteria Bacillus31,32 and Rhodococcus.33 The key question now is: How do solvent-tolerant bacteria overcome the toxic effects of organic solvents? Some of the possible mechanisms involved in solvent tolerance according to various researchers are shown in Fig.14.4.1.3.32 Current research on changes in the structure of the cytoplasmic membrane shows the involvement of: 1) the composition of the fatty acids of the phospholipids like the cis/trans isomerization of unsaturated fatty acids; 2) composition of phospholipid headgroups and 3) rate of turnover of membrane components. Organic solvents cause a shift in the ratio of saturated to unsaturated fatty acids.34,35 In a solvent-tolerant strain, an increase in the saturation degree has been observed during adaptation to the presence of toluene. Solvent-tolerant strains also have the ability to synthesize trans-unsaturated fatty acids from the cis-form in response to the presence of organic solvents.34,36-38 Increases in the saturation degree and the ratio of trans-form change the fluidity of the membrane and the swelling effects caused by solvents are depressed. Alterations in the headgroups of lipids during the adaptation to solvents have also been observed in some solvent-tolerant strains.37,39 The changes in the composition of the headgroups cause changes in the affinity of the lipids with the organic solvents and in the stability of membrane due to an alteration of bilayer surface charge density. These changes compensate the effect caused by the solvents. In one strain, the rate of phospholipid synthesis increases after exposure to a solvent.40 This strain has a repairing system which is faster than the rate of damage caused by the organic solvent. Unlike Gram-positive bacteria, Gram-negative bacteria such as Pseudomonas have an outer membrane. The outer membrane has been shown to play a role in the protection of the cell from solvent toxicity. Ions such as Mg2+ or Ca2+ stabilize the organization of the outer membrane and contribute to solvent tolerance.38 Low cell surface hydrophobicity caused by changes in the lipopolysaccharide (LPS) content has been reported to serve as a defensive mechanism.41,42 It has also been reported that the porins which are embedded in the outer membrane are relevant to solvent tolerance.37,42-44 The metabolism of organic solvents in solvent-tolerant strains contributes to solvent tolerance by degradation of the toxic compounds. This contribution, however, is considered to be limited33,45 because many solvent-tolerant strains show non-specific tolerance against various organic compounds. Non-specific tolerance to toxic compounds is well known in the field of antibiotic resistance. A wide range of structurally dissimilar antibiotics can be exported out of the cell by multidrug-efflux pumps. Could the export of organic solvents contribute to solvent tolerance? Isken and de Bont46 conducted experiments to determine whether the solvent tolerant Pseudomonas putida S12 was able to export toluene by monitoring the accumulation of 14C labeled toluene in the cells. Toluene-adapted cells were able to export toluene from their
14.4.1 Organic solvents in microbial production
863
Figure 14.4.1.4. Schematic picture of toluene penetration and efflux in the solvent-tolerant Pseudomonas putida S12 [After reference 20].
membranes whereas the non-adapted cells were not. Furthermore, it was observed that in the presence of energy coupling inhibitors, toluene accumulation was the same as in the non-adapted cells. The amount of toluene in the cell was concluded to be kept at a low level by an active efflux system. The presence of a toluene-efflux system is supported by genetic research.47,48 The pump has a striking resemblance to multidrug-efflux systems. Active efflux pumps for solvents have also been detected in other Pseudomonas strains.25,26,30,37 It is obvious that solvent tolerance is caused by a combination of the mechanisms described above. Figure 14.4.1.4 shows a schematic picture of toluene penetration and efflux in the solvent tolerant strain P. putida S12.20 Toluene enters the cell through the outer membrane. At present, it is unclear whether toluene passes through porins or through the phospholipid part of the cell. The efflux pump recognizes and interacts with toluene in the cytoplasmic membrane. Toluene is then pumped into the extracellular medium. 14.4.1.4 Biotransformation using solvent-tolerant microorganisms Many important fine chemicals, including catechols, phenols, aldehydes and ketones, low molecular epoxides and diepoxides, medium-chain alcohols, and terpenoids fall within the range of 1 < log PO/W < 4. The discovery of solvent-tolerant bacteria leads to the new possibility of biocatalytic reaction systems containing organic solvents. By using solvent-tolerant bacteria, a variety of fine chemicals can be formed in microbial production processes. The organic solvents used so far in published research had to be very hydrophobic (logPO/W > 5) in order to prevent microbial inactivation.1 Consequently, many fine chemicals can not beneficially produced in the presence of such solvents because they simply would remain in the water phase and would not partition to the organic phase. The use of solvent-tolerant microorganisms enables the use of less hydrophobic solvents (2.5 < logPO/W < 4) and in such a system, chemicals with a 1 < logPO/W < 4 preferentially will go into the organic phase. Up to now, however, only a few applications using solvent-tolerant microorganisms have been reported. Aono et al.49 reported the oxidative bioconversion of cholesterol as a model biocatalytic reaction using a solvent-tolerant Pseudomonas species. Cholesterol and its
864
M Matsumoto, S Isken, J A M de Bont
products are insoluble in an aqueous solution but dissolve in some organic solvents. The attempt was successful. The conversion of cholesterol was more than 98% and the yield of oxidative products was 80%. Speelmans et al.50 reported on the bioconversion of limonene to perillic acid by a solvent-tolerant Pseudomonas putida. The microbial toxicity of limonene is known to be very high. It is a major component of citrus essential oil and is a cheap and readily available base material. By using a solvent-tolerant strain perillic acid was obtained at a high concentration. This finding brings commercial production nearer. The applications of solvent-tolerant strains in microbial production processes are at present limited, but two strategic options are currently available to use such bacteria.20 Relevant genes can be introduced into solvent-tolerant organisms in order to produce the required product. This approach has been followed successfully by J. Wery in our laboratory who employed an 1-octanol-aqueous system. Methylcatechol was produced from toluene by solvent tolerant P. putida S12. Alternatively, the efflux pump can be expressed in a suitable solvent-sensitive host which would then be more tolerant for a particular solvent. Other benefits may arrive from solvent-resistant bacteria. Ogino et al.26 isolated Pseudomonas aeruginosa LST-03 which can grow in organic solvents with logPO/W >2.4 and secrets organic solvent-stable lipolytic enzymes. They were able to purify an organic solvent-stable protease which was more stable than the commercially available proteases.51 Hence, solvent-tolerant strains have become a source for new enzymes.52 In the near future, the use of solvent-tolerant strains will make the application of organic solvents in biotransformations by whole cells a more realistic option. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
R. León, P. Fernandes, H.M. Pinheiro, and J.M.S. Cabral, Enz. Microb. Technol., 23, 483 (1998). B. Angelova and H.S. Schmauder, J. Biotechnol., 67, 13 (1999). C. Laane, S. Boeren, K. Vos, and C. Veeger, Biotechnol. Bioeng., 30, 81 (1987). A.M. Klibanov, Trends Biotechnol., 15, 97 (1997). Y. Inada, A. Matsushima, M. Hiroto, H. Nishimura, and Y. Kodera, Methods Enzymol., 242, 65 (1994). Y. Okahata, Y. Fujimoto, and K. Iijiro, J. Org. Chem., 60, 2244 (1995). R.G. Mathys, A. Schmid, and B. Witholt, Biotechnol. Bioeng., 64, 459 (1999). W.Y. Choi, C.Y. Choi, J.A.M. de Bont, and C.A.G.M. Weijers, Appl. Microbiol. Biotechnol., 53, 7 (1999). R. Bar, J. Chem. Technol. Biotechnol., 43, 49 (1988). J. Sikkema, J.A.M. de Bont, and B. Poolman, Microbiol. Rev., 59, 201 (1995). M.D. Lilly and J.M. Woodley, in Biocatalysis in organic synthesis, J. Tramper, H.C. van der Plas, and P. Linko, Ed., Elsevier, Amsterdam, 1985, pp.179-192. M. Vermuë, J. Sikkema, A. Verheul, and J. Tramper, Biotechnol. Bioeng., 42, 747 (1993). S.D. Doig, A.T. Boam, D.J. Leak, A.G. Livingston, and D.C. Stuckey, Biocatal. Biotransform., 16, 27 (1998). A. Inoue and K. Horikoshi, J. Ferment. Bioeng., 71, 194 (1991). A.J. Harrop, M.D. Hocknull, and M.D. Lilly, Biotechnol. Lett., 11, 807 (1989). A.N. Rajagopal, Enz. Microb. Technol., 19, 606 (1996). J. Sikkema, J.A.M. de Bont, and B. Poolman, J. Biol. Chem., 269, 8022 (1994). H. Sandermann, Jr., Biochim. Biophys. Acta, 515, 209 (1978). P.L. Yeagle, FASEB J., 3, 1833 (1989). J.A.M. de Bont, Trends Biotechnol., 16, 493 (1998). A. Inoue and K. Horikoshi, Nature, 338, 264 (1989). D.L. Cruden, J.H. Wolfram, R.D. Rogers, and D.T. Gibson, Appl. Environ. Microbiol., 58, 2723 (1992). F.J. Weber, L.P. Ooykaas, R.M.W. Schemen, S. Hartmans, and J.A.M. de Bont, Appl. Environ. Microbiol., 59, 3502 (1993). J.L. Ramos, E. Deque, M.J. Huertas, and A. Haïdour, J. Bacteriol., 177, 3911 (1995). K. Kim, S.J. Lee, K.H. Lee, and D.B. Lim, J. Bacteriol., 180, 3692 (1998).
14.4.2 Solvent-resistant microorganisms
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
865
F. Fukumori, H. Hirayama, H. Takami, A. Inoue, and K. Horikoshi, Extremophiles, 2, 395 (1998). H. Nakajima, H. Kobayashi, R. Aono, and K. Horikoshi, Biosci. Biotechnol. Biochem., 56, 1872 (1992). H. Ogino, K. Miyamoto, and H. Ishikawa, Appl. Environ. Microbiol., 60, 3884 (1994). Y. Yoshida, Y. Ikura, and T. Kudo, Biosci. Biotechnol. Biochem., 61, 46 (1997). X.Z. Li, L. Zhang, and K. Poole, J. Bacteriol., 180, 2987 (1998). K. Moriya and K. Horikoshi, J. Ferment. Bioeng., 76, 397 (1993). S. Isken and J.A.M. de Bont, Extremophiles, 2, 229 (1998). M.L. Paje, B.A. Neilan, and I. Couperwhite, Microbiology, 143, 2975 (1997). F.J. Weber, S. Isken, and J.A.M. de Bont, Microbiology, 140, 2013 (1994). H.C. Pinkart, J.W. Wolfram, R. Rogers, and D.C. White, Appl. Environ. Microbiol., 62, 1129 (1996). H.J. Heipieper, G. Meulenbeld, Q. van Oirschot, and J.A.M. de Bont, Chemosphere, 30, 1041 (1995). J.L. Ramos, E. Duque, J.J. Rodriguez-Herva, P. Godoy, A. Haidour, F. Reyes, and A. Fernandez-Barrero, J. Biol. Chem., 272, 3887 (1997). F.J. Weber and J.A.M. de Bont, Biochim. Biophys. Acta, 1286, 225 (1996). V. Pedrotta and B. Witholt, J. Bacteriol., 181, 3256 (1999). H.C. Pinkart and D.C. White, J. Bacteriol., 179, 4219 (1997). R. Aono and H. Kobayashi, Appl. Environ. Microbiol., 63, 3637 (1997). H. Kobayashi, H. Takami, H. Hirayama, K. Kobata, R. Usami, and K. Horikoshi, J. Bacteriol., 181, 4493 (1999). L. Li, T. Komatsu, A. Inoue, and K. Horikoshi, Biosci. Biotechnol. Biochem., 59, 2358 (1995). H. Asano, K. Kobayashi, and R. Aono, Appl. Environ. Microbiol., 65, 294 (1999). G. Mosqueda, M.-S. Ramos-González, and J.L. Ramos, Gene, 232, 69 (1999). S. Isken and J.A.M. de Bont, J. Bacteriol., 178, 6056 (1996). J. Kieboom, J.J. Dennis, G.J. Zylstra, and J.A.M. de Bont, J. Biol. Chem., 273, 85 (1998). J. Kieboom, J.J. Dennis, G.J. Zylstra, and J.A.M. de Bont, J. Bacteriol., 180, 6769 (1998). R. Aono, N. Doukyu, H. Kobayashi, H. Nakajima, and K. Horikoshi, Appl. Environ. Microbiol., 60, 2518 (1994). G. Speelmans, A. Bijlsma, and G. Eggink, Appl. Microbiol. Biotechnol., 50, 538 (1998). H. Ogino, F. Watanabe, M. Yamada, S. Nakagawa, T. Hirose, A. Noguchi, M. Yasuda, and H. Ishikawa, J. Biosci. Bioeng., 87, 61 (1999). N. Doukyu and R. Aono, Appl. Environ. Microbiol., 64, 1929 (1998).
14.4.2 SOLVENT-RESISTANT MICROORGANISMS
Tilman Hahn, Konrad Botzenhart Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Tòbingen, Tòbingen, Germany
14.4.2.1 Introduction Several main properties of microorganisms in relation to solvents can be considered: • Toxic or antimicrobial effects of solvents • Solvent resistance or adaptation of microorganisms • Metabolic activities of microorganisms Antimicrobial effects or solvent-resistant microorganisms are the main topic of this section. 14.4.2.2 Toxicity of solvents for microorganisms 14.4.2.2.1 Spectrum of microorganisms and solvents The growth-inhibiting effects of several solvents on microorganisms are described.1,2 Organic solvents have toxic effect on microorganisms. Table 14.4.2.1 summarizes the relevant organic solvents and their toxicity concerning selected microorganisms.
866
Tilman Hahn, Konrad Botzenhart
Table 14.4.2.1. Toxicity of organic solvents - examples Solvent
Microorganisms
References
Toluene, benzene, ethylbenzene, propylbenzene, xylene, hexane, cyclohexane
Pseudomonas putida
Isken et al. (1999)3 Gibson et al. (1970)4
Terpenes, e.g., alpha-pinene, limonene, β-pinene, terpinolene
Bacillus sp., Saccharomyces cerevisiae, isolated mitochondria
Andrews et al. (1980)5 Uribe et al. (1985)6
Styrene
soil microorganisms
Hartmans et al. (1990)7
Cyclohexane
yeast cells, isolated mitochondria
Uribe et al. (1990)8
Aromatic hydrocarbons
isolated bacterial and liposomal membranes
Sikkema et al. (1992)9 Sikkema et al. (1994)1
Ethanol
yeasts
Cartwright et al. (1986)10 Leao and van Uden (1984)11
14.4.2.2.2 Mechanisms of solvent toxicity for microorganisms The toxicity of organic solvents or hydrophobic substances for microorganisms depends mainly on their effects on biological membranes1,9,12-14 - similar to membrane effects of several anesthetics. This concerns especially effects on cytoplasmatic membranes. The following main changes of membrane structures and functions have been observed: • Accumulation of hydrophobic substances such as organic solvents in cytoplasmatic membranes. This accumulation causes structural and functional changes in the cytoplasmatic membranes and microbial cells. • Structural changes in cytoplasmatic membranes, e.g., swelling of membrane bilayers, increase of surface and thickness of the membranes, changes in the composition of the membrane (e.g., changes in the fatty acid composition), modification of the microviscosity, damage of membrane structures (see below). • Loss of membrane integrity, especially disruption of cytoplasmatic membranes, less damage of outer membranes. Because of these damages often complex cellular structures (e.g., vesicula) or cell functions (decrease of respiratory activities of mitochondria) are destroyed or inhibited. • Interactions of the accumulated lipophilic substances with the cytoplasmatic membranes and especially hydrophobic parts of the cell or cell membranes. Lipid-lipid interactions and interactions between proteins and lipids of the membrane structure (lipid bilayers, membrane-embedded proteins) are discussed. • Effects on passive and active membrane transport systems, e.g., increase of passive efflux and flux of ions such as protons, cations Mg++ and Ca++ or small molecules, stimulation of the leakage of protons and potassium, changes in the uptake of compounds (e.g., solvents) and excretion (e.g., metabolic products), inhibition of active transport systems (e.g., ATP depletion). • Damage of cellular homeostasis and cell physiology, e.g., reduction of transmembrane electrical potentials and proton chemical potentials or proton motive forces as a result of membrane changes (efflux of ions), changes of pH gradients.
14.4.2 Solvent-resistant microorganisms
867
• Changes in enzyme activities, e.g., inhibition of oxidases and depletion of ATP. Of special relevance are various interactions with enzymes (proteins) in the membrane, e.g., lipid-protein interactions. • Loss of particular cellular functions, e.g., respiratory system of mitochondria or active transport systems (see above). • Loss of complex cell functions, e.g., reduced growth rates and activities of microorganisms. The extent of solvent toxicity to microorganisms is determined by various factors: (a) Hydrophobic or lipophilic properties of solvents. The toxicity of solvents to microorganisms can be described by a partition coefficient (log Po/w) between organic compounds (solvents) and water which is specific for the applied substance. This partition coefficient is based on a standard octanol-water system model.1,15,16 The toxicity and the affinity of solvents to cell structures increase with hydrophobic properties of solvents, e.g., high toxicities with Po/w values of 1-5.1,14 The partition coefficient correlates with the membrane-buffer partition coefficient between membrane and aqueous system.1,14,16 They also depend on membrane characteristics.17,18 (b) Accumulation, partitioning and concentrations of solvents in cell structures (membranes). Dissolution and partitioning of solvents depend essentially on solvent properties, e.g., polarity (specific partition coefficients), or membrane characteristics (influence on partition coefficients). Both dissolution and partitioning can be influenced by additional factors, e.g., cosolvents. The effects on microorganisms can depend typically on solvent concentrations, e.g., dose response effects. (c) Biomass, ratio of concentrations of solvents and biomass. Effect of solvents depends on this ratio. (d) Surrounding conditions, e.g., temperature which influences the proton leakage and microbial activities.19 The toxicity of solvents for microorganisms shows positive and negative consequences, e.g.: • Positive aspects such as antibacterial effects20 which are found in several products. • Negative aspects such as reduced stability of biotransformation and bioremediation processes because of the inactivation of microorganisms. 14.4.2.3 Adaption of microorganisms to solvents - solvent-resistant microorganisms 14.4.2.3.1 Spectrum of solvent-resistant microorganisms Different microorganisms are able to adapt and even to grow in the presence of solvents. Some relevant examples are given in Table 14.4.2.2. Table 14.4.2.2. Solvent-tolerant microorganisms and their resistance to organic solvents Solvent-tolerant microorganisms bacteria from deep sea (1.168 m, Japan)
Solvents benzene, toluene, p-xylene, biphenyl, naphthalene
References Abe et al. (1995)21
868
Tilman Hahn, Konrad Botzenhart
Solvent-tolerant microorganisms
Solvents
References
deep sea isolates, Flavobacterium strain DS-711, Bacillus strain DS-994
hydrocarbons
Moriya and Horikoshi (1993)22
marine yeasts
n-alkanes
Fukamaki et al. (1994)23
Pseudomonas putida
toluene, m-, p-xylene, 1,2,4-trimethylbenzene, 3-ethyltoluene
Inoue and Horikoshi (1989)24 Cruden et al. (1992)25
Pseudomonas sp.
alpha-pinene
Sikkema et al. (1995)14
E. coli K-12
p-xylene
Aono et al. (1991)26
The concentration of organic solvents which can be tolerated by the microorganisms varies extensively, e.g., growth of Pseudomonas putida in the presence of more than 50 % toluene27 but tolerance of E. coli K-12 in the presence of only up to 10 % p-xylene.26 The growth characteristics of different species can differ to a great extent, e.g., no growth, growth with or without metabolizing solvents. Generally higher solvent tolerance of Gram-negative bacteria compared to Gram-positive bacteria is observed.28 Some solvent-tolerant microorganisms cannot use organic solvents as substrate for growth and need other substrates in complex media.26,27 Other microorganisms can use organic solvents in minimal media as a source of energy or carbon, e.g., Pseudomonas putida in the presence of xylene or toluene.25 14.4.2.3.2 Adaption mechanisms of microorganisms to solvents The mechanisms of solvent tolerance are only partly known.29-31 Relevant microbial adaptation mechanisms are: - Changes in the composition of the cytoplasmatic membrane. The compounds of the membranes such as lipids or proteins can influence the membrane characteristics and, therefore, the adaption to solvents. Mainly phospholipids in the membrane bilayer determine the partitioning of solutes and especially the resistance to solvents17,18 A reduction of the partition coefficient has been observed when the fatty acid composition was changed.32 An increase of monounsaturated fatty acids and a decrease of saturated fatty acids correlated with a higher ethanol tolerance of S. cerevisiae, E. coli, and Lactobacillus strains.33-35 An increase of unsaturated fatty acids is induced by polar solvents and low temperatures, an increase of saturated fatty acids is connected with more apolar solvents and high temperatures.36 Even changes in the configuration of fatty acids, which are provoked by solvents, can lead to adaption mechanisms, e.g., cis-trans conversions.29 - Changes in the microbial structure. Various structural changes can cause a reduction of toxic solvent effects, e.g., the increase of membrane fluidity which is connected with an increase of unsaturated fatty acids (see above) and results in a decrease of the membrane permeability. - Specific structural characteristics of microorganisms. Typical structures of microorganisms vary according to the microbial species, e.g., outer membrane characteristics of bacteria. For instance, Gram-negative bacteria such as Pseudomonas sp. tolerate higher concentrations of hydrophobic compounds compared to Gram-positive bacteria. The resistance of the outer membrane correlates with the solvent-tolerance.28
14.4.2 Solvent-resistant microorganisms
869
- Alterations of the cell envelope structure (cell wall). Mechanical alterations and chemical modifications of the cell wall can reduce the microbial resistance to solvents. The most interesting chemical modification concerns hydrophobic or hydrophilic abilities of the cell wall.37 Decreasing hydrophobicity of the cell wall enhances the adaption of microorganisms to solvents.38 - Suppression of the effects of solvents on membrane stability. - Limitation of solvent diffusion into the cell (see above). - Repairing mechanisms, e.g. enhanced phospholipid biosynthesis. - Transport or export systems. The excretion of compounds out of the microbial cell and cytoplasmatic membrane is well known but only documented for some substances, e.g., for drugs.39 Passive and active transport systems are relevant, e.g., ATP driven systems. Export systems for the several solvents must be assumed. - Immobilization and mobilization of microorganisms and solvents. The adsorption of solvents to microorganisms can be reduced if the contact is decreased. For instance, immobilization of microorganisms or solvents minimizes the contact. An immobilization and reduction of toxicity was shown if adsorption materials were added.40 - Surrounding conditions, e.g., low temperature, which can induce higher solvent resistance of microorganisms (see above). 14.4.2.4 Solvents and microorganisms in the environment and industry examples Microorganisms are frequently observed in organic-aqueous systems containing solvents are essential in natural and in industrial processes. The occurrence and role of microorganisms and organic compounds in these two-phase organic-aqueous systems are similar to the effects described above (see Section 14.4.2.2). Although toxic effects on microorganisms in these natural and industrial processes are well known, reliable data concerning solvent-resistant microorganisms are not available. 14.4.2.4.1 Examples 14.4.2.4.1.1 Biofilms, biofouling, biocorrosion Important examples for organic-aqueous systems are surface-associated biofilms which are a form of existence of microorganisms. Microorganisms, mostly bacteria, are embedded into a glycocalyx matrix of these biofilms.41,42 This biofilm matrix mediates the adhesion of microorganisms to surfaces, concentrates substances and protects microorganisms from antimicrobial agents.41,42 Several organic-aqueous systems can be observed, especially surface of surrounding materials (pipes, etc.) in relation to water or ingredients (e.g., oil in pipes) related to water between ingredients and surrounding materials (pipes, etc.). Some aspects of solvents in these organic-aqueous biofilm systems are studied. Solvents can occur in water systems emitted from surrounding organic materials.43 It was shown that solvents are important concerning microbial biocorrosion and biofouling processes, e.g., by swelling and hydrolysis of materials.44 Despite these well-known aspects, reliable data and studies concerning solvent-resistant microorganisms in biofilm, biofouling, or biocorrossion processes are not shown. Nevertheless similar mechanisms in biofilms must be assumed as described above (see Section 14.4.2.3) because similar conditions occur (organic-aqueous systems).
870
Tilman Hahn, Konrad Botzenhart
14.4.2.4.1.2 Antimicrobial effects, microbial test systems The toxic mechanisms of solvents to microorganisms described above (see Section 14.4.2.2) are frequently used in effects of antimicrobial agents. The damage of microbial biomembranes is fundamentally connected with the antimicrobial effects of several solvents on bacteria.45 Biomembranes and other microbial structures can be affected by solvents via similar processes as described. Relevant examples are naked viruses which are generally more resistant to viruzi agents because the envelopes of viruses are damaged by viruzi substances such as some solvents. Another example of solvent-like interactions are effects of antimicrobial agents to capsules of bacterial spores, e.g., Bacillus species. Bacterial or enzymatic toxicity tests are used to assay the activity of organic compounds including solvents. A survey of environmental bacterial or enzymatic test systems is given by Bitton and Koopman.46 The principles of these test systems are based on bacterial properties (growth, viability, bioluminescence, etc.) or enzymatic activities and biosynthesis. The toxicity of several solvents were tested in bacterial or enzymatic systems, e.g., pure solvents such as phenol in growth inhibition assays (Aeromonas sp.),46 solvents in complex compounds such as oil derivates,46,47 solvents in environmental samples such as sediments or solvents used in the test systems.46,48,49 The efficiency of several test systems, e.g., Microtox tests or ATP assays, vary, e.g., looking at the effects of solvents.46 14.4.2.4.1.3 Industrial processes In industrial processes the main microbial activities connected with solvents are: • Processes in biotechnology, biotransformation and biocatalysis,50,51 e.g., production of chemicals from hydrophobic substrates or use of solvents as starting materials for microbiological reactions. • Bioremediation: degradation of environmental pollutants, e.g., wastewater treatment or bioremediation in biofilm reactors.52-55 Various microorganisms and microbial mechanisms are relevant in these industrial processes. Examples are conversion processes of organic substances, e.g., by bacterial oxygenases.56,57 Normally low-molecular-weight aromatic hydrocarbons including solvents are converted in these biotransformation processes.58 References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
J. Sikkema, J.A.M. deBont, B.Poolman, J. Biol. Chem., 269, 8022 (1994). G.J. Salter, D.B. Kell, Crit. Rev. Biotechnol., 15, 139 (1995). S. Isken, A. Derks, P.F.G. Wolffs, J.A.M. deBont, Appl. Environ. Microbiol., 65, 2631 (1999). D.T. Gibson, G.E. Cardini, F.C. Maseles, R.E. Kallio, Biochemistry, 9, 1631 (1970). R.E. Andrews, L.W. Parks, K.D. Spence, Appl. Environ. Microbiol., 40, 301 (1980). S. Uribe, J. Ramirez, A. Pena, J. Bacteriol., 161, 1195 (1985). S. Hartmans, M.J. van der Werf, J.A.M. de Bont, Appl. Environ. Microbiol., 56, 1347 (1990). S. Uribe, P. Rangel, G. Espinola, G. Aguirre, Appl. Environ. Microbiol., 56, 2114 (1990). J. Sikkema, B. Poolman, W.N. Konigs, J.A.M. deBont, J. Bacteriol., 174, 2986 (1992). C.P. Cartwright, J.R. Juroszek, M.J. Beavan, F.M.S. Ruby, S.M.F. DeMorais, A.H. Rose, J. gen. Microbiol., 132, 369 (1986). C. Leao, N. vanUden, Biochim. Biophys. Acta, 774, 43 (1984). M.J. DeSmet, J. Kingma, B. Witholt, Biochem. Biophys. Acta, 506, 64 (1978). M.R. Smith in Biochemistryof Microbial Degradation, C. Ratledge, Ed, Kluwer Academic Press, Dordrecht, 1993, pp. 347-378. J. Sikkema, J.A.M. deBont, B. Poolman, B., FEMS Microbiol. Rev., 59, 201 (1995). A. Leo, C. Hansch, D. Elkins, Chem. Rev., 71, 525 (1971).
14.4.2 Solvent-resistant microorganisms
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
871
W.R. Lieb and W.D. Stein in Transport and Diffusion Across Cell Membranes, W.D. Stein, Ed., Academic Press, N.Y., 1986, pp. 69-112. M.C. Antunes-Madeira, V.M.C. Madeira, Biochim. Biophys. Acta, 861, 159 (1986). M.C. Antunes-Madeira, V.M.C. Madeira, Biochim. Biophys. Acta, 901, 61 (1987). W. DeVrij, A. Bulthuis, W.N. Konings, J. Bacteriol., 170, 2359 (1988). F.M. Harold, Adv. Microb. Physiol., 4, 45 (1970). A. Abe, A. Inoue, R. Usami, K. Moriya, K. Horikoshi, Biosci. Biotechnol. Biochem., 59, 1154 (1995). K. Moriya, K. Horikoshi, J. Ferment. Bioeng., 76, 168 (1993). T. Fukamaki, A. Inoue, K. Moriya, K. Horikoshi, Biosci. Biotechnol. Biochem., 58, 1784 (1994). A. Inoue, K. Horikoshi, Nature, 338, 264 (1989). D.L. Cruden, J.H. Wolfram, R. Rogers, D.T. Gibson, Appl. Environ. Microbiol., 58, 2723 (1992). R. Aono, K. Albe, A. Inoue, K. Horikoshi, Agric. Biol. Chem., 55, 1935 (1991). A. Inoue, M. Yamamoto, K. Horikoshi, K., Appl. Environ. Microbiol., 57, 1560 (1991). A.J. Harrop, M.D. Hocknull, M.D. Lilly, Biotechnol. Lett., 11, 807 (1989). H.J. Heipieper, R. Diefenbach, H. Keweloh, Appl. Environ. Microbiol., 58, 1847 (1992). H.C. Pinkart, D.C. White, J. Bacteriol., 179, 4219 (1997). J.L. Ramos, E. Duque, J. Rodriguez-Herva, P. Godoy, A. Haidour, F. Reyes, A. Fernandez-Barrero, J. Biol. Chem., 272, 3887 (1997). M.C. Antunes-Madeira, V.M.C., Madeira, Biochim. Biophys. Acta, 778, 49 (1984). P. Mishra, S. Kaur, Appl. Microbiol. Biotechnol., 34, 697 (1991). L.O. Ingram, J. Bacteriol., 125, 670 (1976). K. Uchida, Agric. Biol. Chem., 39, 1515 (1975). L.O. Ingram, Appl. Environ. Microbiol., 33, 1233 (1977). V. Jarlier, H. Nikaido, FEMS Microbiol. Lett., 123, 11 (1994). Y.S. Park, H.N. Chang, B.H. Kim, Biotechnol. Lett., 10, 261 (1988). D. Molenaar, T. Bolhuis, T. Abee, B. Poolman, W.N. Konings, J. Bacteriol., 174, 3118 (1992). H. Bettmann, H.J. Rehm, Appl. Microbiol. Biotechnol., 20, 285 (1984). J.W. Costerton, J. Ind. Microbiol. Biotechnol., 22, 551 (1999). J.W. Costerton, Int. J. Antimicrobial Agents, 11, 217 (1999). K. Frensch, H.F. Scholer, D. Schoenen, Zbl. Hyg. Umweltmed., 190, 72 (1990). W. Sand, Int. Biodeterioration Biodegradation, 40, 183 (1997). K. Wallhäußer, Praxis der Sterilisation, Desinfektion, Konservierung, Thieme Verlag, Stuttgart, 1998. G. Bitton, B. Koopman, Rev. Environ. Contamin. Toxicol., 125, 1 (1992). K.L.E. Kaiser, J.M. Ribo, Tox. Assess, 3, 195 (1988). B.J. Dutka, K.K. Kwan, Tox. Assess, 3, 303 (1988). M.H. Schiewe, E.G. Hawk, D.I. Actor, M.M. Krahn, Can. J. Fish Aquat., 42, 1244 (1985). O. Favre-Bulle, J. Schouten, J. Kingma, B. Witholt, Biotechnology, 9, 367, 1991. C. Laane, R. Boeren, R. Hilhorst, C. Veeger, in C. Laane, J. Tramper, M.D. Lilly, Eds., Biocatalysis in organic media, Elsevier Publishers, B.V., Amsterdam, 1987. C. Adami, R. Kummel, Gefahrstoffe Reinhaltung Luft, 57, 365 (1997). A.R. Pedersen, S. Moller, S. Molin, E. Arvin, Biotechnol. Bioengn., 54, 131 (1997). P.J. Hirl, R.L. Irvine, Wat. Sci. Technol., 35, 49 (1997). M.W. Fitch, D. Weissman, P. Phelps, G. Georgiou, Wat. Res., 30, 2655 (1996). S.V. Ley, F. Sternfeld, S.C. Taylor, Tetrahedron Lett., 28, 225 (1987). G.M. Whited, W.R. McComble, L.D. Kwart, D.T. Gibson, J. Bacteriol., 166, 1028 (1986). M.R. Smith, J. Bacteriol., 170, 2891 (1990)
872
Tsuneo Yamane
14.4.3 CHOICE OF SOLVENT FOR ENZYMATIC REACTION IN ORGANIC SOLVENT
Tsuneo Yamane Graduate School of Bio- and Agro-Sciences, Nagoya University, Nagoya, Japan
14.4.3.1 Introduction The ability of enzymes to catalyze useful synthetic biotransformations in organic media is now beyond doubt. There are some advantages in using enzymes in organic media as opposed to aqueous medium, including 1) shifting thermodynamic equilibrium to favor synthesis over hydrolysis, 2) reduction in water-dependent side reaction, 3) immobilization of the enzyme is often unnecessary (even if it is desired, merely physical deposition onto solid surfaces is enough), 4) elimination of microbial contamination, 5) suitable for reaction of substrates insoluble and/or unstable in water, etc. Here organic media as the reaction system are classified into two categories: substrates dissolved in neat organic solvents and solvent-free liquid substrates. Although the latter seems preferable to the former, if it works, there are a number of cases where the former is the system of choice: for example, when the substrate is solid at the temperature of the reaction, when high concentration of the substrate is inhibitory for the reaction, when the solvent used gives better environment (accelerating effect) for the enzyme, and so forth. Prior to carrying out an enzymatic reaction in an organic solvent, one faces choice of a suitable solvent in the vast kinds of organic solvents. From active basic researches having been carried out in the past two decades, there has been a remarkable progress in our understanding of properties of enzymes in organic media, and in how organic solvents affect them. Some researchers call the achievement ‘medium engineering’. A comprehensive monograph was published in 1996 reviewing the progress.1 In this article organic solvents often used for enzymatic reactions are roughly classified, followed by influence of solvent properties on enzymatic reactions and then properties of the enzymes affected by the nature of the organic solvents are briefly summarized. 14.4.3.2 Classification of organic solvents Among numerous kinds of organic solvents, the ones often used for enzymatic reactions are not so many, and may be classified into three categories (Table 14.4.3.1),2 in view of the importance of water content of the organic solvent concerned (see Section 14.4.3.3). 1) water-miscible organic solvents These organic solvents are miscible with water at the temperature of the reaction. Any cosolvent system having 0 - 100% ratio of the solvent/water can be prepared from this kind of solvent. Note that some organic solvents having limited water solubility at ambient temperature, and hence are not regarded as water-miscible, become miscible at elevated temperature. 2) Water-immiscible organic solvents
14.4.3 Choice of solvent for enzymatic reaction
873
Table 14.4.3.1. Classification of solvents commonly used for enzymatic reactions in organic media. [Adapted, by permission, from T. Yamane, Nippon Nogeikagaku Kaishi, 65, 1104(1991)] 1) Water-miscible organic solvents Methanol, ethanol, ethylene glycol, glycerol, N,N’-dimethylformamide, dimethylsulfoxide, acetone, formaldehyde, dioxane, etc. 2) Water-immiscible organic solvents (water solubility [g/l] at the temperature indicated)
•
alcohols (n-, iso-) propyl alcohol, (n-, s-, t-) butyl alcohol, (n-, s-, t-) amyl alcohol, n-octanol, etc. • esters methyl acetate, ethyl acetate (37.8, 40oC), n-butyl acetate, hexyl acetate, etc. • alkyl halides dichloromethane (2, 30oC), chloroform, carbon tetrachloride, trichloroethane (0.4, 40oC), etc. • ethers diethyl ether (12, 20oC), dipropyl ether, diisopropyl ether, dibutyl ether, dipentyl ether, etc. 3) Water-insoluble organic solvents (water solubility [ppm] at the temperature indicated)
• • •
aliphatic hydrocarbons n-hexane (320, 40oC), n-heptane (310, 30oC), isooctane (180, 30oC), etc. aromatic hydrocarbons benzene (1200, 40oC), toluene (880, 30oC), etc. allicyclic hydrocarbons cyclohexane (160, 30oC), etc.
These organic solvents have noticeable but limited solubility of water, ranging roughly 0.1 - 10 % its solubility. The water solubility is of course increased as the temperature is raised. 3) Water-insoluble organic solvents These solvents are also water-immiscible and have very low water solubility so that they are regarded as water-insoluble, i.e., water is practically insoluble in the organic solvents. Most aliphatic and aromatic hydrocarbons belong to this category. In Table 14.4.3.1 organic solvents often used for enzymatic reactions are listed together with their water solubilities (although not for all of them). 14.4.3.3 Influence of solvent parameters on nature of enzymatic reactions in organic media Factors that influence the activity and stability of enzymes in organic media have been mostly elucidated. Several of them are mentioned below. 1) Water activity, aw Among factors, the amount of water existing in the reaction system is no doubt the most influential. To emphasize the effect of water, the author once proposed to say ‘enzymatic reaction in microaqueous organic solvent’, instead of merely say ‘enzymatic reaction in organic solvent’.3-5
874
Tsuneo Yamane
Very trace amount of water or nearly anhydrous state renders practically no enzymatic reaction. In this context, it should be reminded that commercially available enzyme preparations, or the enzymes even after lyophilization or other drying procedures, contain some water bound to the enzyme proteins. Whereas, excess water in the reaction system results in hydrolysis of the substrate, which is often unfavorable side reaction, giving rise to lower yield of product. Thus, there exist usually the optimal water content for each enzymatic reaction of concern. Water molecules in the microaqueous system exist in three different states: 1) water bound to the enzyme protein, 2) water dissolved freely in the solvent (plus dissolved substrate), and 3) water bound to impurities existing in the enzyme preparation or bound to the support materials if immobilized enzyme particles are used. Therefore, the following equation with respect to water holds: Total water = (Water bound to the enzyme) + (Water dissolved in the solvent)
[14.4.3.1]
+ (Water bound to the immobilization support or to impurities of the enzyme preparation)
Water affecting most of the catalytic activity of the enzyme is the one bound to the enzyme protein.6 From the above equation, it can be well understood that the effect of water varies depending on the amount of enzyme used and/or its purity, kind of solvent, and nature of immobilization support, etc. as far as the total water content is used as the sole variable. Also, it is often asked what is the minimal water content sufficient for enzymatic activity? It should be recognized that a relation between the degree of hydration of the enzyme and its catalytic activity changes continuously. There exist a thermodynamic isotherm-type equilibrium between the protein-bound water and freely dissolved water, and its relationship is quite different between water-miscible and water-insoluble solvents.4 A parameter better than the water content, water activity, aw, was proposed to generalize the degree of hydration of a biocatalyst in organic media.7 aw is a thermodynamic parameter which determines how much water is bound to the enzyme and in turn decides the catalytic activity to a large extent among different kinds of the organic solvents. aw is especially useful when water-insoluble organic solvent is used because the precise water content is hard to be measured due to its low solubility. It was shown that profiles between aw Figure 14.4.3.1 Activity of Lipozyme catalyst as a function of water activity in a range of solvents. [Adapted, by and the reaction rate were similar when the permission, from R.H. Valivety, P.J. Halling and A.R. same reaction was carried out in different Macrae, Biochim. Biophys. Acta, 1118, 221 (1992)]. solvents at varying water contents.8 In dif-
14.4.3 Choice of solvent for enzymatic reaction
875
ferent solvents, maximum reaction rate was observed at widely different water content, but if water content was qualified in terms of aw, the optimum was observed at almost the same aw (Figure 14.4.3.1). However, as seen from Figure 14.4.3.1, the profile does not lay on a single curve, and the absolute optimal reaction rate varies depending on the kind of the solvent, implying that aw is not almighty. 2) Hydrophobicity (or polarity), logP A hydrophobicity parameter, logP, was first proposed for microbial epoxidation of propene and 1-butene.9,10 logP is the logarithm of P, where P is defined as the partition coefficient of a given compound in the standard n-octanol/water two phase: P=
solubility of a given compound in n − octanol phase solubility of a given compound in water phase
[14.4.3.2]
Laane et al.10 concluded as a general rule that biocatalysis in organic solvents is low in polar solvents having a logP < 2, is moderate in solvents having a logP between 2 and 4, and is high in apolar solvents having a logP > 4. They also stated that this correlation between polarity and activity paralleled the ability of organic solvents to distort the essential water layer bound to the enzyme that stabilized the enzyme. Since logP can easily be determined experimentally, or be estimated from hydrophobic fragmental constants, many biotechnologists have tried since then to correlate effects of organic solvents on biocatalysts they studied with logP approach. Their results have been successful, not completely but only partially. A number of exceptions to the ‘logP rule’ have been in fact reported. 3) Dielectric constant (or dipole moment), ε (or D) Interactions between an enzyme and a solvent in which the enzyme is suspended are mostly non-covalent ones as opposed to interactions in water. These strong non-covalent interactions are essentially of electrostatic origin, and thus according to Coulomb’s law, their strength is imposed dependent on the dielectric constant, ε, (which is higher for water than for almost all organic solvents). It is likely that enzymes are more rigid in anhydrous solvents of low ε than in those of high ε. Thus, ε of a solvent can be used as a criterion of rigidity of the enzyme molecule. For the enzyme to exhibit its activity, it must be dynamically flexible during its whole catalytic action so that its activity in a solvent of lower ε should be less than in a solvent of higher ε. On the other hand, its selectivity or specificity becomes higher when its flexibility decreases so that the selectivity in a solvent of lower ε should be higher than in a solvent of higher ε. 14.4.3.4 Properties of enzymes affected by organic solvents 1) Thermal stability (half-life), t1/2 Stability of an enzyme in an organic solvent is estimated by its half-life, t1/2, when its activity is plotted as a function of incubation time. Although, t1/2 during the enzymatic reaction is more informative for practical purposes, t1/2 under no substrate is often reported because of its easiness of measurement. Inactivation of an enzyme is caused mostly by change in its native conformation, or irreversible unfolding of its native structure. Water, especially enzyme-bound water makes a major contribution to the protein folding through van der Waals interaction, salt-bridges, hydrogen bonds, hydrophobic interaction, etc. When the enzyme molecule is put into organic solvent, water molecules bound to the enzyme molecule are more or less re-equilibrated, depending on the free water content. Therefore, both the nature of organic solvent and the free water content have profound effects on its stability.
876
Tsuneo Yamane
It has been shown that a number of enzymes suspended in anhydrous (dry) organic solvents exhibit thermal stability far superior to that in aqueous solutions (Table 14.4.3.2).11 This is because most of chemical processes that occur in the thermal inactivation involve water, and therefore do not take place in a water-free environment. Furthermore, increased rigidity in dry organic solvents hinders any unfolding process. The increased thermal stability in the dry organic solvent drops down to the stability in aqueous solution by adding small amount of water as demonstrated by Zaks and Klibanov12 and others. Thus, thermal stability of the enzyme in organic solvent strongly depends on its free water content. Table 14.4.3.2. Stability of enzymes in non-aqueous vs. aqueous media. [Adapted, by permission, from ‘Enzymatic Reactions in organic Media’, A. M. P. Koskinen and A. M. Klibanov, Blackie Academic & Professional (An Imprint of Chapman & Hall), Glasgow, 1996, p. 84] Enzyme
Conditions
Thermal property
References
PLL
tributyrin aqueous, pH 7.0
t1/2 < 26 h t1/2 < 2 min
Zaks & Klibanov (1984)
Candida lipase
tributyrin/heptanol aqueous, pH 7.0
t1/2 = 1.5 h t1/2 < 2 min
Zaks & Klibanov (1984)
Chymotrypsin
octane, 100oC aqueous, pH 8.0, 55oC
t1/2 = 80 min t1/2 =15 min
Zaks & Klibanov (1988) Martinek et al. (1977)
Subtilisin
octane, 110oC
t1/2 = 80 min
Russell & Klibanov (1988)
o
Lysozyme
cyclohexane, 110 C aqueous
t1/2 = 140 h t1/2 < 10 min
Ahen & Klibanov (1986)
Ribonuclease
nonane, 110oC, 6 h aqueous, pH 8.0, 90oC
95% activity remains t1/2 < 10 min
Volkin & Klibanov (1990)
F1-ATPase
toluene, 70oC aqueous, 70oC
t1/2 > 24 h t1/2 < 10 min
Garza-Ramos et al. (1989)
Alcohol dehydrogenase
heptane, 55oC
t1/2 > 50 days
Kaul &Mattiasson (1993)
o
HindIII
heptane, 55 C, 30 days
no loss of activity
Kaul & Mattiasson (1993)
Lipoprotein lipase
toluene, 90oC, 400 h
40% activity remains
Ottoline et al. (1992)
β-Glucosidase
2-propanol, 50oC, 30 h
80% activity remains
Tsitsimpikou et al. (1994)
Tyrosinase
chloroform, 50oC aqueous solution, 50oC
t1/2 = 90 min t1/2 = 10 min
Yang & Robb (1993)
Acid phosphatase
hexadecane, 80oC aqueous, 70oC
t1/2 = 8 min t1/2 = 1 min
Toscano et al. (1990)
Cytochrome oxidase
toluene, 0.3% water toluene, 1.3% water
t1/2 = 4.0 h t1/2 = 1.7 min
Ayala et al. (1986)
For References, refer to Ref. 11.
14.4.3 Choice of solvent for enzymatic reaction
877
2) Specificity and selectivity, kcat/Km It is the most exciting and significant feature that the substrate specificity, enantioselectivity and regioselectivity can be profoundly affected by nature of solvents in which the enzyme molecule exists. This phenomenon has opened an alternative approach for changing specificity and selectivity of an enzyme other than both screening from nature and protein engineering in the field of synthetic organic chemistry. The ability of enzymes to discriminate substrate specificity among different, but structurally similar substrates, enantioselectivity among enantiomers, eantiofaces or identical functional groups linked to a prochiral center, and regioselectivity among identical functional groups on the same molecule, is expressed quantitatively on E value, which is the ratio of the specificity constants, kcat/Km, for the two kinds of substrate (or entiomers), i.e., (kcat/Km)1/(kcat/Km)2. For kinetic resolution of racemic mixture by the enzyme, E is called enantiomeric ratio. The higher E, the higher the enantiomeric excess (i.e., the optical purity), ee, of the product (or remaining substrate). It is said that an E value higher than 100 is preferable for pharmaceutical or biotechnological applications. For overview of this topic, see Refs. 13 and 14. (2a) Substrate specificity Zaks and Klibanov reported that the substrate specificity of α-chymotrypsin, subtilisin, and esterase changed with an organic solvent.15 The substrate specificity in octane was reversed compared to that in Figure 14.4.3.2. The dependence of (A) subtilisin water. A thermodynamical model that preCarlsberg and (B) a-chymotrypsin substrate specificity dicted the substrate specificity of subtilisin for substrates 1 and 2 on the ratio of their Raoult’s law ac- Carlsberg and α-chymotrypsin in organic tivity coefficients. For the structures of the substrates 1 media on the basis of specificity of the enand 2, and the solvents a through m in (A) and a to g in (B), refer to Ref. 16. [Adapted, by permission, from C.R. zyme in water and physicochemical charWescott and A.M. Klibanov, Biotechnol. Bioeng., 56, acteristics of the solvents was developed 343(1997)]. by Wescott and Klibanov.16 They determined kcat/Km for the transesterification of N-acetyl-L-phenylalanine and N-acetyl-L-serine with propanol in 20 anhydrous solvents, and correlated the data of (kcat/Km)Ser/(kcat/Km)Phe, first with the solvent to water partition coefficients for the substrate, PPhe/PSer. Later they examined the selectivity of subtilisin toward two different substrates with the Raoult’s law activity coefficients, γ, by the following equation:15
878
Tsuneo Yamane
log[(k cat / K m )1 / (k cat / K m ) 2 ] = log(γ 1 / γ 2 ) + constant
[14.4.3.3]
The correlation was unexpectedly high as seen in Figure 14.4.3.2,15 implying that the change of substrate specificity of enzyme in organic solvent stems to a large extent from the energy of desolvation of the substrate. (2b) Enantioselectivity Changes in enantioselectivity in various organic solvents was first discovered by Sakurai et al.17 Later Fitzpatrick and Klibanov studied enantioselectivity of subtilisin, Carsberg in the transesterification between the sec-phenethyl alcohol (a chiral alcohol) and vinyl butyrate to find that it was greatly affected by the solvent. Only the correlations with ε or with D gave good agreements.18 The enzyme enantioselectivity was inversed by changing solvents.19 Nakamura et al. studied lipase (Amano AK from Pseudomonas sp.)-catalyzed transesterification of cis- and trans-methylcycolhexanols with vinyl acetate in various organic solvents, and investigated the effect of solvent on activity and stereoselectivity of the lipase.20 They correlated their stereoselectivity with good linearity (except for dioxane and dibutyl ether) by the following two-parameter equation (Figure 14.4.3.3): E = a(ε − 1) / (2ε + 1) + bVm + c
[14.4.3.4]
where ε and Vm are dielectric constant and molar volume of the solvent, respectively, and a, b, and c are constants which should be experimentally determined. Bianchi et al. also reported that for the resolution of antitussive agent, Dropropizine, using both hydrolysis in aqueous buffer and transesterification techniques in various organic solvents, by a lipase (Amano PS from Pseudomonas cepacia), E depended very much on organic solvents (Table 14.4.3.3), with the highest E value (589) in n-amyl alcohol and the lowest one (17) in water.21 In this case, however, there was no correlation between the enantioselectivity and the Figure 14.4.3.3. Linear relationship between f(ε, Vm) and E for physico-chemical properties of lipase-catalyzed transesterification of cis- and trans-4-methylthe solvents such as logP or ε. cyclohexanols with vinyl acetate in various organic solvents. For the organic solvents 1 through 18, refer to Ref. 20. [Adapted, by permission, from K. Nakamura, M. Kinoshita and A. Ohno, Tetrahedron, 50, 4686(1994)].
14.4.3 Choice of solvent for enzymatic reaction
879
Table 14.4.3.3. Effect of the solvent on enantioselectivity of lipase PS. [Adapted, by permission, from D. Bianchi, A. Bosetti, P. Cesti and P. Golini, Tetrahedron Lett., 33, 3233(1992)]. Solvent
logP
H 2O
ε
Nucleophile
E
78.54
H2O
17
hexane
3.5
1.89
n-propanol
146
CCl4
3.0
2.24
n-propanol
502
toluene
2.5
2.37
n-propanol
120
iso-propyl ether
1.9
3.88
n-propanol
152
2-methyl-2-butanol
1.45
5.82
n-propanol
589
2-methyl-2-butanol
1.45
5.82
H2O
63
n-propanol
0.28
20.1
n-propanol
181
acetonitrile
-0.33
36.2
n-propanol
82
1,4-dioxane
-1.1
2.2
n-propanol
164
As mentioned above, no correlation was reported in a large number of articles on the effects of solvents on the enzyme specificity/selectivity, although correlations between the specificity/selectivity and physico-chemical properties of the solvents were successful in some combinations of an enzymatic reaction for a set of solvents. Therefore, attempts to rationalize the phenomena based on either physico-chemical properties of the solvents or on their structure, are at present clearly unsatisfactory from the point of view of predictable value. Further experiments carried out under more strictly defined condition are necessary to reach the quantitative explanation of the whole phenomena. 14.4.3.5 Concluding remarks Activity, stability, and selectivity of an enzyme are affected considerably by nature of organic solvents as well as free water content in the enzyme-catalyzed reaction in the organic solvents. However, rational criteria for the selection of a proper solvent among vast variety of the organic solvents are very limited so far. Researchers are obliged to resort to empirical approach by examining some kinds of solvent for the enzyme-catalyzed reaction concerned at the present state of art. References 1 2 3 4 5 6 7 8 9 10
A.M.P. Koskinen and A.M. Klibasnov, Ed., Enzymatic Reactions in Organic Media, Blackie Academic & Professional (An Imprint of Chapman & Hall), London, 1996. T. Yamane, Nippon Nogeikagaku Kaishi, 65, 1103(1991). T. Yamane, J. Am. Oil Chem. Soc., 64, 1657(1987). T. Yamane, Y. Kozima, T. Ichiryu and S. Shimizu, Ann. New York Acad. Sci., 542, 282(1988). T. Yamane, Biocatalysis, 2, 1(1988). J.A. Rapley, E. Gratton and G. Careri, Trends Biol. Sci., Jan., 18(1983). P.J. Halling, Biochim. Biophys. Acta, 1040, 225(1990). R.H. Valivety, P.J. Halling, A.R. Macrae, Biochim. Biophys. Acta, 1118, 218(1992). C. Laane, S. Boeren and K. Vos, Trends Biotechnol., 3, 251(1985). C. Laane, S. Boeren, K. Vos and C. Veeger, Biotechnol. Bioeng., 30, 81(1987).
880
11 12 13 14 15 16 17 18 19 20 21
George Wypych A. Zaks, in Enzymatic Reactions in Organic Media ed. by A.M.P. Koskinen and A.M. Klibanov, Blackie Academic & Professional (An Imprint of Chapman & Hall), London, 1996, p. 84. A. Zaks and A.M. Klibanov, Science, 224, 1249(1984). C.R. Wescott and A.M. Klibanov, Biocheim. Biophys. Acta, 1206, 1(1994). G. Carrea, G. Ottolina and S. Riva, Trends Biotechnol., 13, 63(1995). A. Zaks and A.M. Klibanov, J. Am. Chem. Soc., 108, 2767(1986). C.R. Wescott and A.M. Klibanov, J. Am. Chem. Soc., 108, 2767(1986). T. Sakurai, A.L. Margolin, A.J. Russell and A.M. Klibanov, J. Am. Chem. Soc., 110, 7236(1988). P.A. Fitzpatrick and A.M. Klibanov, J. Am. Chem. Soc., 113, 3166(1991). S. Tawaki and A.M. Klibanov, J. Am. Chem. Soc., 114, 1882(1992). K. Nakamura, M. Kinoshita and A. Ohno, Terehedron, 50, 4681(1994). D. Bianchi, A. Bosetti, P. Cesti and P. Golini, Tetrahedron Lett., 33, 3231(1992).
14.5 COIL COATING George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The coil coating industry is under pressure to eliminate the use of solvents. Polyester coil coatings contain up to 40% of solvents such as glycol esters, aromatic hydrocarbons (e.g., Solvesso 150), alcohols, ketones, and butyl glycol.1,2 A recent book1 predicted that the solvent-based technology will not change during the next decade because the industry heavily invested in equipment to deal with solvents. Such changes in technology require long testing before they can be implemented. The coil coating industry normally recovers energy from evaporated solvents either by at-source incineration or by a recycling process which lowers emissions. Because of the large amount of solvents used, the use of PVC and fluorocarbon resins in some formulations, and the use of chromates in pretreatments the pressure remains on the industry to make improvements.3 The coil coating industry is estimated to be consuming about 50,000 tons of solvents both in Europe and in the USA.1 About half of these solvents are hydrocarbons. According to the published studies,3,4 efforts to change this situation did start in the early 1990s and by mid nineties research data were available to show that the technology can be changed. Two directions will most likely challenge the current technology: radiation curing and powder coating. Coil coats are thin (about 30 µm wet thickness) but contain a high pigment loading. Consequently, UV curing is less suitable than electron beam curing. The application of this technology requires a change to the polymer system and acrylic oligomers are the most suitable for this application. This system can be processed without solvents. If a reduction of viscosity is required, it can be accomplished by the use of plasticizers (the best candidates are branched phthalates and linear adipates) and/or reactive diluents such as multifunctional monomers. Results3 show that the UV stability of the system needs to be improved by using a polyester top coat or fluoropolymer. With top coat, material performs very well as learned from laboratory exposures and exposures in industrial environment.3 At the time of the study (about 6 years ago), process of coating was less efficient than solvent-based system because production speed was about 6 times slower than the highest production rates in the industry (120 m/min). At the same time, it is known2 that the quality of solvent-based coatings suffers from excessive production rates as well. Radiation curing has a disadvantage because of its high capital investment but it does have an economical advantage because the
14.6 Cosmetics and personal care products
881
process is very energy efficient. Previous experiences with radiation curing technology show that the process has been successfully implemented in several industries such as paper, plastic processing, and wood coating where long term economic gains made the changes viable. Comparison of solvent-based fluoropolymer and fluorocarbon powder coating developed in Japan4 shows that elimination of solvent is not only good for environment but also improves performance (UV stability especially is improved). The study was carried out with a very well designed testing program to evaluate the weathering performance of the material. These two technologies show that there is extensive activity to improve coil coatings with simultaneous elimination of solvents. Two recent patents contribute more information on the developments in the coil coating industry.5,6 One problem in the industry is with the poor adhesion of the coating to steel.2,3 A primer developed contains dipropylene glycol methyl ether and PM acetate which allows the deposition of relatively thick layers (20-40 µm) without blistering and at suitable rate of processing. However, the primer has a low solids content (30-45%).5 A new retroreflective coating was also developed6 which is based on ethyl acrylate-styrene copolymer and contains a mixture of xylene with another aromatic hydrocarbon (Solvesso 150) at relatively low concentration (11-12%).
REFERENCES 1 2 3 4 5 6
B P Whim, P G Johnson, Directory of Solvents, Blackie Academic & Professional, London, 1996. A L Perou, J M Vergnaud, Polym. Testing, 16, No.1, 19-31 (1997). G M Miasik, Surface Coatings International, 79, No.6, 258-67 (1996). C Sagawa, T Suzuki, T Tsujita, K Maeda, S Okamoto, Surface Coatings International, 78, No.3, 94-8 (1995). M T Keck, R J Lewarchik, J C Allman, US Patent 5,688,598, Morton International, Inc., 1997. G L Crocker, R L Beam, US Patent 5,736,602, 1998.
14.6 COSMETICS AND PERSONAL CARE PRODUCTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada Several cosmetic products contain solvents. These include nail polish, nail polish remover, fragrances, hair dyes, general cleaners, hair sprays and setting lotions. In most cases, ethanol is the only solvent. Nail polish and nail polish remover contain a large variety of solvents. Several recent patents3-6 give information on current developmental work in this area. Nitrocellulose, polyester, acrylic and methacrylic ester copolymer, formaldehyde resin, rosin, cellulose acetate butyrate are the most frequently used polymers in nail polish formulations. Solvents were selected to suit the polymer used. These include acetone, methyl acetate, ethyl acetate, butyl acetate, methyl glycol acetate, methyl ethyl ketone, methyl isobutyl ketone, toluene, xylene, isopropyl alcohol, methyl chloroform, and naphtha. Solvents constitute a substantial fraction of the composition usually around 70%. Reformulation is ongoing to improve the flexibility and durability of the nail polish.3 Other efforts are directed to improve antifungal properties,4 to eliminate ketones and formaldehyde resin (ketones because of their toxicity and irritating smell and formaldehyde resins because they contribute to dermatitis),5 and elimination of yellowing.6 All efforts are di-
882
George Wypych
rected towards improvements in drying properties and adhesion to the nails. These properties are partially influenced by solvent selection. The current trend is to greater use of ethyl acetate, butyl acetate, naphtha, and isopropanol which are preferable combinations to the solvents listed above. Acetone used to be the sole component of many nail polish removers. It is still in use but there is a current effort to eliminate the use of ketones in nail polish removers. The combinations used most frequently are isopropanol/ethyl acetate and ethyl acetate/isopropanol/ 1,3-butanediol. General cleaners used in hairdressing salons contain isopropanol and ethanol. Hair spray contains ethanol and propellants which are mixtures of ethane, propane, isobutane, and butane. The reported study2 of chemical exposure in hairdresser salons found that although there were high concentrations of ethanol the detected levels were still below the NIOSH limit. The concentrations were substantially higher in non-ventilated salons (about 3 times higher) than those measured in well ventilated salons. Small concentrations of toluene were found as well, probably coming from dye components. Recent patents7,8 show that solvents may enter cosmetic products from other ingredients, such as components of powders and thickening agents. Some solvents such as dichloromethane and benzene, even though they are present is smaller quantities are reason for concern.
REFERENCES 1 2 3 4 5 6 7 8
B P Whim, P G Johnson, Directory of Solvents, Blackie Academic & Professional, London, 1996. B E Hollund, B E Moen, Ann. Occup. Hyg., 42 (4), 277-281 (1998). M F Sojka, US Patent 5,374,674, Dow Corning Corporation. 1994. M Nimni, US Patent 5,487,776, 1996. F L Martin, US Patent 5,662,891, Almell Ltd., 1997. S J Sirdesai, G Schaeffer, US Patent 5,785,958, OPI Products, Inc., 1998. R S Rebre, C Collete, T. Guertin, US Patent 5,563,218, Elf Atochem S. A., 1996. A Bresciani, US Patent 5,342,911, 3V Inc., 1994.
14.7 Dry cleaning
883
14.7 DRY CLEANING - TREATMENT OF TEXTILES IN SOLVENTS Kaspar D. Hasenclever Kreussler & Co.GmbH, Wiesbaden, Germany
Most processes in manufacturing and finishing of textiles are aqueous. In order to prevent water pollution, some years ago developments were made to transform dyeing-, cleaningand finishing-processes from water to solvents. During 1970-1980 solvent processes for degreasing, milling, dyeing and waterproofing of textiles could get limited economic importance. All these processes were done with tetrachloroethylene (TCE). After getting knowledge about the quality of TCE penetrating solid floors, stone and ground, contaminating groundwater, this technology was stopped. Today the main importance of solvents in connection with textiles is given to dry cleaning, spotting and some special textile finishing processes. 14.7.1 DRY CLEANING 14.7.1.1 History of dry cleaning Development of dry cleaning solvents The exact date of discovery of dry cleaning is not known. An anecdote tells us that in about 1820 in Paris, a lamp filled with turpentine fell down by accident and wetted a textile. After the turpentine was vaporized, the wetted areas of the textile were clean, because the turpentine dissolved oily and greasy stains from it. In 1825, Jolly Belin founded the first commercial “dry laundry” in Paris. He soaked textile apparel in a wooden tub filled with turpentine, cleaned them by manual mechanical action, and dried them by evaporating the turpentine in the air. After getting the know-how to distill benzene from tar of hard coal in 1849, this was used as a solvent for dry cleaning because of its far better cleaning power. But benzene is a strong poison, so it was changed some decades later to petrol, which is explosive. In order to reduce this risk, petrol as dry cleaning solvent was changed to white spirit (USA: Stoddard solvent) with a flash point of 40 - 60°C (100 - 140°F) in 1925. The flammability of the hydrocarbon solvents in dry cleaning plants was judged to be risky because of fire accidents. After finding the technology for producing inflammable chlorinated hydrocarbons, trichloroethane and tetrachloroethylene (TCE) were introduced in dry cleaning since about 1925. These solvents gave the opportunity for good cleaning results and economic handling. Up to 1980, TCE was the most important solvent for dry cleaning worldwide. Compared to TCE, fluorinated chlorinated hydrocarbons (CFC) offer benefits to dry cleaning because of their lower boiling points and their more gentle action to dyestuffs and fabrics. So since 1960 these solvents have had some importance in North America, Western Europe, and the Far East. They were banned because of their influence on the ozone layer in the stratosphere by the UNESCO’s Montreal Protocol in 1985. At the same time, TCE was classified as a contaminant to groundwater and as a dangerous chemical to human health with the possible potential of cancerogenic properties. As a result of this, hydrocarbon solvents on the basis of isoparaffins with a flash point higher
884
Kaspar D. Hasenclever
than 55°C (130°F) were at first used in Japanese and German dry cleaning operations since 1990. Also, alternatives to conventional dry cleaning were developed. Wet cleaning was introduced by Kreussler in 1991 (Miele System Kreussler) and textile cleaning in liquid carbon dioxide was exhibited by Global Technologies at the Clean Show Las Vegas in 1997. Development of dry cleaning machines In the beginning, dry cleaning was done manually in wooden tubs filled with turpentine or benzene. About 1860, the Frenchman Petitdidier developed a wooden cylindrical cage, which was rotated in a tub filled with solvent. The apparel to be dry cleaned was brought into the cage and moved through the solvent by the rotation of the cylinder. This machine got the name “La Turbulente”. The next step was the addition of a centrifuge to the wooden machine. The dry cleaned apparel was transferred from the machine into the centrifuge and then dried by vaporizing the solvent in the open air. About 1920, a tumble dryer was used for drying the dry cleaned textiles. Fresh air was heated up, blown through the dryer, where the air was saturated with solvent vapor and then blown out to be exhausted into the environment. The used solvent was cleaned to be recycled by separation of solid matter by centrifugal power with a separator and to be cleaned from dissolved contamination by distilling. In 1950, dry to dry machines were developed by Wacker in Germany for use with TCE. The principle of working is as follows: The cage, filled with textiles, is rotating in a closed steel cylinder. The dry cleaning solvent is pumped from the storage tank into the cylinder so that the textiles in the cage are swimming in the solvent. After ending the cleaning process, the solvent is pumped back into the storage tank and the cage rotates with high speed (spinning) in order to separate the rest of the solvent from the textile. Then air is circulated through a heat exchanger to be heated up. This hot air is blown into the rotating cage in order to vaporize the remaining solvent from the textiles. The air saturated with solvent vapor is cleaned in a condenser, where the solvent is condensed and separated from the air. The air now goes back into the heat exchanger to be heated up again. This circulation continues until textiles are dry. The separated solvent is collected in a tank to be reused. In order to reduce solvent losses and solvent emissions, since 1970 charcoal filters have been used in the drying cycle of dry cleaning machines, so that modern dry cleaning systems are separated from the surrounding air. 14.7.1.2 Basis of dry cleaning Dry cleaning means a cleaning process for textiles, which is done in apolar solvents instead of water. If water is used, such cleaning process is called “washing” or “laundering”. Natural textile fibers, such as wool, cotton, silk and linen, swell in water because of their tendency to absorb water molecules in themselves. This causes an increase of their diameter and a change of the surface of yarns and fabrics. The result is shrinkage, felting and creasing. Apolar solvents, such as hydrocarbons, are not absorbed by natural textile fibers because of the high polarity of the fibers. So there is no swelling, no shrinkage, no felting and no creasing. From the solvent activity, dry cleaning is very gentle to textiles, with the result that the risk of damaging garments is very low. Because of the apolar character of dry cleaning solvents cleaning activity also deals with apolar contamination. Oils, fats, grease and other similar substances are dissolved in
14.7 Dry cleaning
885
the dry cleaning operation. Polar contamination, such as, salt, sugar, and most nutrition and body excrements are not dissolved. In the same way that water has no cleaning activity with regard to oils, fat, or grease, in dry cleaning there is no activity with regard to salts, sugar, nutrition and body excrements. And in the same way that water dissolves these polar substances, cleaning of textiles from these contaminants does not pose any problems in a washing process. To make washing active to clean oils, fat and grease from textiles, soap (detergent) has to be added to water. To make dry cleaning active to clean salts, sugar and the like from textiles, dry cleaning detergent has to be added to the solvent. In washing, fresh water is used for the process. After being used, the dirty washing liquid is drained off. In dry cleaning the solvent is stored in a tank. To be used for cleaning it is pumped into the dry cleaning machine. After being used, the solvent is pumped back into the storage tank. To keep the solvent clean it is constantly filtered during the cleaning time. In addition to this, a part of the solvent is pumped into a distilling vessel after each batch to be cleaned by distilling. Dry cleaning solvents are recycled. The solvent consumption in modern machines is in the range of about 1 - 2 % per weight of the dry cleaned textiles. 14.7.1.3 Behavior of textiles in solvents and water Fibers used for manufacturing textiles can be classified into three main groups: • Cellulosic fibers: cotton, linen, rayon, acetate. • Albumin fibers: wool, silk, mohair, camelhair, cashmere. • Synthetic fibers: polyamide, polyester, acrylic. Textiles made from cellulosic fibers and synthetics can be washed without problems. Apparel and higher class garments are made from wool and silk. Washing very often bears a high risk. So these kinds of textiles are typically dry cleaned. Dependent to the relative moisture of the surrounding air fabrics absorb different quantities of water. The higher the polarity of the the fiber, the higher is their moisture content. The higher the swelling (% increase of fiber diameter) under moisture influence, the higher is the tendency of shrinkage in a washing or dry cleaning process. Table 14.7.1. Water content (%) in textile fibers dependent on relative humidity
Fiber
Relative humidity, %
Swelling, %
70
90
max.
Viscose
14.1
23.5
24.8
115
Wool
15.6
22.2
28.7
39
Silk
11.2
16.2
17.7
31
Cotton
8.1
11.8
12.9
43
Acetate
5.4
8.5
9.3
62
Polyamide
5.1
7.5
8.5
11
Acrylic
2.1
4.0
4.8
9
Polyester
0.5
0.6
0.7
0
886
Kaspar D. Hasenclever
The same absorption of water occurs if textiles are immersed in solvent in a dry cleaning machine. If the relative humidity in the air space of the cylinder of a dry cleaning machine is higher, textiles absorb more water. The water content of textiles in solvent is equal to the water content of textiles in the open air if the relative humidity is the same. If water additions are given to the dry cleaning solvent in order to intensify the cleaning effect with regard to polar contamination, then the relative humidity in the air space of the cylinder of the dry cleaning machine increases. As a result of this, the water content in the dry cleaned textiles increases, too, so that swelling begins and shrinkage may occur. Woolen fabrics are particularly sensitive to shrinkage and felting because of the scales on the surface of the wool fiber. Not only in washing, but also in dry cleaning there is a risk of shrinkage on woolen garments. This risk is higher when the dry cleaning process is influenced by water addition. If the relative humidity in the air space of a dry cleaning machine is more than 70%, shrinkage and felting may occur, if the dry cleaning solvent is tetrachloroethylene (TCE). In hydrocarbon solvents wool is safe up to 80% relative humidity, because of the lower density of hydrocarbons compared to TCE, which reduces mechanical action. 14.7.1.4 Removal of soiling in dry cleaning “Soiling” means all the contamination on textiles during their use. This contamination is of very different sources. For cleaning purposes, the easiest way of classification of “soiling” is by solubility of soiling components. The classification can be done by definition of four groups: • Pigments: Substances are insoluble in water and in solvents. Examples are: dust, particles of stone, metal, rubber; soot, scale of skin. • Water soluble material: Examples are: salts, sugar, body excrements, sap and juice. • Polymers: Substances are insoluble in solvents but can be soaked and swell in water. Examples are: starch, albumin and those containing material such as blood, milk, eggs, sauce. • Solvent soluble material: Examples are: oil, fat, grease, wax, resins. Table 14.7.2 Average soiling of garments (apparel) in Europe Soil type
Proportion, %
Solubility
Components
Pigments
50
not
Water soluble
30
water
sugar, salt, drinks, body excretions
Polymers
10
water
starch, albumin, milk, food
Solvent soluble
10
solvents
dust, soot, metal oxides, rub-off, pollen, aerosols
skin grease, resin, wax, oils, fats
In practice, the situation is not so simple as it seems to appear after this classification. That is because soiling on garments almost always contains a mixture of different substances. For example, a spot of motor oil on a pair of trousers consists of solvent soluble oil, but also pigments of soot, metal oxides and other particles. The oil works as an adhesive for
14.7 Dry cleaning
887
the pigments, binding them to fabric. In order to remove this spot, first the oil has to be dissolved then pigments can be removed. The removal of solvent soluble “soiling” in dry cleaning is very simple. It will be just dissolved by physical action. Polymers can be removed by the combined activity of detergents, water and mechanical action. The efficiency of this process depends on the quality and concentration of detergent, the amount of water added into the system, and the operating time. Higher water additions and longer operating time increase the risk of shrinkage and felting of the textiles. Water soluble “soiling” can be removed by water, emulsified in the solvent. The efficiency here depends on the emulsifying character of the detergent and the amount of water addition. The more water is emulsified in the solvent, the higher is the efficiency of the process and the higher is the risk of shrinkage and felting of the textiles. Pigments can be removed by mechanical action and by dispersing activity of detergents. The higher the intensity of mechanical action, influenced by cage diameter, rotation, gravity of the solvent and operating time and the better the dispersing activity of the detergent, the better is the removal of pigments. The same parameters influence the care of the textiles. The better the cleaning efficiency, the higher the risk for textile damage. 14.7.1.5 Activity of detergents in dry cleaning The main component of detergents is surfactant. The eldest known surfactant is soap. Chemically soap is an alkaline salt of fatty acid. Characteristic of soap (and surfactant) is the molecular structure consisting of apolar - hydrophobic - part (fatty acid) and a polar - hydrophilic - part (-COONa), causing surface activity in aqueous solution. Surface activity has its function in the insolubility of the hydrophobic part of molecule in water and the hydrophilic part of molecule influences water solubility. This gives a tension within soap molecules in water forming layers on every available surface and forming micelles if there is a surplus of soap molecules compared to the available surface. This soap behavior stands as an example of mechanism of action of surfactants in general. Micelles of surfactants in water are formed by molecular aggregates of surfactants oriented in such a way that the hydrophobic parts are directed internally, so that the hydrophilic parts are directed outwards. In this way the aggregates form spheres, cylinders, or laminar layers, dependent on its concentration. Because of this behavior, it is possible to remove oil, fat or grease from substrates in aqueous solutions, if surfactants are present. The surfactants act to disperse the oil into small particles and build up micelles around these particles, so that oil, fat or grease incorporated inside the micelle (Figure 14.7.1). If aggregates are small, the solution is clear. If aggregates are larger, the solution (emulsion - type oil in water) becomes milky. In the same way, but in the opposite direction, surfactants form micelles in solvents (Figure 14.7.2). In this case, not the hydrophobic, but the hydrophilic part is directed internally and the hydrophobic outwards. Figure 14.7.2. Schematic Figure 14.7.1. Schematic Emulsions in this case are not formed by oil diagram of surfactants agdiagram of surfactants aggregated in micelle in solgregated in micelle in in water, but by water in oil (solvent). aqueous solution.
vent solution.
888
Kaspar D. Hasenclever
Because of this behavior, it is possible to remove water soluble (polar) material from substrates in solvents, if surfactants are at present. The surfactants disperse the polar material into very small particles and build up micelles around the particles, so that the polar material is totally incorporated inside micelle. If the aggregates are small, the solution is clear. If they are large, the solution (emulsion type water in oil) becomes milky. Drycleaning detergents consist of surfactants, cosolvents, lubricants, antistatic compounds and Figure 14.7.3. Aggregation of surfactants in emulsions of type oil in water (O/W) and water water. They are used in order to increase the cleaning efficiency, to improve the handle of dry cleaned in oil (W/O). textiles, and to prevent electrostatic charge on textiles. They are formulated as liquids in order to be easily added to the solvent by automatic dosing equipment. In dry cleaning the drycleaning detergents play the same role as soap or laundry detergents in textile washing. 14.7.1.6 Dry cleaning processes Process technology in dry cleaning has the target to clean garments as good as possible without damaging them, at lowest possible costs and with highest possible safety. Because of the environmental risks dependent on the use of TCE or hydrocarbon solvents, safe operation is the most important. Modern dry cleaning machines are hermetically enclosed, preventing infiltration of surrounding air, operating with electronic sensor systems, and they are computer controlled. The main part is the cylinder with the cleaning cage, the solvent storage tanks, the solvent recovering system, the solvent filter system and the distillation equipment (Figure 4.9.4). Cylinder/cage: A cage capacity is 20 liters per kg load capacity of the machine (about 3 gal per lb). That means, an average sized dry cleaning machine with load capacity of 25 kg (50 lbs) has a cage capacity of 500 liters (130 gal). This size is necessary because the same cage is used for cleaning and drying. During the cleaning cycle, 3 - 5 liters of solvent per kg load (0.5 - 0.8 gal/lb) are pumped into the cylinder. In a 25 kg machine, it is 80 - 125 liters (25-40 gal). This solvent is filtered during the cleaning time. Solvent storage tanks: A dry cleaning machine has 2 - 3 solvent storage tanks. The biggest - the working tank - has a capacity in liters ten times the load capacity in kg. The clean solvent tank and optional retex tank have half the capacity. All the tanks are connected to each other with an overflow pipe. The working tank with its inlet and outlet is connected to the cylinder of the machine. The inlet of the clean tank is connected to the distilling equipment, the outlet to the cylinder. Filter system: The filter is fed with solvent from the cylinder by pump pressure. The filtered solvent goes back into the cylinder. The filter has a further connection to the distilling equipment used as drain for the residue.
14.7 Dry cleaning
889
Figure 14.7.4. Scheme of a dry cleaning machine. 1 textiles to be cleaned, 2 cylinder, 3 cage, 4 fan, 5 heater, 6 drive for cage, 7 air canal, 8 cooler, 9 lint filter, 10 button trap, 11 filter, 12 distilling vessel, 13 pump, 14 solvent storage tanks, 15 water separator, 16 condenser, 17 steam pipe, 18 outlet to charcoal filter, 19 dosing unit.
Distilling equipment: The distilling vessel has the same capacity as the biggest storage tank. Solvent to be distilled is pumped into the vessel from the cylinder or from the filter. The vessel is steam or electrical heated. The solvent vapor is directed into a condenser (water cooling) and then to a water separator, from where it flows into the clean solvent tank. The residue after distillation remains in the vessel and is pumped into special residue drums. Recovering system: In order to dry the dry cleaned garments from the solvent residue which remains after spinning, a fan extracts air out of the cylinder into a condenser, where solvent vapor is condensed out of the air. From there the air is blown into a heater and directed by fan back into the cylinder. In this way the solvent is vaporized away from the garment and after condensing, reused for cleaning. Dosing equipment: In order to get the right additions of dry cleaning detergent into the solvent, a dry cleaning machine has dosing equipment working on the basis of a piston pump, which doses the right amount of detergent at the right time into the system.
890
Kaspar D. Hasenclever
Computer control: All the processes are computer controlled. Drying/Recovering, Distilling/Filtration operate according to fixed programs; the cleaning cycle can be varied with regard to the requirements of the garment. The most common program works as follows: Loading the machine, closing the door, starting computer control: 1 min pumping solvent from working tank into cylinder up to low dip level (3 l/kg); 3 min pre-cleaning by rotating the cage in the cylinder; 1 min pumping the solvent from the cylinder into the distillation vessel; 1 min spinning; 2 min pumping solvent from working tank into cylinder up to high dip level (5 l/kg); addition of detergent (2-5 ml/l of solvent); 8 min cleaning by rotating the cage in the cylinder together with filter action; 1 min pumping the solvent from the cylinder back to the working tank (the solvent in working tank is filled by overflow from clean tank); 3 min spinning 8 min drying /recovering solvent with recovery system 5 min drying/recovering solvent with charcoal filter. After the sensor has indicated that the load is free of solvent residue, opening the door and unloading the machine. After the cleaning process, the garment is controlled for cleaning quality and can be finished or if necessary, it undergoes spotting/recleaning before finishing. 14.7.1.7 Recycling of solvents in dry cleaning The recycling of solvents in dry cleaning is very important, because solvents are too expensive for single use. Three different systems are used for the cleaning of the solvent in order for it to be recycled: • Filtration • Adsorption • Distillation Filtration is a simple physical process, separating insoluble parts from the solvent. It is done during the cleaning cycle. Adsorption is mainly used together with hydrocarbon solvents, because their high boiling temperature is insufficient to separate lower boiling contaminants from the solvent. Adsorption systems use charcoal or bentonites. The solvent is pumped to filters where contaminants with higher polarity than solvent are adsorbed by the adsorbing material. The adsorbing material can adsorb contaminants in a quantity of about 20% of its own weight. After being saturated, the adsorbing material must be replaced and changed to fresh material. The charged adsorbing material is disposed according to regulations, which is cost intensive. Distillation is the best cleaning method if the boiling point of solvent is significantly lower than the boiling point of possible contaminants. With dry cleaning machines using TCE, distilling is the normal recycling method. The boiling point of TCE is 122°C, which makes steam heating possible, so that the process can be done safely and cost effectively. The distillation residue consists of removed soil and detergents. Its quantity is much lower
14.7 Dry cleaning
891
than in the adsorption systems, because there is no waste from adsorbing material. Disposal costs are much lower than with adsorption. 14.7.2 SPOTTING 14.7.2.1 Spotting in dry cleaning Spotting is the removal of stains from textiles during professional textile cleaning. Correct cleaning, good lighting conditions, appropriate equipment, effective spotting agents, and expert knowledge are indispensable. Lighting conditions Lighting of the spotting table with composite artificial light consisting of a bluish and a yellowish fluorescent tube attached above the table approx. 80 cm from the standing point is better suited than daylight. This ensures high-contrast, shadow-free lighting of the work surface and allows for working without fatigue. Equipment Basic equipment should comprise a spotting table with vacuum facility, sleeve board, steam and compressed air guns, and a spray gun for water. Spotting brushes should have soft bristles for gentle treatment of textiles. Use brushes with bright bristles for bright textiles and brushes with dark bristles for dark textiles. Use spatulas with rounded edges for removing substantial staining. 14.7.2.2 Spotting agents In dry cleaning, normally three groups of spotting agents are used: Brushing agents Brushing agents containing surfactants and glycol ethers dissolved in low viscosity mineral oil and water. Brushing agents are used for pre-spotting to remove large stains from textiles. They are applied undiluted with a soft brush or sprayed onto the heavily stained areas before dry cleaning. Special spotting agents Special spotting agents are used for removing particular stains from textiles. The range consists of three different products in order to cover a wide range of different stains. The products are applied as drops directly from special spotting bottles onto the stain and are allowed to react. The three products are: • Acidic solution of citric acid, glycerol surfactants, alcohol and water for removing stains originating from tannin, tanning agent, and fruit dye. • Basic solution of ammonia, enzymes, surfactants, glycol ethers and water for removing stains originating from blood, albumin, starch, and pigments. • Neutral solution of esters, glycol ethers, hydrocarbon solvents and surfactants for removing stains originating from paint, lacquer, resin, and adhesives. Post-spotting agents Stains that could not be removed during basic cleaning must be treated with post-spotting agents. Most common is a range of six products, which are used in the same way, as the special spotting agents: • Alkaline spotting agent for stains originating from starch, albumin, blood, pigments • Neutral spotting agent for stains originating from paint, lacquer, grease, and make-up. • Acidic spotting agent for stains originating from tannin, fruits, beverages, and rust. • Acidic rust remover without hydrofluoric acid.
892
Kaspar D. Hasenclever
• Solvent combination for removing oily and greasy stains from textiles. • Bleaching percarbonate as color and ink remover. 14.7.2.3 Spotting procedure Correct procedure for spotting It is recommended to integrate the three stages of spotting • Brushing • Special spotting • Post-spotting into the process of textile cleaning. Perform brushing and special spotting when examining and sorting the textiles to be cleaned. Brushing Examine textiles for excessive dirt, particularly at collars, pockets, sleeves, and trouser legs. If textiles are to be dry cleaned check them particularly for stains originating from food or body secretions. If the textiles are to be wet-cleaned check them particularly for greasy stains. Apply a small quantity of brushing agent onto the stained areas and allow to react for 10-20 minutes before loading the cleaning equipment. Special spotting Intensive staining found during the examination of the textiles can be treated with special spotting agents. The stain substance must be identified and related to one of the following categories: • coffee, tea, fruit, red wine, grass, urine • blood, food, pigments, sweat • wax, paint, lacquer, make-up, pen ink, adhesives Depending on the category of the stain substance, apply the special product with the dripping spouts in the work bottle onto the stain and tamp it gently with a soft spotting brush. Allow to react for 10-20 minutes before loading the cleaning equipment. Post-spotting Stains that could not be removed during dry-cleaning or wet cleaning with machines are subject to post-spotting. Proceed as follows: • Place the garment with the stain area onto the perforated vacuum surface of the spotting table. • Identify stain, drip appropriate product undiluted onto the stain and tamp it gently with the brush. • For stubborn stains, allow product to react for up to 3 minutes. • Use vacuum to remove product and rinse spotted area with steam gun. • Dry with compressed air, moving the air gun from the edges to the center of the spotted area. Hidden spot test If it is suspected that a stain cannot be removed safely due to the textile material, the compatibility of the spotting agent should be checked by applying a small quantity of the agent at a hidden part of the garment. If the garment passes the test, the spotting agent is expected to be successful without damaging the garment.
14.7 Dry cleaning
893
14.7.3 TEXTILE FINISHING The use of solvents for textile finishing has only some importance for the treatment of fully fashioned articles. The processes are done in industrial dry cleaning machines. The advantages are the same as in dry cleaning compared to washing: lower risk against shrinkage and damage of sensitive garments. 14.7.3.1 Waterproofing In Northern Europe, North America and Japan, some kinds of sportsware need to be waterproofed. The treatment is done in dry cleaning machines with a load capacity larger than 30 kg ( > 60 lbs). The machines need to be equipped with a special spraying unit, which allows one to spray a solution of waterproofing agent into the cage of the machine. The waterproofing agents consists of fluorocarbon resins dissolved in a mixture of glycol ethers, hydrocarbon solvents and TCE. This solution is sprayed onto the garments, which are brought into the cage of the machine. The spraying process needs about 5 - 10 min. After spraying, the solvent is vaporized in the same way as drying in dry cleaning, so that the fluorocarbon resin will stay on the fibers of the garments. In order to get good results and the highest possible permanence of water resistance, the resin needs to be thermally fixed. In order to meet these requirements, a drying temperature of > 80°C (= 175°F) is necessary. 14.7.3.2 Milling Solvent milling has some importance for the treatment of fully fashioned woolen knitwear. The process is very similar to normal dry cleaning. The specific difference is the addition of water together with the detergent, in order to force an exact degree of shrinkage and/or felting. Milling agents are similar to dry cleaning detergent. They have specific emulsifying behavior, but no cleaning efficiency. The process runs like this: textiles are loaded into the machine, then solvent (TCE) is filled in before the milling agent diluted with water is added. After this addition, the cage rotates for 10 - 20 min. The higher the water addition, the higher the shrinkage; the longer the process time, the higher the felting. After this treatment, the solvent is distilled and textile load is dried. High drying temperature causes a rather stiff handle, low drying temperatures give more elastic handle. 14.7.3.3 Antistatic finishing Antistatic finishing is used for fully fashioned knitwear - pullovers made from wool or mixtures of wool and acrylic. The process is equal to dry cleaning. Instead of dry cleaning detergent, the antistatic agent is added. Antistatic agents for the treatment of wool consist of cationic surfactants such as dialkyl-dimethylammonium chloride, imidazolidione or etherquats. Antistatic agents for the treatment of acrylic fibers are based on phosphoric acid esters.
894
Martin Hanek, Norbert Löw, Andreas Mühlbauer
14.8 ELECTRONIC INDUSTRY - CFC-FREE ALTERNATIVES FOR CLEANING IN ELECTRONIC INDUSTRY Martin Hanek, Norbert Löw Dr. O. K. Wack Chemie, Ingolstadt, Germany
Andreas Mühlbauer Zestron Corporation, Ashburn, VA, USA
14.8.1 CLEANING REQUIREMENTS IN THE ELECTRONIC INDUSTRY The global ban of the CFCs has fundamentally changed cleaning in the electronic industry. Manufacturing processes were developed with the goal of avoiding cleaning. However, this objective has only be partially realized. Furthermore, there are still a lot of different areas in the electronic industry where cleaning of assemblies is highly recommended and necessary. Examples of such applications can be found in the aviation and space industries and in addition in the rapidly growing telecommunication industry. When investigating the production process of printed circuit assemblies, there are several important cleaning applications that have varying degrees of impact on the quality of the manufactured assemblies (see Figure 14.8.1).
Figure 14.8.1. Cleaning applications in the electronic production.
Cleaning applications in the SMT assembly line typically start with the cleaning of the bare boards (1). The preliminary substrates have to be cleaned in order to remove a variety of residues (particles, chemicals used in the manufacturing process of the bare boards). These residues may have a negative impact on the subsequent production steps. A detailed analysis1 of different kinds of contamination which might occur on the bare boards revealed
14.8 Electronic industry
895
that these residues might cause failures in production steps far beyond the initial printing process. Therefore it is important to assure clean surfaces in each production step. The cleaning applications (2-4) outlined in the SMT process line, shown in Figure 14.8.1, refer to the soldering process. Within this production step the components are placed on the bare board using solder paste that is printed on the solder pads through a stencil. The stencil can be cleaned continuously during the printing process by using a stencil printer underside wiping system. However, it has definitely to be cleaned after removing the stencil from the printer. The solder paste is removed from the apertures of the stencils in order to assure an accurate printing image (3).2,3 If the assemblies are to be mounted on both sides, the components are placed on one side by using an epoxy adhesive (SMT adhesive), in order to prevent them from being accidentally removed during the reflow process (3).3 The soldering process is carried out in a specially designed reflow oven. The cooling coils which contain condensed flux residue of such reflow ovens also have to be cleaned (5). In the case of wave soldering processes, the flux contaminated solder frames in addition must be cleaned. The contamination that must be removed after the soldering process is predominantly from flux residues. The removal of flux residues from the soldered assemblies (6) is generally the most critical application. Thus it is the important that the assemblies are cleaned. A subsequent coating process demands a very clean and residue-free surface to assure long term stability of the coating against environmental stresses such as humidity.1 Asked for the reasons for cleaning most of the process engineers give the following answers: 1) Stencils and other tools, such as squeegees, are cleaned in order to assure a reproductive and qualitative satisfying printing process.2 2) Contaminations from printed circuit assemblies are removed to achieve long-term reliability. 3) Cleaning is an important manufacturing step within the SMT process line if subsequent processes such as coating and bonding are required. In recent times an additional process step, which cleaning definitely is, was regarded to be time and cost inexpedient. Therefore a lot off optimization techniques have been applied to the SMT production process to try to avoid cleaning. However, considering new technologies in the electronic industry like fine pitch, flip chip and micro-BGA applications, the above statement is no longer true. The question if the removal of flux residues from printed circuit boards is necessary can only be answered by carrying out a detailed process analysis including the costs for cleaning and the increased reliability of the products. A large number of global players in the electronic industry prefer the advantages of better field reliability of their products. This aspect is even more important taking into consideration the outstanding competition in these industries. Whether a product has to be cleaned, and, especially, when cleaning is necessary in the SMT production depends on the following factors. • Process costs: cleaning is an additional process step that demands additional investment and resources. However, failures and downtime in other process steps can be diminished. • Process reliability: due to a larger process window, the manufacturing process as a whole is more stable. • Product reliability: the functionality and long term reliability of the products has to be assured.
896
Martin Hanek, Norbert Löw, Andreas Mühlbauer
After the decision has been made in order to choose the most suitable cleaning process, the following two questions have to be answered in advance: • Which contamination has to be removed? • What are the requirements regarding cleanliness of the surface, especially taking the subsequent process steps into account. The most important consideration in achieving an optimal cleaning result, is the interaction between the chemistry of the cleaning agent and the type of contamination. This is explicitly shown in the below presented lock and key principle. However, the cleaning result is not only depended on the nature of the chemistry used for cleaning. Other important factors in the cleaning process are the type of cleaning equipment used and the applied Figure 14.8.2. Factors influencing the cleaning process. process parameters such as temperature, time and method of application of the cleaning agent (see Figure 14.8.2). In order to optimize a cleaning process for the specific requirements, detailed information on the nature and composition of the kind of contamination that has to be removed is necessary. This tailoring of the chemistry is achieved by carrying out time consuming screenings matching typical contamination with suitable solvents. Another very important aspect is the material compatibility of the cleaning agent and the substrates that need to be cleaned. The cleaning chemistry should not attack the different materials of the components in any manner. Consequently, a number of long-term compatibility tests have to be carried out to assure this very important process requirement. Also, more environmental and worker safety issues are ever increasing when evaluating a new cleaning process. Cleaning processes approved by the industry have to pass even higher standards than are demanded from the government. They have to show, under the worst case scenario, a large process window and a large process margin of safety with respect to personal and environmental exposure. If all of the above mentioned process requirements such as • excellent cleaning result, • long-term functional reliability of the substrates after cleaning, • material compatibility between cleaning chemistry and the substrates, and • no drawbacks due to environmental and safety issues should be fulfilled, a lot of time and know-how has to be invested. 14.8.2 AVAILABLE ALTERNATIVES Due to the global ban of the CFCs, a large number of different cleaning processes4 have been developed. Each of them has advantages and disadvantages regarding the above mentioned process requirements. The available cleaning processes can be divided into the following main groups: 1) aqueous processes, 2) semi-aqueous processes, 3) water-free processes based on solvents, and
14.8 Electronic industry
897
4) special cleaning processes. In an aqueous cleaning process,5 cleaning is achieved by using water or a water-based cleaning agent. In order to avoid drying spots, rinsing is carried out with deionized water. By soldering with water-soluble solder pastes only hot water is necessary for cleaning. However, the use of an alkaline water-based cleaning agent is effective to remove more critical flux residues with low or high pressure spray in Figure 14.8.3. PCB cleaning. cleaning equipment. In semi-aqueous processes,4 cleaning is carried out using a high boiling mixture of organic solvents. Rinsing is achieved by using deionized water. The organic solvents used for these cleaners are predominantly based on alkoxypropanols, terpenes, high-boiling alcohols, or mixtures of different chemistries. In applications based on solvents such as isopropanol or alkoxypropanols6 a rinse with deionized water is not imperative, since these chemistries do not leave any residues if used for rinsing. Especially when using an uniformly boiling solvent-based cleaning agent, vapor rinsing with solvent vapor is possible. All there mentioned cleaning process will be discussed more detailed in the following sections. Less widespread applications in the electronic industry include: • Cleaning using supercritical carbon dioxide (SCF).7 • Cleaning using plasma.7 SCF cleaning alternative is especially useful for extremely sensitive and complex assemblies. At temperatures above 31°C and pressures above 73 bar, carbon dioxide transforms into a supercritical phase. Supercritical carbon dioxide reveals an extremely low surface tension. Consequently, the wetting of small gaps and complex assembly architectures can be achieved. However, the contamination that should be removed has to be nonpolar, and the compounds should be resistant to high pressures. Due to these basic limitations and the high costs of this cleaning process, this process does not play an important part in the electronic industry. The advantage of plasma cleaning is in almost waste-free removal of contaminations. The contaminations are transformed into carbon dioxide and water. However, this cleaning process is only suitable for removal of the small amounts of residue from the substrate surface. Combinations with aqueous cleaning processes are feasible and common in the field. 14.8.2.1 Water based systems; advantages and disadvantages 14.8.2.1.1 Cleaning with DI - water Cleaning without using any kind of cleaning agent in the electronic assembly process is possible when water-based fluxes are used for the soldering process. Since these aqueous fluxes are highly activated, aggressive residues occur after the soldering process. These residues have to be removed immediately to avoid severe corrosion of solder joints. The use of water-based fluxes is very common in the US electronic market due to the main advantage of a large soldering process window for soldering. DI-water is used for this purpose since it is able to solubilize the polar residues. However, physical energy, such as high-spray-pres-
898
Martin Hanek, Norbert Löw, Andreas Mühlbauer
sure, is in most cases necessary to assure complete and residue-free removal of the contamination. However, since pure deionized water reveals a relatively high surface tension, the wetting ability is extremely low.8 Consequently, being confronted with the situation of constant increase of the packaging densities on PCBs, flux residues especially, between or under components, are very difficult to be removed.8 As a result this presently predominant cleaning process will be in decline in the future. 14.8.2.1.2 Cleaning with alkaline water-based media The principle of this kind of cleaning process is based on the presence of saponifiers in the cleaning fluid. These saponifiers are able to react with the non-soluble organic acids, that are the main ingredients of the flux residues. In other words, the non-soluble flux residues are transformed into water-soluble soaps due to a chemical reaction between the saponifier and the flux. This reaction is called saponification. For such a cleaning process an effective rinse with deionized water is imperative to minimize the level of ionic contamination.1,5 However, the presence of alkaline components in the aqueous cleaning fluid may cause the phenomenon of corrosion on different metals (Al, Cu, Sn) resulting in the visible dulling or discoloration of solder joints and pads.9,10 The chemical nature of the saponifiers is either organic or inorganic. Organic saponifiers are for example monoethanolamine or morpholine.5 These soft bases transform the insoluble organic carboxylic acids to soluble soaps acting, in addition, as a kind of buffer to stabilize the pH of the cleaning solution. However, the above mentioned organic amines have an associated odor, and more important, are classified as volatile organic compounds (VOCs). Inorganic saponifiers based on the system sodium carbonate/sodium hydrogen carbonate are also very attractive alternatives exhibiting sufficient cleaning ability at pH-values below 11. By using these chemistries the VOC and odor issue can be avoided.5 Besides organic or inorganic saponifiers a large scale of different anionic, cationic or nonionic surfactants are present in water-based alkaline cleaning solutions. The main reason for adding surfactants is the requirement for lower surface tension in order to increase the wetting of the substrates. However, using surfactants in cleaning median in high-pressure spray in air cleaning applications, the issue of foaming has to be taken into consideration. 14.8.2.1.3 Aqueous-based cleaning agents containing water soluble organic components There is a large number of products on the market that are mixtures of water and water-soluble organic solvents. These media are recommended for the removal of solder paste, SMT adhesives and flux residues from stencils, misprints and populated reflowed PCBs.11 The water-soluble organic solvents are predominantly natural long chain alcohols, glycol ether derivatives or furfuryl alcohol. The basic principle of the cleaning process using such fluids is based on the removal of lipophilic contamination such as oil, grease, flux or adhesive through the organic components of the mixtures. After the removal from the surface of the substrates, the lipophilic contamination precipitates out of the water-based cleaning agent and can be removed by using a filtration devices. In order to guarantee a sufficient cleaning performance, these mixtures contain organic components up to 50%. Consequently, these formulations tend to be classified as cleaning agents with high content of volatile organic compounds (VOCs).
14.8 Electronic industry
899
14.8.2.1.4 Water-based cleaning agents based on MPC® Technology (MPC = Micro Phase Cleaning) Cleaning agents based on MPC Technology can be formulated either neutral or alkaline. The media combine the advantages of solvent and aqueous-based formulations revealing an extremely large process window.2,3 The principle of MPC systems is shown in Figure 14.8.4. The MPC media are based on formulations consisting of alkoxypropanols and water. Their phase behavior is dependent Figure 14.8.4. MPC® cleaning technology. on the temperature, agitation, and the dragged in contamination. The microphase, which is generated through temperature or media agitation, is responsible for the removal of the lipophilic components of the contamination. The ionics are removed by the aqueous phase.12 Since the ability to keep the organic contamination in the microphase is limited through the degree of loading, the lipophilic substances are expedited into the aqueous phase. Due to their lipophilic character, they precipitate out of the fluid and can be removed by a simple filtration device. As a result the loading of the cleaning fluid with contamination is minimal thereby minimizing the amount of organic components that finally pollute the subsequent rinsing stages.3 Alkaline pH, that is essential for the saponification process, can be realized through biochemical buffer systems, predominately aminoalkanoles. Due to the described cleaning principle, extremely long bath life times can be guaranteed with excellent cleaning results.2,3,12 The MPC® media are tailor made for all kinds of different cleaning equipments such as high and low pressure spray systems, spray under immersion, and ultrasonic applications.3 These cleaning agents based on MPC® Technology also contain low quantities of volatile organic compounds. However, due to this state of the art technology the percentage of VOCs are reduced to a minimal level. 14.8.2.1.5 Advantages and disadvantages of aqueous cleaning media Prior to the implementation of a cleaning process in a production line, the following criteria need to be taken into consideration: 1) Cleaning performance The substrates that have to be cleaned should be free of all kinds of residues that might have some negative influence on the functionality of the substrates. The contamination consist predominantly of flux residues, solder paste or SMT adhesive. Another important aspect with respect to the cleaning result is the compatibility between substrates and the cleaning chemistry. It is painless to remove the contamination while destroying the substrates. 2) Process reliability and process costs Every engineer tries to implement a process with a wide process window, and minimizing potential problems during production. While evaluating a cleaning process, the process parameters should be stable over a long period of time. As a result the equipment down time can be minimized, due to maintenance thereby optimizing the overall process cost.
900
Martin Hanek, Norbert Löw, Andreas Mühlbauert
3) Environmental aspects A cleaning agent should not only be considered based on its content of volatile organic compounds. A medium that does not contain any VOCs may on the one hand have no influence on air pollution. However, on the other hand, soil and water may be influenced. A VOC containing cleaning fluid has an inherently longer bath life. Consequently, the overall environmental calculation for some VOC containing cleaning fluid are better than by using VOC-free media. 4) Personal safety Another very important aspect for the evaluation of a new cleaning process is the issue on personal safety. The following topics have to be taken into consideration: • No risks during handling of the cleaning agent, especially regarding toxicity and flash point. • Low odor of the cleaning fluid. If the evaluated cleaning agent fulfils these important requirements, the chemistry will be accepted by the operator. Taking the discussed issues (1-4) into account, the advantages (+) and disadvantages (-) for the different aqueous cleaning fluids are as follows: a) Aqueous fluids using amines for saponification: + excellent removal of flux residues and unsoldered solder paste + predominantly VOC-free formulations low solid flux residues are difficult to remove limited removal of adhesives due to curing of the adhesive short bath life times resulting in high costs for media disposal b) Aqueous fluids using inorganic saponifiers and buffer systems: + satisfying cleaning results on flux residues and solder paste + VOC-free formulations + longer bath life due to constant pH level caused by the buffering systems low solid flux residues and SMT adhesives are difficult to remove c) Mixtures of organic solvents and water: small process window regarding flux residues and SMT adhesives VOC containing mixtures short bath life d) pH neutral and alkaline fluids based on MPC® Technology: + extremely large process window regarding flux residues, solder paste and adhesive removal + extremely long bath life-times due to previous described cleaning principle + economical cleaning process small amounts of VOCs. 14.8.2.2 Semi-aqueous cleaners based on halogen-free solvents, advantages and disadvantages A semi-aqueous cleaning process consists of a wash cycle using a mixture of different organic solvents followed by a rinse with deionized water. Organic contaminations, predominantly rosin flux residues, are removed through the lipophilic solvents, whereas the ionics are minimized by the polar rinse media. The different kinds of solvent formulations that are
14.8 Electronic industry
901
available on the market reveal an extremely wide process window. Almost all different fluxes that occur in the electronic assembly process can be removed efficiently. In principle, the media that can be used for semi-aqueous cleaning processes can be classified into two main groups:4 • solvent-based mixtures that are not soluble in water, and • water-soluble cleaning fluids. The water solubility of the solvent-based cleaning agent is very important for the treatment of the subsequent water rinse. 14.8.2.2.1 Water insoluble cleaning fluids The first media designed for this application were predominantly based on terpenes13 and different mixtures of hydrocarbons and esters. In order to increase the rinse ability of the solvent-based mixtures with deionized water, different surfactants were added.14 The principle of using a water insoluble cleaning fluid was based on the idea to avoid contaminating the rinse sections through the cleaning fluid. The lipophilic contamination should be kept in the organic cleaning media that can be easily separated from the rinse water by a skimming device. However, experience in the field revealed that this kind of media have some critical drawbacks. Since the surface tension of the organic cleaning formulations are lower than the surface tension of deionized water (approx. 72 mN/m), the cleaning fluid stays on the surface and will not rinse off. Consequently, the cleaning result is not acceptable with regard to ionic contamination and surface resistivities.1 Acceptable cleaning results can only be realized through complicated multi-stage rinse processes. Another important disadvantage is the potential risk of re-contamination of the substrate when lifting the substrate out of the rinse section, since the top layer of the rinse section consists of polluted cleaning media. Consequently, effective agitation and skimming of the rinse water is imperative to avoid this phenomenon.4 14.8.2.2.2 Water-soluble, water-based cleaning agents The solvents used for this application are based on water soluble alcohols, alkoxypropanols and aminoalcohols.4,6 The chemical structure of these compounds combines a hydrophobic and hydrophilic groups. Due to this ambivalent structure it is possible to optimize the water solubility and lipophilic character for an optimal cleaning performance. Consequently, cleaning fluids based on this principle reveal the largest process window. All different kinds of flux residues, especially low rosin fluxes, are easily rinseable with water resulting in an residue-free cleaning process. The quality and long-term stability of these processes are proven through many applications in the field.8,15,16 Applications and processes will be discussed more detailed later on. 14.8.2.2.3 Comparison of the advantages (+) and disadvantages (-) of semiaqueous cleaning fluids In order to quantify the effectiveness of the two main groups of solvent based cleaning agents that can be used for semi-aqueous processes, the following key aspects has to be taken into consideration: • cleaning performance, • process reliability, • environmental aspects, and
902
Martin Hanek, Norbert Löw, Andreas Mühlbauer
• personal safety. a) Non water soluble cleaning agents: + wide process window regarding fluxes, solder pastes and adhesives low rosin flux residues are critical to remove possibility of re-contamination of the substrate in the rinsing stage cleaning fluids contain volatile organic compounds. b) Water soluble solvent based cleaning agents: + extremely wide process window regarding all kinds of fluxes, solder pastes and adhesives + residue-free removal of the cleaning fluid due to excellent rinse ability VOC containing formulations 14.8.2.3 Other solvent based cleaning systems Besides the already mentioned solvent based cleaners, there are some special solvents on the market. However, due to their physical and toxicological character, most of them are used for special applications. The advantages and disadvantages of the different products are summarized below. 14.8.2.3.1 Isopropanol
Figure. 14.8.5. Zestron FA/IPA.
Isopropanol is sometimes used for the cleaning of assemblies. More common, however, is the use of this solvent in stencil cleaning applications. Isopropanol is often also used as a rinsing media when cleaning with a solvent for hybrid or ceramic applications prior to bonding (see Figure 14.8.5). However, when using isopropanol for any kind of cleaning application there are some essential disadvantages that have to be taken into consideration: • Flammability due to low flash point, • Requires explosion-proof cleaning equipment, • Small process window with respect to the ability to flux residues, • Limited removal of adhesives due to curing of the adhesives, and • White residues formation with some flux residues. 14.8.2.3.2 Volatile siloxanes Due to their extremely low surface tensions combined with excellent wetting character, siloxanes were recommended for electronic cleaning applications. While siloxanes can be
14.8 Electronic industry
903
used for the removal of non-polar contaminations, ionic contaminations cannot be readily removed. Consequently, for an effective cleaning process, formulations containing siloxanes and alkoxypropanols are needed. However, the use of siloxanes containing cleaning fluids does not play an important role in electronic assembly cleaning applications with respect to the above mentioned disadvantages. 14.8.2.3.3 Chlorinated solvents Chlorinated solvents such as trichloroethylene or 1.1.1-trichloroethane are nonflammable and excellent cleaning results can be achieved on especially non-polar residues. However, ionic contamination cannot be removed sufficiently. There are predominant environmental issues, in particular, their ozone depletion potential, which makes their use in electronic cleaning applications obsolete. 14.8.2.3.4 n-Propylbromide (nPB) Cleaning fluids based on formulations using n-propylbromide (nPB) have to be regarded very critically. Since these products also contain halogens, they cannot be considered to be real alternatives for chlorinated or fluorinated hydrocarbon mixtures. Due to these open environmental questions, cleaning applications using n-propylbromide formulations are still not very common, especially in Europe. 14.8.2.3.5 Alkoxypropanols Cleaning agents based on alkoxypropanols show extremely satisfying cleaning results due to their chemical structure. The lipophilic part of the molecule is capable of removing organic soils such as greases or oils from the substrate surface, whereas the hydrophilic part is able to remove the ionic contaminations. Furthermore, products based on this kind of chemistry show a large process window on all different types of flux residues. Different applications and processes will be discussed in detail in the following sections. 14.8.2.3.6 Hydrofluorinated ethers Hydroflourinated ethers are a new generation of cleaning fluids revealing no ozone depletion potential. Since they have no flash point there is no need to use explosion-proof cleaning equipment, thereby the cost of cleaning equipment is dramatically reduced. In addition, with their extremely low surface tension and viscosity, they are able to penetrate the tightest spaces of assemblies with high packing densities. Physical properties and different applications and processes will be discussed in more detail below. 14.8.2.3.7 Advantages and disadvantages of solvent-based cleaning agents Advantages: • Solvents exhibit a large process window on all different kinds of fluxes, adhesives and solder pastes • Most solvents can be regenerated through distillation, • There is no need of extensive treatment systems for the rinse water, Disadvantages: • Higher proven costs due to investment in explosion-proof cleaning equipment, • VOC containing solvents, Process: • Handling risks for the operator due to flammability, toxicity and odor, and
904
Martin Hanek, Norbert Löw, Andreas Mühlbauer
• Good material compatibility with respect to materials commonly found on assemblies. 14.8.3 CLEANING OF TOOLS AND AUXILIARIES The prohibition of CFCs in 1993 in Europe found the electronic industry scrambling to undertake appropriate actions. Various alternative technologies have asserted themselves during the conversion to environmentally friendly processes. Basically, two courses have evolved as alternative to the use of CFCs for changing applications: • New soldering techniques that avoid cleaning as a process have been developed and introduced in the electronic industry. • New environmentally friendly cleaning processes have been introduced. A large number of European companies decided in favor of no-clean processes, in spite of the associated disadvantages. In the USA cleaning has not be abandoned; it is still an important step in the SMT production line. However, with the increase of packing density and increased quality demands, more and more European companies are now returning to cleaning processes. In general, the purpose of changing is for the removal of contamination which might otherwise harm the operation of the electronic circuit assembly. Ensuring high reliability for a circuit assembly by effective cleaning is extremely important in such domains as the military, aviation, telecommunications, and in the automotive industry. The need for change to high reliability can be expected to grow appreciably because of different factors. These factors range from more complicated assemblies with higher density to the greatly accelerating use of conformal coatings in the automotive, niche computer, consumer electronics, and telecommunications markets. Dramatic growth is expected to occur in the USA, Europe and Asia. The applications where cleaning is necessary can be divided into three main applications: • the removal of flux residues from printed circuit assemblies (PCBs) after reflow or wave soldering; • the removal of solder paste and/or SMT-adhesive from stencils screens or misprints, and • the tool cleaning (dispensing nozzles, solder frames, cooling coils from reflow or wave soldering ovens, and squeegees). 14.8.3.1 Cleaning substrates and contamination In recent years stencil and misprint cleaning has become a significant issue for many users. Normally, in a typical PCB assembly line, the solder paste is printed through screens or stencils. As an alternative to glue dispensing, specially formulated SMT adhesives can now be printed through a stencil. Consequently, the stencils must be cleaned periodically to maintain the quality and yield of the process. However, the removal of SMT adhesive gives rise to completely new demands on the cleaning process and the condition of the stencils. The following specifies the points that must be observed in connection with stencil cleaning. Great importance has therefore been given to comparison of different cleaning chemistries, cleaning equipment, and comparison of manual and automated cleaning. Ever-increasing component density and the result expected from the printed image give rise to stringent requirements regarding stencil cleanliness. A flawless printed image is very important because misprinted boards can no longer be used without cleaning. This can be very expensive, particularly if components have already been mounted onto one side of a board. As a consequence, stencil printing for the application of solder paste on SMT boards
14.8 Electronic industry
905
has become a standard. Stainless steel stencils with apertures that are either etched or cut with a laser are used for this purpose. They can also be used for fine-pitch apertures. Studies have revealed that more than 60% of the defects arising in the SMT process are caused by inadequate paste printing. This, in return, can be traced to insufficient cleaning that is often done manually. Consequently, a well working stencil cleaning process is an imperative. With regard to the cleaning process it is very important that no paste or cleaner residues remain in a stencil aperture, as this will inevitably cause misprints (see Figure 14.8.6, Figure 14.8.6. Left: Residues of SMT adhesive cured in the aperture of a stain- right). The printing of adless steel stencil. Right: Stainless steel stencil encrusted with solder paste. hesives, on the other hand, requires some points to be taken into consideration since this method has to produce different dot heights. The ratio between the stencil thickness and a hole opening is the decisive factor with this technology. Stencils for adhesive application are made of either stainless steel or plastic and are usually thicker than those used for paste printing. Cleaning is complicated by the thickness of the stencils (often exceeding 1 mm). Consequently, small holes require a cleaner with intense dissolving properties to ensure that the adhesives are completely removed (see Figure 14.8.6, left). Although cleaning of stencils improves the printing results drastically, the production of misprints during the printing process cannot be avoided completely. However, the cleaning of misprinted assemblies is an application that is still frequently ignored. This inFigure 14.8.7. Left: Misprinted PCB, Right: Via filled with solder paste. volves the removal of misprinted or smeared solder paste or SMT adhesive. The removal of solder paste as the most widespread application gives rise to the following demands: • No solder paste must remain on the surface, or the vias (see Figure 14.8.7). • The surface must be perfectly clean and dry to ensure immaculate renewed printing. • The solder flux on double-sided PCBs must also be removed from the side where soldering has been finished. 14.8.3.2 Compatibility of stencil and cleaning agent In some cases, repeated washing can damage stencils cleaned in automated washing stations. This damage usually occurs through degradation of the frame-to-screen glue. This degradation occurs at the junction of the stencil and the mesh and also at the junction of the mesh and the frame. The causes for this degradation lay in two factors. First, the high temperatures that are needed to clean stencils with aqueous cleaners cause thermal stresses
906
Martin Hanek, Norbert Löw, Andreas Mühlbauer
along the glue boundaries. Secondly, the glue itself, which in most cases is a cured epoxy material, may dissolve in the cleaning medium. The combination of these two factors can lead to severe stencil defects. A protective coating can be applied to the glue in cases where it is known that the glue is not stable towards the cleaning medium. Advanced cleaning agent manufacturers work closely with stencil manufacturers to avoid this problem. However, testing in advance of installed cleaning operation ensures that no unexpected stencil defects will affect the production process. 14.8.3.3 Different cleaning media When selecting a cleaning process, the chemistry should be selected first, because the cleaning chemistry has to be adopted to the chemical ingredients of the contamination. Only after the evaluation of the proper chemistry, can the method of applying the cleaning agent be evaluated. Solvent-based cleaning agents such as hydrocarbons, alcohols, terpenes, and esters offer properties that make them effective cleaning agents for the electronics industry. Low surface tension, high solubility, and ease of drying allow organic solvents to clean effectively where water-based cleaning agents are less effective. These benefits are very similar to the physical properties of the chlorofluorocarbons (CFC) that were once in widespread use in electronics manufacturing operations. Solvents, however, have several significant drawbacks that have limited their use including: • Flame and explosion hazards, • Higher equipment costs, due to required explosion protection, • Personal exposure concerns, • Material costs, and • Disposal of used solvents. And while water-based cleaning agents have been much improved in recent times, their performance does not always match that of their solvent forerunners. In fact, an ideal water-based cleaning agent would combine the physical properties of solvent-based cleaning agents with the safety and convenience of water-based materials. Solvent-based cleaning agents Cleaning with solvents such as isopropanol or acetone still remains the most widely applied method because any type of solder paste or SMT adhesive can be removed from the stencils by specially formulated mixtures. Consequently, the cleaning process window for such modern solvents is sufficiently wide to ensure the required results when changing over to another paste or adhesive. Normally, such cleaning processes are performed in machines designed specifically for solvent cleaning. The use of cleaners with flash points in excess of 104ºF (40°C) means that the machines can be set up without having to maintain an explosion protected area. Moreover installation in a special room may be necessary. However, the handling requirements of highly flammable cleaners, for such as isopropanol (IPA, flash point of 54ºF, 12°C), are far more critical, and cleaning is not so effective, particularly with fine-pitch stencils and SMT adhesives. Besides, modern solvent cleaners have been formulated in such a manner that health hazards are minimized - provided, they are properly used. Water-based cleaning agents The benefits of cleaning with water-based agents are straightforward. They include: • Ease of use, • Absence of fire or explosive hazards, • Simple disposal,
14.8 Electronic industry
907
• Environmental and personal safety rules. In some cases even pure water can be used as the cleaning medium (presupposing that the contaminants are fully water-soluble). In most cases, however, water requires the addition of chemical or mechanical energy to obtain good cleaning performance. This can come in the form of: • A cosolvent, • A chemical activator, • Elevated temperature, • Mechanical action, or • High-pressure spray or ultrasonics. Contaminants such as oils, greases and SMT adhesives that feature high organic matter content are not readily soluble in water. Adding chemical activators such as saponifiers to the formulation can increase their solubility. Unfortunately, saponifiers require in most cases elevated temperatures (38-60°C) in order to react with the contamination. However, temperatures in this range will most likely damage the stencil. Saponifiers improve cleaning by reducing the surface tension of the cleaning solution, permitting the agent to penetrate void spaces and stand-off gaps that pure water cannot reach. The saponifiers in water-based cleaning agents are either organic or inorganic. Organic saponifiers are, in most cases, fully miscible with water and can easily be rinsed from the stencil. Any residues remaining after rinsing evaporate during drying. Inorganic saponifiers are usually water-soluble materials applied as a solution in water. They are nonvolatile, and residues remaining after rinsing might leave contamination on the substrate. The alkaline saponifier is consumed during the cleaning process and requires constant replenishment. In addition, the high pH value and the elevated application temperatures can also cause an oxidative attack of the aluminum frame of the stencil. Elevated temperatures are also commonly used to increase cleaning performance. An increase in temperature will lead to a corresponding increase in the solubilizing properties of the cleaning medium. There is of course a point at which the thermal stress induced by high temperature cleaning can damage the stencil adhesion. Lastly, high pressures or ultrasonics are used as a way to remove contaminants by transferring mechanical energy to particulate contamination. Pressure helps to force the cleaning agent into void spaces and stand-off gaps. Pressure can only marginally improve cleaning performance when the contaminant is not in particulate form or when solubility is the limiting factor. Water-based cleaners based on MPC® technology Satisfactory results in stencil cleaning with aqueous systems can only be achieved for the removal of solder paste. Most aqueous systems used to remove adhesives failed in the past because the water caused the adhesives to set. Micro Phase Cleaning (MPC)2,3,12 refers to the use of a cleaning agent formulated to undergo a phase change at elevated temperature. When heated above a threshold temperature (typically 100-120oF, 40-50°C) a microphase cleaner changes from a clear colorless solution to a turbid milky mixture. The phase transition produces a cleaning mixture that exhibits the properties of both solvent-based and water-based cleaners. “Water-like” properties allow the cleaning agent to effectively dissolve ionic contaminants and remove them from the substrate’s surface.
908
Martin Hanek, Norbert Löw, Andreas Mühlbauer
Non-polar and organic residues are removed by contact with the hydrophobic or “solvent-like” phase.12 No specialized equipment is needed to handle microphase cleaning agents. In most cases a microphase cleaning agent can be used as a drop-in replacement in existing equipment. Microphase cleaners are effective in all types of cleaning equipment capable of applying the cleaning agent in a liquid form. This includes spray and ultrasonic equipment. Table 14.8.1. Comparison of different cleaner types Cleaning Agent
Pros
Cons
Modern solvents
Removal of adhesives and solder paste Wide process window Cleaning at room temperature Do not oxidize/corrode stencil
Flammable Emits solvent vapors into work area Slow drying VOC (volatile organic compound) Ozone depletion potential Require explosion proof equipment
Aqueous alkaline cleaner
Cheap Non-flammable Non ozone depleting No VOCs Mild odor
Do not remove adhesives Short bath life times Narrow process window Water rinse necessary
Aqueous based cleaners of MPC Technology
Removal of adhesives and solder paste Wide process window Non-flammable Rapid drying Residue-free drying Long bath life
Agitation of the cleaner (spray, ultrasonic, spray under immersion, overflow) necessary
Table 14.8.2 presents the typical technical characteristics of different cleaning agents developed by Dr. O.K. Wack Chemie for stencil and misprint cleaning applications. 14.8.3.4 Comparison of manual cleaning vs. automated cleaning Printing screens are often cleaned manually (Figure 14.8.8) as users still shy away from the purchase of a cleaning machine because of the associated investment costs ($6,500 to $60,000). However, a more accurate assessment must take the following points into account: Table 14.8.2. Technical data of typical (1) The mechanical rubbing action of cleaning agents (water-based and solvent) the repeated manual cleaning impairs the stencil surface. This is particularly apparent ® ® Cleaning Agent Vigon SC200 Zestron SD300 with plastic stencils. These changes to the Chemistry water based solvent based stencil surface can result in misprints and shorten the service life of the stencil, makFlash point none 106°F ing it very difficult to track the source of the Appl. Temp. 77°F 77°F problem. (2) Precise repeatability of the cleanspray in air ultra- spray in air (ex Process sonic manual proofed) manual ing results is not guaranteed with manual
14.8 Electronic industry
909
cleaning because manual applications are not based on precisely defined cleaning processes. This can also result in fluctuations of the printing result. (3) Manual cleaning must be completed with the utmost care. Such care costs time and money, thus making it very expensive. The investments for a cleaning machine can be recovered within a very short period, independent of the number of substrates that are to be cleaned. Example: A worker requires 1 hour per day for stencil cleaning. On the basis of $40.00/h and 220 workdays, a small machine ($8,700) can be a worthwhile investment within a year. Some users clean misprinted assemblies manually. This can prove to be very labor and cost intensive when larger quantities are involved. The cleaning results and their reproducibility will be rather inconsistent. Manual cleaning of PCBs with vias and blind holes with a brush or cloth can mean that solder beads are actually rubbed into the holes. Such holes can only be satisfactorily freed by ultrasonic or other mechanical means. It is imperative that the via are free of solder particles, otherwise the consequential damage can be very expensive. 14.8.3.5 Cleaning equipment for stencil cleaning applications As previously mentioned, it may well prove to be an economical proposition to invest in a specially designed cleaning machine if this can be justified by the number of substrates that have to be cleaned. In general, PCB assemblers use the following different types of cleaning equipment for stencil and misprint cleaning: • Spray in air, • Spray under immersion, and • Ultrasonic. Spray in air cleaning equipment Spray in-air machines consist of either a single or multiple reservoirs containing wash and rinse fluids (detergents, solvents, saponifiers, water, etc.). The stencils are placed in a separate chamber, and the wash/rinse fluids are then pumped from the reservoirs into the stencil chamber, and delivered onto the stencil via low pressure (<70 psi) spray wands. These arms may be either co- or counter-rotating. When designed for using a solvent as a wash medium, these machines must be designed to be explosion protected (intrinsically safe electronics, pneumatic or TEFC pumps, sealed conduits, etc.). These machines may also be outfitted to use an inert atmosphere, such as nitrogen, in the stencil chamber. Figure 14.8.8. Scheme of typical spray in air type of When used with aqueous-based mecleaning equipment. dia, they may often be equipped with a wash and or rinse tank heater. Advantages of this type of machine are:
910
Martin Hanek, Norbert Löw, Andreas Mühlbauer
• Relatively inexpensive (although explosion protecting oftentimes significantly increases the cost of the machine),
• Easy to use, • A large number of options (add-ons) available (in-line filtration, closed loop water recirculation, programming options, etc.),
• Do not require as much media as other types of stencil cleaners, because they do not require total immersion of the stencil,
• The stencil is also washed, rinsed, and dried in the same chamber, thus eliminating the need to move the stencil around manually. Disadvantages include the sometimes large footprint that results from having to provide up to three separate chambers (wash, rinse, and stencil), and the sometimes inadequate agitation (spray pressure) generated by the machine. This can make it difficult to clean fine-pitch apertures. Spray under immersion cleaning equipment In spray under immersion equipment, the stencil is placed into a bath (immersed) containing the cleaning medium. The medium is then recirculated using submerged spargers to provide agitation to the stencil surface. A separate chamber is then used for rinsing and drying. Advantages of this type of machine are that it is: • Most gentle on the stencil, and • The capital expense is relatively low. However, because of the low pressure, Figure 14.8.9. Scheme of typical spray under immercleaning cycles may require longer cleaning sion batch cleaner. times. Also, the size of the machine and the quantity of cleaner required are dictated by the size of the stencil to be cleaned. Ultrasonic cleaning equipment Ultrasonic machines, as the name implies, use ultrasonic “vibratory” energy to clean stencils. The stencil is immersed in a bath containing the cleaning medium, and exposed to ultrasonic energy. The ultrasonic frequency varies, but is typically between 40 and 100 kHz. Options for the ultrasonic cleaning machine include overflow recirculation of the medium, as well as spray under immersion jets to Figure 14.8.10. Scheme of typical ultrasonic batch cleaning aid cleaning and bath agitation. The equipment also varies greatly in size equipment. (from tabletop models to large footprint models that include rinsing baths, air dryers, and robot arms). Advantages of this unit are that: • The ultrasonic energy is much better suited to cleaning fine pitch openings, and
14.8 Electronic industry
911
• External heaters often may not be needed when using aqueous medium, because of the bath heating that the ultrasonic energy accomplishes. However, because of this heating, these units are not recommended for use with solvent cleaners. Also, precise control of the ultrasonic energy is required to prevent possible damage to the stencil and apertures. Comparatively, also, ultrasonic machines are more expensive than either type of spray equipment. The size, and thus the cost, is directly proportional to the size of the stencil to be cleaned. 14.8.3.6 Stencil cleaning in screen printing machines Today, many screen printers incorporate an automatic cleaning system for the undersides of stencils. During the process a wiper is moistened with a cleaner that wipes the stencil. On high-end printers this unit is programmable to permit subsequent dry wiping and vacuum drying. Cleaning the underside of stencils may not replace machine stencil cleaning, but it does extend the time, in some cases significantly, until machine cleaning becomes necessary. A disadvantage of these automatic-cleaning units is the fact that only solder paste removal is possible. The main problem associated with cleaning the stencil underside of adhesive stencils is that it does not remove the adhesive from the stencil holes. This means that only the surface is cleaned, while the holes are untouched. Cleaning agents suitable for this application already exist. 14.8.3.7 Summary The cleaning process, which a user finally selects, depends to a major extent on the operating conditions. Each process has its own specific advantages and disadvantages, so users are supported by advice and testing. These can be accomplished under tightly simulated process conditions by a technical center. 14.8.4 CLEANING AGENTS AND PROCESS TECHNOLOGY AVAILABLE FOR CLEANING PCBs 14.8.4.1 Flux remove and aqueous process With the banning of CFCs - enforced in Germany since January 1993 - users have been confronted with the problem of ensuring cost effective elimination of ozone depleting chemicals from the production process while having to satisfy constantly rising quality requirements. In the electronic industry it is customary to set certain quality standards in the manufacturing process of electronic assemblies. Nevertheless, it is often not enough to check the quality of a part immediately after it has been manufactured as long-term quality must also be taken into account. However, technically proficient and efficient processes are available that can exclude long-term risks and even reduce overall costs. A large number of companies have decided in favor of no-clean processes. Though, with the expansion of packing density and increased quality demands, more and more companies are now returning to cleaning processes. 14.8.4.1.1 The limits of a no-clean process The no-clean concept (i.e., no cleaning after soldering) proceeds from the following assumptions: (1) The used soldering fluxes have a low solid content of approx. 2 to 3% and contain strong activators. These activators, however, are critical as they can easily form highly conductive electrolytes under humid climatic conditions.
912
Martin Hanek, Norbert Löw, Andreas Mühlbauer
Figure 14.8.11. Soldering flux residues on a component.
Figure 14.8.12. Dentride growth between connecting contacts.
(2) Reduction of the resin content and quantity of activators simultaneously diminishes the risk of leakage current, electro-migration and corrosion. After soldering, cleaning removes all residues from the pc-boards, whereas the no-clean soldering process leaves residues on the pc-boards surface, thereby exposing them to the influence of humidity. Consequently, the presence of soldering flux residues on assemblies (Figure 14.8.11) can cause significant functional disruptions. To guarantee quality assurance with no-clean fluxes, the standard tests previously employed in the days of CFC cleaning where transferred accordingly to no-clean pc-boards, including: • Testing for impurity, • Testing for surface resistance (SIR test), • Testing for electro-migration, • Testing for bondability, and • Testing under climatic stresses. Testing for ionic contamination and establishing the surface resistance usually produces good results with pc-boards soldered in a no-clean process. Nevertheless, the values are still significantly poorer than with cleaned assemblies. The migration and precipitation of metal ions can cause the dangerous growth of dendrites (Figure 14.8.12). These metal dendrites are outstanding electric conductors that diminish the surface resistance and, in the long-term, result in short-circuits. Coating can significantly reduce the risk of electro-migration. However, this demands a clean surface, otherwise homogeneous adhesion of the coating is not assured. 14.8.4.1.2 Different cleaning media and cleaning processes When selecting a cleaning process, the cleaning chemistry is the most sensitive parameter, because the cleaning agent has to be adapted to the chemical ingredients of the contamination. Only after determining the most suitable chemistry, should the method of application be evaluated. In principle cleaning processes can be split into three separate categories: (1) Semi-aqueous cleaning, (2) Aqueous cleaning, and (3) Solvent cleaning.
14.8 Electronic industry
913
14.8.4.1.3 Semi-aqueous cleaning A process that has found widespread use in practice is semi-aqueous cleaning with ZESTRON® FA. This cleaner was especially developed for removal of no-clean soldering fluxes in ultrasonic batch or spray under immersion cleaning systems. It is used in the cleaning bath, rinsed with demineralized water, and then dried with hot air (see scheme in Figure 14.8.13.). Figure 14.8.13. Diagram of a semi-aqueous cleaning process with ® The carried-in flux constituZESTRON FA. ents can be removed by a flux absorber system to ensure the cleaning medium maintains a constant quality. Cartridge filters also remove precipitated flux residues and particles. Two cascaded rinsing units - a pre-rinse and a fine rinse - follow the wash stage. The rinsing water contaminated with carried-over cleaning medium is passed through different sized particle filters, mixed-bed ion exchangers, and activated carbon filters. The quality of the rinsing water can be continuously monitored by conductivity measurement to establish its residual ion content. The conductivity value should be below 1 µS/cm. The complete process in based on a closed-loop principle. Consequently, the cleaning process merely liberates those substances that are to be removed from the pc-boards (i.e., soldering flux residues). Closed-loop operation is not only advisable for ecological reasons, but is also a very sound economic proposition as it significantly minimizes process costs. 14.8.4.1.4 Aqueous cleaning in spray in air cleaning equipment Low pressure spray in air batch cleaning systems (see Figure 14.8.14) can be attractive propositions for users who only have a low volume of pc-boards to clean and are required to furnish proof of a specified cleaning process to their customers. The water-based mildly alkaline cleaning agent VIGON® A200 was specifically developed for this application. High-pressure spray in air in-line cleaning processes that remove flux residues from PCBs, are often employed by manufactures to clean large quantities of boards. This still represents the most economical way to clean large quantities of PCBs. Especially for contract manufactures that are required to clean different types of PCBs, often with several different flux residues. The parts to be cleaned are fed, one after the other, into a continuous spray cleaner where they pass through the different proFigure 14.8.14. Diagram of an aqueous in-line cleaning cess stages indicated in Figure 14.8.15. The ® process with VIGON A200.
914
Martin Hanek, Norbert Löw, Andreas Mühlbauer
Figure 14.8.15. Scheme of a low-pressure batch cleaning system.
high-pressure cleaning unit is supplied with VIGON® A200 by a chemical dosing station. Cleaning medium quality can be maintained by installing a flux absorber system. The cleaned parts are blown-out with an air knife to minimize carry-over losses from the high-pressure cleaning section. Pre-rinsing the parts with deionized water follows the cleaning section. A fine rinse unit ensures minimum ionic contamination on the PCBs. Finally, the cleaned parts are then blown off with hot air. Complete drying, particularly of parts that retain a great deal of liquid, can be performed with an infrared dryer.20,21 14.8.4.2 Flux removal from printed circuit boards - water-free cleaning processes In the European Community, the production and sales of chlorinated fluorocarbons (CFCs) have been banned since 1995. Due to their ozone depletion potential (ODP) hydrochlorinated fluorocarbons (HCFCs) are also subject to be phased-out. The revision of EC Direction 3094/95 proposed in Article 5 the ban of HCFCs as solvents for parts cleaning. HCFCs are not allowed in open solvent applications and their use in contained cleaning systems, such as industrial equipment with ultrasonic agitation, will be banned by January 1, 2002. However, the aerospace and aeronautic industries are allowed to use HCFC until January 1, 2002. Furthermore, in some countries, including Germany and Switzerland, national ordinances have prohibited the use of HCFCs to replace CFC solvents for several years. Consequently, the search for alternative solutions to avoid cleaning (no clean technology) or to use environmentally acceptable solvents leads to new cleaning processes including new cleaning agents and new cleaning equipment. • NovecTM hydrofluoroether (HFE) in combination with a cosolvent, and • High boiling cleaning agents that can be recycled continuously by vacuum distillation. These two new processes can permanently replace both 1,1,1-trichlorochlorethane and CFC-113, as well as, chlorinated solvents such as dichloromethane, trichloroethylene (TCE) , perchloroethylene and HCFC-141b.
14.8 Electronic industry
915
14.8.4.2.1 Water-free cleaning processes using HFE (hydrofluoroethers) in combination with a cosolvent The cleaning process: A cosolvent cleaning process is one, in which the cleaning and rinsing solvents are of significantly different composition. For example, in semi-aqueous systems (see Cleaning Technologies on the PCB Assembly Shop Floor (Part II) in EPP Europe, October 1999) cleaning is done with organic solvents and water is used for rinsing. In the cosolvent cleaning process, cleaning is accomplished primarily by the organic cosolvent, and rinsing is effected by a fluorochemical. Typically, in cosolvent systems, the boil (wash) sump contains a mixture of cosolvent and fluorochemical rinsing agent. The rinse sump normally contains essentially 100% fluorochemical rinsing agent. Referring to Figure 14.8.16, a cosolvent process operation can be summarized as follows. Printed circuit boards containing flux residues are immersed into the boil sump, which contains a mixture of cosolvent and fluorochemical rinsing agent, typically about equal volumes of each. After being cleaned in the boil sump, the substitutes are immersed in the rinse sump, which contains nearly 100% fluorochemical. Figure 14.8.16. Typical vapor degreasing equipment. The primary function of the rinse sump is to remove the fluorochemical/cosolvent mixture and the soil dissolved in it. Following immersion in the rinse liquid, the parts are moved into the vapor phase for a final rinse with pure fluorochemical vapor. Finally, the parts are lifted into the freeboard zone, where any remaining fluorochemical rinsing agent evaporates from the parts and returns to the sump by condensation from cooling coils. At this point, the cleaning cycle is complete and the parts are clean and dry. Furthermore, pure NovecTM HFEs can be used for the removal of light soils, halogenated compounds like fluorinated greases or oils, and other particles. The increasing popularity of the cosolvent process is mainly due to its flexibility in allowing independent selection of both the solvating and rinsing agents that best meet the needs of a particular cleaning application. The fluorochemical cosolvent process is even capable of defluxing components with very complex geometry due to the advanced physical properties of the cosolvent. The cleaning chemistry: NovecTM hydrofluoroethers (NovecTM HFEs) are rapidly becoming recognized as the solvents of choice for superior cleaning in the electronics and precision engineering industries. The combined properties of these new materials have proven to be very effective alternatives to ozone depleting substances (ODSs) like the previously mentioned CFCs and HCFCs. The new developed chemical compounds methoxyfluorobuthylether (HFE 7100) and ethoxyfluorobuthylether (HFE 7200). HFE rinsing agents can be used in combination
916
Martin Hanek, Norbert Löw, Andreas Mühlbauer
with respective cosolvents. The most effective cosolvents for defluxing applications in the market are based on alcoxypropanols which are specially designed for applications in combination with HFEs. Since the boiling points of these two HFEs lie at the upper range covered by the traditional ODSs such as HCFC-141b, CFC-113 and 1,1,1-trichloroethane, a better cleaning efficiency is provided due to the increase in solubility of most soils with temperature. Like ODS solvents, the HFEs are non-flammable, which leads to very simple and thus inexpensive equipment designs. The low surface tension of HFEs helps these fluids to easily penetrate tight spaces, especially between and under components of populated printed circuit boards. In addition, this is supported by the low viscosity of the HFEs that also allow it to quickly drain from the parts that have been cleaned. One useful parameter for assessing the potential performance of a cleaning agent is the wetting index, which is defined as the ratio of the solvent’s density to its viscosity and surface tension. The wetting index indicates how well a solvent will wet a surface and penetrate into tight spaces of complex cleaning substrates. The higher the index, the better the surface penetration. Table 14.8.3. Properties of HFEs compared to common CFCs and HCFCs Property
HFE-7100
HFE-7200
HCFC-141b
CFC-113
60
73
32
48
Flash point, °C
none
none
none
none
Surface tension at 25°C
13.6
13.6
19.3
17.3
Viscosity, Pa s
0.61
0.61
0.43
0.7
Wetting index
181
172
149
133
Boiling point, °C
Since HFEs contain no chlorine nor bromine, they do not have an ozone depletion potential. In addition, their atmospheric lifetime is short compared to CFCs. Furthermore, the Global Warming Potential of HFE fluids is significantly lower than other proposed fluorinated solvent replacements. Practical studies as well as laboratory tests have found that HFE solvent emissions are 5-10 times lower than the emissions of CFCs or HCFCs. Although, HFEs are volatile organic compounds (VOC), they are not controlled by related US directive regulations. Finally, toxicological tests have shown, that these products have extremely low toxicity which allows for higher worker exposure guidelines, unlike many of the chlorinated and brominated solvents offered as ODS alternatives. 14.8.4.2.2 Water-free cleaning processes in closed, one-chamber vapor defluxing systems Liquid vapor degreasing using solvents has been an accepted method of precision cleaning for over 50 years. This cleaning process incorporates washing, rinsing, drying and solvent reclamation in a compact, cost-effective unit. Consequently, this makes it a very attractive process for many production cleaning applications. Environmental concerns raised in the last decade have changed both the chemistry and equipment technology for degreasing. Solvents have been modified or replaced to eliminate ozone depletion potential and other haz-
14.8 Electronic industry
917
Figure 14.8.17. Closed-loop batch vapor degreasing process.
ards. Equipment designs and operating processes have been totally re-engineered to reduce solvent losses to near zero. Referring to Figure 14.8.17, a modern vapor defluxing process can be described as follows. The cleaning equipment contains a cylindrical process chamber within which the complete cleaning, rinsing and drying process takes place. The process chamber may include spray manifolds and/or ultrasonic transducers (frequency of 40kHz) to provide the mechanical energy necessary for advanced precision cleaning. Following the final vapor rinse, remaining vapors are passed through a heat exchanger to condense the vapors to liquid, which is returned to the solvent recovering device. The vacuum then extracts any remaining solvent molecules, which are either exhausted from the machines directly or passed through carbon absorption system in order to limit solvent emissions to extremely low levels. The use of vacuum distillation for solvent recovery and vacuum drying assures minimal solvent consumption. Distillable solvents with a flash point above 55°C like Zestron® VD by Dr.O.K. Wack Chemie are suitable for such a process. Many benefits provided by solvent cleaning are evident. Among these are: • no process chemistry to mix or maintain, • solvent is reclaimed automatically in the cleaning equipment, • only few process variables to be managed, • very easy and cost-effective to operate, and • typically small footprint of the cleaning equipment. 14.8.5 CRITERIA FOR ASSESSMENT AND EVALUATION OF CLEANING RESULTS There are a number of different processes that can be customized to monitor the cleaning process and cleaning results. Depending upon the existing requirements, simple “qualitative tests” can be applied to produce a “Good” or “Poor” rating, or more elaborate quantita-
918
Martin Hanek, Norbert Löw, Andreas Mühlbauer
tive tests may be required to furnish a more in-depth statement of regarding the condition of the cleaning bath. Refractive index: An optical method by which a change of the refractive index of liquids, depending upon the content of liquid or solid contaminants, is used to measure the level of bath contamination. Solid residue from evaporation: This method establishes the non-volatile residues in cleaning and rinsing baths. This produces a very accurate statement of the bath contamination level. Electrochemical potential: The electrochemical potential is determined by the pH value, which furnishes an indirect statement of the level of contamination. Conductivity measurement: This is used to check the quality of the rinsing bath in assembly cleaning processes. The conductivity measurement indicates the amount of process-related ionic, and therefore conductive, contaminants carried over into the rinse medium. Ionic contamination measurement: Ionic contamination-expressed in sodium chloride equivalents is primarily used to assess the climatic resistance of electronic assemblies. Ionic measurement is suitable for statistical process control, but it does not furnish an absolute statement concerning the climatic resistance of an assembly. Ink test: Most organic residues on assemblies are not detected by ionic contamination meters. However, such residues do change the surface tension of assemblies, thereby influencing the adhesion of subsequent coatings. This method determines the surface tension by iterative use of corresponding testing inks. Furthermore, the remaining protective ability of organic solder (OSP-coatings) on copper can be estimated. This test also provides an indication of the adhesion of lacquers and conformal coatings to the assembly. Water-immersion test: This economical test exposes the climatically unstable points of electronic components. Due to the nature of the test, the entire board is evaluated. This test accelerates the mechanisms of electrochemical migration. Consequently, faults that previously would appear after months or even years can be detected during the development process. To identify potential weak points, the assembly is operated in standby mode and immersed in deionized water. Testing while the assembly is in full operation is even more effective. The sensitivity of the circuit to moisture exposure is assessed on the basis of the recorded test current, combined with a subsequent examination of the assembly. Through weak point analysis, a Yes/No decision can be determined concerning the expected service life, of the assembly. The growing diversity of cleaning media, cleaning equipment, and cleaning processes, together with more stringent legal requirements and quality expectations, are making it increasingly difficult for companies to develop their own cost-effective and result-optimized cleaning processes. Consequently, a lot of well known manufacturing companies opted for turnkey solutions for the entire development and installation of complete cleaning processes.
14.8 Electronic industry
919
14.8.6 COST COMPARISON OF DIFFERENT CLEANING PROCESSES Long-term cost analysis experience on different cleaning processes have been summarized in Figure 14.8.18 which describes the overall process costs of the different cleaning applications discussed. Whereby the overall process costs are divided into four main sections: • Investment costs, • Costs for the cleaning Figure 14.8.18. Comparison of overall process costs between different cleaning agent, processes. • Costs for the disposal of used media/ recycling cost, and • Energy costs. By analyzing the numbers, the following key conclusions on the different cleaning processes are substantial: • The total annual process costs are always connected with the throughput, but not necessarily in a linear function. • The lowest investment costs can be achieved by using an aqueous-based or NovecTM HFE-cosolvent cleaning process. • Closed vapor degreasing and NovecTM HFE-cosolvent processes are more competitive regarding waste disposal. • A water-based cleaning process is the most competitive regarding media costs. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
C. J. Tautscher, Effects on Electronic Products, Marcel Dekker Inc., New York, 1991. M. Hanek, System integration in Micro Electronics, Tutorial 4, Nürnberg, May 4-6, 1999. N. Loew, EPP Europe, 10 ,44 (1999). M. Hanek, R. Schreinert, Galvanotechnik, 83, 1370 (1992). F. Cala, A. E. Winston, Handbook of Aqueous Cleaning Technology for Electronic Assemblies, Elektrochemical Publications Ltd., Ashai House, Church Road, Port Erin, Isle of Man, British Isles, 1996. N. Loew, EPP Europe, 1/2, 24 (2000). S. J. Sackinger, F. G. Yost, Material Technology, 9, 58 (1994). J.M. Price, Proceedings Nepcon East`94, 365 (1994). B. N. Ellis, Cleaning and Contamination of Electronics Components and Assemblies, Electrochemical Publications Ltd, Port Erin, Isle of Man, British Isles, 1986. B. N. Ellis, Circuit World, 19, 1, (1992). http:\\www.Kyzen.com. M. Hanek, N. Loew, Unveröffentliche Ergebnisse, Dr. Wack Chemie, Ingolstadt, Germany, 1999. M. E. Hayes, R. E. Miller, US Patent No. 4,640,719, 1987. L.Futch, K. R. Hrebenar, M. E. Hayes, US Patent No. 4,934,391, 1990. M. Hanek, O. K. Wack, Patent No. 0587917, 1995. K. Lessmann, Proceedings, Nepcon West `94, 2219 (1994). H. Gruebel, Galvanotechnik, 87, (1996). M. Hanek, Productronic, 11, 39, 1997. P.Schlossarek, Surface Mount Technology, 7, (1994).
920
20 21 22
George Wypych N. LØw, EPP Europe, 8, 42 (1999).. N. LØw, P. Bucher, Surface Mount Technology, 6, 36 (1998). J. Owens, K. Warren, H. Yanome, D. Mibrath, International CFC Halone Alternatives Conference, Washington DC, October 1995.
14.9 FABRICATED METAL PRODUCTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Figure 14.9.1. Schematic diagram of operations in fabricated metal products manufacturing process. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Fabricated Metal Products Industry. US Environmental Protection Agency, 1995.]
14.9 Fabricated metal products
921
Table 14.9.1. Reported solvent releases from the metal fabricating and finishing facilities in 1993 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
acetone
870,400
2-methoxyethanol
22,800
benzene
1,800
methyl ethyl ketone
4,540,000
n-butyl alcohol
495,000
methyl isobutyl ketone
934,000
sec-butyl alcohol
13,600
tetrachloroethylene
844,000
cyclohexanone
303,000
toluene
3,015,000
1,2-dichlorobenzene
10,900
1,1,1-trichloroethane
2,889,000
dichlorofluoromethane
10,800
trichloroethylene
3,187,000
dichloromethane
1,350,000
trichlorofluoromethane
82,000
ethylbenzene
299,000
1,2,4-trimethylbenzene
262,000
ethylene glycol
99,000
xylene (mixture)
4,814,000
2-ethoxy ethanol
18,600
m-xylene
8,800
isopropyl alcohol
23,400
o-xylene
17,000
methanol
199,000
p-xylene
10
Figure 14.9.1 shows a schematic diagram of the operations in a fabricated metal products manufacturing process.1 The diagram shows that solvent emissions, waste waters, and solid wastes are produced from three operations: the metal cutting and forming process, metal cleaning, and painting. In metal cutting and forming the major solvents used are 1,1,1-trichloethane, acetone, toluene, and xylene. In surface cleaning, the straight solvents are used for cleaning or some solvents such as kerosene or glycols are emulsified in water and the emulsion is used for cleaning. When emulsions are used the amount of solvent is decreased. Solvents are released due to evaporation, volatilization during storage, and direct ventilation of fumes. Waste waters are generated from rinse waters. These waste waters are typically cleaned on-site by conventional hydroxide precipitation. Solid wastes are generated from waste water cleaning sludge, still bottoms, cleaning tank residues, and machining fluid residues. In painting, two methods are used: spray painting and electrodeposition. Painting operations release benzene, methyl ethyl ketone, methyl isobutyl ketone, toluene, and xylene. Paint cleanup operations also contribute to emissions and waste generation. The predominant solvents used for equipment cleaning are tetrachloride, dichloromethane, 1,1,1-trichloroethane, and perchloroethylene. Sources of solid and liquid waste include components of emission controlling devices (e.g., paint booth collection system and ventilation filters), equipment washing, paint disposal, overspray, and excess paint discarded after expiration of paint’s shelf-life. Table 14.9.1 shows reported solvent releases from metal fabricating and finishing facilities and Table 14.9.2 shows reported solvent transfers from the same industry. Xylene, methyl ethyl ketone, trichloroethylene, 1,1,1-trichloroethane, and dichloromethane are released in the largest quantities. Methyl ethyl ketone, xylene, toluene,
922
George Wypych
Table 14.9.2. Reported solvent transfers from the metal fabricating and finishing facilities in 1993 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
acetone
767,000
methyl ethyl ketone
5,130,000
n-butyl alcohol
282,000
methyl isobutyl ketone
846,000
sec-butyl alcohol
495
tetrachloroethylene
415,000
cyclohexanone
1,025
toluene
1,935,000
dichloromethane
237,000
1,1,1-trichloroethane
920,000
ethylbenzene
263,000
trichloroethylene
707,000
ethylene glycol
68,000
1,2,4-trimethylbenzene
42,000
2-ethoxy ethanol
1,800
xylene (mixture)
2,335,000
isopropyl alcohol
41,000
m-xylene
27,000
methanol
138,000
o-xylene
37
2-methoxyethanol
7,800
p-xylene
23
trichloroethylene, and methyl isobutyl ketone are transferred in largest quantities outside industry for treatment and disposal. The metal fabrication and finishing industry’s contribution to reported releases and transfers in 1995 was 9.3% of the total for 21 analyzed industries. This industry is the fourth largest contributor to total reported releases and transfers. Among the potential actions to reduce pollution, these are the most effective: recycling of solvents, employing better waste control techniques, and substituting raw materials. For solvent cleaning, several techniques are suggested. Employees should be required to obtain solvent from their supervisor. It is estimated that this will reduce waste by 49%. Pre-cleaning (wipe, blow part with air, etc.) should be adopted or the parts should be first washed with used solvent. Equipment for vapor degreasing should be modified by increasing its head space, covering degreasing units, installing refrigerator coils, rotating parts before removing, installing thermostatic controls, and adding filters. Trichloroethylene and other halogenated solvents should be replaced with liquid aqueous alkali cleaning compounds. Other waste reduction can be realized by the better production scheduling, use of dry filters in the spray booth, prevention of leakage from the spray gun, separation of wastes, and recycling of solvents by distillation.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Fabricated Metal Products Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.10 Food industry
923
14.10 FOOD INDUSTRY - SOLVENTS FOR EXTRACTING VEGETABLE OILS Phillip J. Wakelyn National Cotton Council, Washington, DC, USA
Peter J. Wan USDA, ARS, SRRC, New Orleans, LA, USA
14.10.1 INTRODUCTION Many materials including oils, fats, and proteins, for both food and nonfood use, are recovered from diverse biological sources by solvent extraction.1,2 These materials include animal tissues (e.g., beef, chicken, pork and fish); crops specifically produced for oil or protein (e.g., soy, sunflower, safflower, rape/canola, palm, and olive); by-products of crops grown for fiber (e.g., cottonseed and flax); food (e.g., corn germ, wheat germ, rice bran, and coconut); confections (e.g., peanuts, sesame, walnuts, and almonds); nonedible oils and fats (castor, tung, jojoba); and other oil sources (oils and fats from microbial products, algae, and seaweed). There are many physical and chemical differences between these diverse biological materials. However, the similarities are that oils (edible and industrial) and other useful materials (e.g., vitamins, nutriceuticals, fatty acids, phytosterols, etc.) can be extracted from these materials by mechanical pressing, solvent extracting, or a combination of pressing and solvent extraction. The preparation of the various materials to be extracted varies. Some need extensive cleaning, drying, fiber removal (cottonseed), dehulling, flaking, extruding, etc., all of which affect the solvent-substrate interaction and, therefore, the yield, composition, and quality of the oils and other materials obtained. Historically, the advancement of processing technology for recovering oils and other useful materials has been primarily driven by economics. For thousands of years stone mills, and for several centuries simple hydraulic or lever presses were used as batch systems. The continuous mechanical presses only became reality during the early 1900s. It was not until the 1930s, that extraction solvents were used more widely, which greatly enhanced the recovery of oil from oilseeds or other oil bearing materials. In solvent extraction crude vegetable oil and other useful materials are dissolved in a solvent to separate them from the insoluble meal. Many solvents have been evaluated for commercial extraction. Commercial hexane has been the main solvent for the oilseed processing industry since the 1940s1,2 because of its availability at reasonable cost and its suitable functional characteristics for oil extraction. However, the interest in alternative solvents to hexane has continued and is motivated by one of a combination of factors: desire for a nonflammable solvent, more efficient solvent, more energy efficient solvent, less hazardous and environmentally friendly solvent, solvent with improved product quality, and solvents for niche/specialty markets. Today commercial isohexane (hexane isomers) is replacing commercial hexane in a few oilseed extraction operations and other solvents (e.g., isopropanol, ethanol, acetone, etc.) are also being used for various extraction processes or have been evaluated for use as extracting solvents.3-6 With greater flexibility of operating hardware and availability of various solvents with tailored composition, the oilseed industry does have expanded options to choose the unique composition of solvents to obtain the desired final products. While the availability, cost,
924
Phillip J. Wakelyn, Peter J. Wan
ease-of-use (i.e., no specialty processing hardware required and with low retrofitting costs), and acceptable product quality continue to be the principle factors used to choose a solvent for oil extraction, in recent years environmental concerns and health risk have become increasingly important criteria to be considered in selecting an oil extraction solvent. This chapter presents information on solvents for extracting oilseeds and other biological materials for oils, fats, and other materials. 14.10.2 REGULATORY CONCERNS Many workplace, environmental, food safety, and other regulations (see Tables 14.10.1 and 14.10.2 for summary information on U.S. laws and regulations) apply to the use of solvents for the extraction of oilseeds and other diverse biological materials. Some solvent may have fewer workplace and environmental regulatory requirements; offer benefits, such as an improved product quality and energy efficiency;7 as well as other advantages, such as the ability to remove undesirable (e.g., gossypol from cottonseed) and desirable (e.g., vitamin E, tocopherols, lecithins, phytosterols, and long chain polyunsaturated fatty acids) constituents of oilseeds and other biological materials more selectively. Some of the regulations required in the United States are discussed. Many other countries have similar requirements, but if they do not, it would be prudent to consider meeting these regulations in solvent extraction and to have environmental, health and safety, and quality management programs.19,44 Table 14.10.1. U.S. Worker health and safety laws and regulations Laws: Occupational Safety and Health Act of 1970 (OSHA) (PL 91-596 as amended by PL 101 552; 29 U.S. Code 651 et. seq.) OSHA Health Standards: Air Contaminants Rule, 29 CFR 1910.1000 Hazard Communication Standard, 29 CFR 1910.1200 OSHA Safety Standards: Process Safety Management, 29 CFR 1910.119 Emergency Action Plan, 29 CFR 1910.38(a)(1) Fire Prevention Plan, 29 CFR 1910.38(b)(1) Fire Brigades, 29 CFR 1910.156 Personal Protection Equipment (General Requirements 29 CFR 1910.132; eye and face protection, .133; respiratory protection, .134)
Table 14.10.2. U.S. Environmental laws and regulations Law or Regulation
Purpose
Environmental Protection Agency (EPA) (established 1970)
To protect human health and welfare and the environment
Clean Air Act (CAA) (42 U.S. Code 7401 et seq.)
To protect the public health and welfare. Provides EPA with the authority to set NAAQS, to control emissions from new stationary sources, and to control HAP.
Federal Water Pollution Control Act (known as the Clean Water Act) (CWA) (33 U.S. Code 1251 et seq.)
The major law protecting the “chemical, physical and biological integrity of the nation’s waters." Allows the EPA to establish federal Limits on the amounts of specific Pollutants that can be released by municipal and industrial facilities.
14.10 Food industry
925
Toxic Substances Control Act (TSCA) (15 U.S. Code 2601 et seq.)
Provides a system for identifying and evaluating the environmental and health effects of new chemicals and chemicals already in commerce.
Resource Conservation and Recovery Act (RCRA) (42 U.S. Code 6901 et seq.)
A system for handling and disposal of non-hazardous and hazardous waste.
Comprehensive Environmental Response, Compensation and Liability Act (CERCLA) (42 U.S. Code 9601 et seq.)
Known as “Superfund”, gives the EPA power to recover costs for containment, other response actions, and cleanup of hazardous waste disposal sites and other hazardous substance releases.
Emergency Planning and Community Right-to-Know Act (EPCRA; also “SARA 313") (42 U.S. Code 1101 et seq.)
(Part of Superfund) Provides authority for communities to devise plans for preventing and responding to chemical spills and release into the environment; requires public notification of the types of hazardous substances handled or release by facilities; requires state and local emergency plans.
14.10.2.1 Workplace regulations Workplace regulations (see Table 14.10.1) are promulgated and enforced in the U.S. by the Occupational Safety and Health Administration (OSHA), which is in the Department of Labor. The purpose of OSHA is to ensure that the employers maintain a safe and healthful workplace. Several workplace standards that affect extraction solvents are discussed. 14.10.2.1.1 Air Contaminants Standard (29 CFR 1910.1000) The purpose of the air contaminants standards are to reduce risk of occupational illness for workers by reducing permissible exposure limits (PEL) for chemicals. Table 14.10.3 lists the PELs [8-hr time-weighted average (TWA) exposure] for the solvents discussed. To achieve compliance with the PEL, administrative or engineering controls must first be determined and implemented, whenever feasible. When such controls are not feasible to achieve full compliance, personal protective equipment, work practices, or any other protective measures are to be used to keep employee exposure below the PEL. Table 14.10.3. U.S. Workplace regulations,a air contaminants Chemical Name (CAS No.)
Permissible Exposure Limit (PEL) [Health Risk: Basis for the PEL]
n-Hexane (110-54-3)
500 ppm/1800 mg/m3; new PEL was 50 ppm/180 mg/m3 same as ACGIH (TLV); [neuropathy]
Commercial hexaneb (none)
(Same as n-hexane)
n-Heptane (142-82-5)
500 ppm/200 mg/m3; new PEL was 400 ppm/1640 mg/m3, (500 ppm STEL) same as ACGIH (TLV); [narcosis]
Cyclohexane (110-82-7)
300 ppm/1050 mg/m3; ACGIH (TLV) 300 ppm/1030 mg/m3; [sensory irritation]
Cyclopentane (287-92-3)
None; new PEL was 600 ppm, same as ACGIH (TLV); [narcosis]
Hexane isomers
None; new PEL was 500 ppm/1760 mg/m3 (1000 ppm STEL) same as ACGIH (TLV); [narosis]
Commercial isohexanec (none)
(Same as hexane isomer)
926
Phillip J. Wakelyn, Peter J. Wan
Chemical Name (CAS No.)
Permissible Exposure Limit (PEL) [Health Risk: Basis for the PEL]
2-Methyl pentane (isohexane) (2-MP) (107-83-5)
(Same as hexane isomer)
3-Methyl pentane (3-MP) (96-14-0)
(Same as hexane isomer)
Methyl cyclopentane (MCP) (96-37-7)
(Same as hexane isomer)
2,2 Dimethyl butane (neohexane) (2,2-DMB) (75-83-2)
(Same as hexane isomer)
2,3 Dimethyl butane (2,3-DMB) (79-29-8)
(Same as hexane isomer)
Methyl cyclohexane (108-87-2)
500 ppm; new PEL was 400 ppm/1610 mg/m3, same as ACGIH (TLV); [narcosis]
Isopropyl alcohol (2-propanol) (IPA) (67-17-5)
400 ppm/980 mg/m3; ACGIH (TLV) same plus 500 ppm/1230 mg/m3 STEL; [sensory irritation]
Ethyl alcohol (ethanol) (64-17-5)
1000 ppm/1880 mg/m3; ACGIH (TLV) same; [narcosis, irritation]
Acetone (67-64-1)
1000 ppm/2400 mg/m3; ACGIH (TLV) 750 ppm (1000 ppm STEL); [sensory irritation]
a
CAS No. is the Chemical Abstracts Service Registry Number; PEL is from 29 CFR 1910.1000, Table Z-1; American Conference on Governmental Industrial Hygienists (ACGIH), threshold limit value (TLV); under the HCS, a MSDS is required for all of the compounds (physical and/or chemical hazard); all of the solvents are flammable liquids or gasses, under the OSHA definition, and are regulated under the PSM Standard. b Commercial hexane as used in the U.S. is usually about 65% n-hexane, and the rest is hexane isomers (e.g., methyl cyclopentane (MCP), 2-methyl pentane (2-MP), and 3-methyl pentane (3-MP)), and it contains less than 10 ppm benzene. c Mixture of 2-MP (45-50%), 3-MP, 2,2-DMB, and 2,3-DMB. (ref 7)
In the case of a mixture of contaminants, an employer has to compute the equivalent exposure when the components in the mixture pose a synergistic threat (toxic effect on the same target organ) to worker health.8,9 14.10.2.1.2 Hazard Communication Standard (HCS) (29 CFR 1910.1200) The HCS requires information on hazardous chemicals to be transmitted to employees through labels, material safety data sheets (MSDS), and training programs. A written hazard communications program and record keeping are also required. A substance is a “hazardous chemical” if it is a “physical hazard” or a “health hazard”. A flammable or explosive liquid is a “physical hazard”. A flammable liquid means “any liquid having a flash point below 110oF (37.8oC), except any mixture having components with flash points of 100oF (37.8oC) or higher, the total of which make up 99% or more of the total volume of the mixture”. “Health hazard” means “a chemical for which there is statistically significant evidence based on at least one valid study that acute or chronic health effects may occur in exposed employees”. Hexane and all the solvents listed in Table 14.10.3 would require a MSDS, since all are flammable liquids (physical hazards) as defined by OSHA and/or possible health hazards because all, except hexane isomers, have an U.S. OSHA PEL. However, hexane isomers have an American Conference of Industrial Hygien-
14.10 Food industry
927
ist (ACGIH) threshold limit value (TLV),10 which many states and countries enforce as a mandatory standard. Chemical manufacturers and importers are required to review the available scientific evidence concerning the hazards of chemicals they produce or import, and to report the information to manufacturing employers who use their products. If a chemical mixture has not been tested as a whole to determine whether the mixture is a hazardous chemical, the mixture is assumed to present the same hazards as do the components that comprise 1% or greater of the mixture or a carcinogenic hazard if it contains a component in concentration of 0.1% or greater that is a carcinogen. Commercial hexane containing 52% n-hexane has been tested and found not to be neurotoxic unlike pure n-hexane.11-13 So mixtures with less than 52% n-hexane should not be considered to be a neurotoxin, although n-hexane would have to be listed on the MSDS, if in greater quantity than 1% of the mixture. 14.10.2.1.3 Process Safety Management (PSM) Standard (29 CFR 1910.119) PSM is for the prevention or minimization of the consequences of catastrophic releases of toxic, reactive, flammable, or explosive chemicals. This regulation applies to all processes that involve one or more of 137 listed chemicals (29 CFR 1910.119, Appendix A) above their threshold quantities or have 10,000 lbs. or more of a flammable liquid or gas, as defined by the U.S. OSHA HCS [29 CFR 1910.1200(c)]. This includes n-hexane, hexane isomers, and all solvents listed in Table 14.10.3. In addition to the PSM standard, U.S. OSHA has been enforcing two other regulations for operations/processes with flammable liquids. First, under Personal Protective Equipment - General Requirements (29 CFR 1910.132), OSHA has cited or obtained voluntary agreement from organizations relative to flame-resistant (FR) clothing. Operators and other employees working in the area of a flammable process are being required to wear flame-resistant work clothing. For facilities that use hexane and other flammable solvents, it would be prudent to require FR clothing for all personnel working in areas where there is an exposure to a flammable liquid. Second, OSHA has cited organizations for failure to meet related safety regulations under Fire Brigades (29 CFR 1910.156), specifically for standards such as: training, both initial and annual refresher training; protective equipment availability and testing; and fitness for duty including periodic physicals. If an on-site fire brigade is part of the site’s Emergency Response Plan (29 CFR 1910.38), then these requirements must also be met. In addition, the requirement of the PSM standard for an Emergency Response Plan triggers the requirements of Emergency Action Plan [29 CFR 1910.38(a)]. 14.10.2.2 Environmental regulations The role of the U.S. Environmental Protection Agency (EPA) is to protect human health and welfare and the environment. The U.S. EPA administers all regulations affecting the environment and chemicals in commerce. The individual states and state environmental regulatory control boards implement and enforce most of the regulations. The legislation that serves as the basis for the regulations can be divided into: statutes that are media-specific [Clean Air Act (CAA) and Clean Water Act (CWA)]; statutes that manage solid and hazardous waste [Resources Conservation and Recovery Act (RCRA) and Comprehensive Environmental Response, Compensation and Liability Act (CERCLA; “Superfund”)]; and, statutes that directly limit the production rather than the release of chemical substance [Toxic Substances Control Act (TSCA) and Federal Insecticide, Fungicide and Rodenicide Act (FIFRA)]. See Table 14.10.2 for a summary of the information on environmental laws
928
Phillip J. Wakelyn, Peter J. Wan
and regulations, and Tables 14.10.4 and 14.10.5 for an overview of environmental requirements that apply to each solvent. Table 14.10.4. U.S. Environmental regulations, air and water VOC
HAP
CWAa
Sol. in H2O
n-Hexane (110-54-3)
Yes
Yes
Yes
I
Commercial hexane(none)
Yes
n-Heptane (148-82-5)
Yes
No
Yes
I
Cyclohexane (110-82-7)
Yes
No
Yes
0.01
Cyclopentane (287-92-3)
Yes
No
Yes
I
Hexane isomers (none)
Yes
No
Yes
Commercial isohexane (none)
(same as hexane isomers)
No
Yes
2-Methyl pentane (isohexane) (107-83-5)
(a hexane isomer) Yes
No
Yes
0.0014
3-Methyl pentane (96-14-0)
(a hexane isomer) Yes
No
Yes
0.0013
Methyl cyclopentane (96-37-7)
(a hexane isomer) Yes
No
Yes
0.0013
2,2 dimethyl butane (neohexane) (75-83-2)
(a hexane isomer) Yes
No
Yes
0.0018
2,3 dimethyl butane (79-29-8)
(a hexane isomer) Yes
No
Yes
0.0011
Methyl cyclohexane(108-87-2)
Yes
No
Yes
0.0014
Isopropyl alcohol (2-propanol) (67-17-5)
Yes
No
Yes
Misc.
Ethyl alcohol (ethanol) (64-17-5)
Yes
No
Yes
Misc.
b
No
Yes
Misc.
Chemical Name (CAS No.)
Acetone (67-64-1)
No
a
Under the Clean Water Act there could be stormwater and NPDES permit requirements; none of the solvents are listed as priority toxic pollutants in 40 CFR 401.15. b Acetone is considered by the U.S. EPA not be a VOC (60 FR 31643; June 16, 1995) Abbreviations: CAS No., Chemical Abstracts Service Registry number; VOC, volatile organic chemical; HAP, hazardous air pollutant; CWA, Clean Water Act; I, insoluble in H2O; MISC., miscible in H2O; and 0.01, 0.01 parts soluble in 100 parts H2O. Source for water solubility: Ref 47.
Table 14.10.5 U.S. Environmental regulations, waste Chemical Name (CAS No.) n-hexane (110-54-3)
a
(RCRA) RCRA Codeb
(EPCRA/SARA Title III)
Sec. 304 CERCLA RQ
Sec.311/312
Sec.313 (TRI)
5000c
Yes
Yes Yesd
Commercial hexane (none) n-heptane (142-82-5)
Yes
No
14.10 Food industry
Chemical Name (CAS No.)
929
a
(RCRA)
(EPCRA/SARA Title III)
RCRA Codeb
Sec. 304 CERCLA RQ
Sec.311/312
Sec.313 (TRI)
U056
1000
Yes
Yes
Cyclopentane (287-92-3)
Yes
No
Hexane isomers (none)
Yes
Noe
Cyclohexane (110-87-7)
Noe
Commercial isohexane (none) 2-Methyl pentane (also called isohexane) (107-83-5)
Yes
Noe
3-Methyl pentane (96-14-0)
Yes
Noe
Methyl cyclopentane (96-37-7)
Yes
Noe
2,2 Dimethyl butane (neohexane) (75-83-2)
Yes
Noe
2,3 Dimethyl butane (79-29-8)
Yes
Noe
Methyl cyclohexane (108-87-2)
Yes
No
Isopropyl alcohol (2-propanol) (67-63-0)
Yes
Nof
Ethyl alcohol (64-17-5)
Yes
No
Yes
No
Acetone (67-64-1)
U002
5000
a
From Title III Lists of Lists, U.S. EPA, EPA 740-R-95-001 (April 1995); 40 CFR 52-99; (59 FR 4478; January 31, 1994) hexane added to TRI list; (60 FR 31633; June 16, 1995) acetone removed from TRI list. b 40 CFR 261.33, listed hazardous waste - EPA RCRA Hazardous Waste Number. All the solvents that are on the RCRA list are listed because of Section 3001 of RCRA (part for identification and listing of hazardous waste) except hexane which is on because of CAA Section 112 (HAP). c RQ for hexane finalized June 12, 1995 (60 FR 30939). d Only the amount of commercial hexane that is n-hexane has to be reported (e.g., if the commercial hexane is 62% n-hexane, only 62% of the emissions have to be reported for TRI). e The EPA clarrified that the listing for hexane was only for n-hexane, other isomers of hexane are not included. (59 FR 61457; Nov. 30, 1994). f The EPA has indicated (62 FR 22318, April 25, 1997) that IPA itself does not meet the criteria for listing on the TRI list. The EPA will remove IPA from the TRI list. Abbreviations: RQ, reportable quantity in pounds/24 h.
14.10.2.2.1 Clean Air Act (CAA; 42 U.S. Code 7401 et seq.) To satisfy the CAA requirements, states and state air control boards are required to implement regulations and develop state implementation plans (SIP).14,15 Criteria pollutants (e.g., ozone, oxides of nitrogen, carbon monoxide) are regulated with National Ambient Air Quality Standards (NAAQS) and hazardous air pollutants (HAP), such as hexane, with National Emissions Standards for Hazardous Air Pollutants (NESHAP). The NAAQS are set at levels sufficient to protect public health (primary air quality standards) and welfare (secondary air quality standards; “welfare effects” include wildlife, visibility, climate, damage to and deterioration of property and effects on economic value and on personal comfort and well being) from any known or anticipated adverse effect of the pollutant with an adequate (appropriate) margin of safety. The 1990 CAA expanded the list of HAP to 188, including hexane, and more strictly regulates nonattainment areas for
930
Phillip J. Wakelyn, Peter J. Wan
criteria pollutants such as ozone, particulate matter, carbon monoxide and oxides of nitrogen. NAAQS: Volatile organic compounds (VOC) are essentially considered the same as the criteria pollutant ozone.16,17 VOCs are very broadly defined by the U.S. EPA (40 CFR 51.100): any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates, and ammonium carbonate, that participates in atmospheric photochemical reactions. This includes any organic compound other than those specifically listed as having been determined to have negligible photochemical reactivity. Reactive VOCs are essentially all those judged to be clearly more reactive than ethane - the most reactive member of the “negligibly reactive” class. C4 - C6 paraffins are of relatively low kinetic reactivity but produce NO2, and potentially ozone.18 Hexane and all of the solvents discussed, except acetone, would be considered VOCs (see Table 14.10.4) that can undergo photochemical oxidation in the atmosphere to form ozone. In the U.S., acetone was added to the list of compounds excluded from the definition of VOCs in 1995, because it was determined to have negligible photochemical reactivity.19 Most U.S. vegetable-oil extracting facilities would be major sources of VOCs and would be covered by the requirements for ozone emissions and attainment, unless they used a solvent that was not classified as a VOC. (The definition of “major source” changes as the severity of the ozone nonattainment area increases. Plants in marginal and moderate areas are major if they emit 100 tons VOC/yr; in serious areas, 50 tons/yr; in severe areas, 25 tons/yr; and in extreme areas 10 tons/yr). All facilities in ozone non-attainment areas could be required to reduce emissions through implementing Reasonable Available Control Measures (RACM) standards or Best Available Control Measures (BACM). Any new or significantly modified facility would have to comply with the new source review (NSR) requirements and prevention of significant deterioration (PSD) requirements. Hazardous air pollutants (HAP) or air toxics: If a facility is a major emitter of any of the chemicals on the CAA list of HAPs, EPA requires sources to meet emissions standards.14,16,17 n-Hexane is on the HAP list but isohexane, acetone and other solvents listed in Table 14.10.4 are not. The air toxic requirements of the CAA for establishing control measures for source categories are technology-based emission standards (not health based) established for major sources (10 tons/yr of one HAP or 25 tons/yr of total HAP) that require the maximum degree of reduction emissions, taking costs, other health and environmental impacts, and energy requirements into account. Standards are set based on known or anticipated effects of pollutants on the public health and the environment, the quantity emitted, and the location of emissions. Compliance involves the installation of Maximum Achievable Control Technology (MACT) - MACT essentially is maximum achievable emission reduction. For new sources, MACT standards must be no less stringent than the emission control achieved in practice by the best controlled similar source. The MACT standards for vegetable oil processing using n-hexane are expected to be issued in 2001. Once a standard has been promulgated for a source category, a source will have three years after the due date to comply. The requirements cover normal operations and startup, shutdown, and malfunction (SSM). The allowable emissions for solvent extraction for vegetable oil production in the U.S., as a 12-month rolling average based on a 64% n-hexane content, will vary from 0.2 gal/ton to greater than 0.7 gal/ton depending on the oilseed (65FR34252; May 26, 2000) (see Table
14.10 Food industry
931
12.14.6). There are also variable emission requirements depending on the oilseed for allowable emissions for solvent extraction for vegetable oil production in Europe. Table 14.10.6. Oilseed solvent loss factors for allowable HAP loss (12-mo. rolling ave.)
Type of Oilseed Process
A source that...
Oilseed Solvent Loss Factor (gal/ton) Existing Sources
New Sources
Corn Germ, Wet Milling
processes corn germ that has been separated from other corn components using a wet process of centrifuging a slurry steeped in a dilute sulfurous acid solution
0.4
0.3
Corn Germ, Dry Milling
processes corn germ that has been separated from the other corn components using a dry process of mechanical chafing and air sifting
0.7
0.7
Cottonseed, Large
processes 120,000 tons or more of a combination of cottonseed and other listed oilseeds during all normal operating periods in a 12 operating month period
0.5
0.4
Cottonseed, Small
processes less than 120,000 tons of a combination of cottonseed and other listed oilseeds during all normal operating periods in a 12 operating month period
0.7
0.4
Flax
processes flax
0.6
0.6
Peanuts
processes peanuts
1.2
0.7
Rapeseed
processes rapeseed
0.7
0.3
Safflower
processes safflower
0.7
0.7
Soybean, Conventional
uses a conventional style desolventizer to produce crude soybean oil products and soybean animal feed products
0.2
0.2
Soybean, Specialty
uses a special style desolventizer to produce soybean meal products for human and animal consumption
1.7
1.5
Soybean, Small Combination Plant
processes soybeans in both specialty and conventional desolventizers and the quantity of soybeans processed in specialty desolventizers during normal operating periods is less than 3.3 percent of total soybeans processed during all normal operating periods in a 12 operating month period. The corresponding solvent loss factor is an overall value and applies to the total quantity of soybeans processed
0.25
0.25
Sunflower
processes sunflower
0.4
0.3
A health-based standard would be for a boundary line level of a solvent (e.g., n-hexane) based on the inhalation reference concentration (RfC).20 The current RfC for n-hexane
932
Phillip J. Wakelyn, Peter J. Wan
is 200µg/m3. Recent research suggests that the RfC for hexane should be at least 10 times higher (> 2000 µg/m3). Federal permits: All major sources of regulated solvents are required to have federally enforceable operating permits (FOP)14,15 (also referred to as Title V permits). State permits: Most states require state permits for facilities that emit listed air pollutants.14,15 In some states federal permits and state permits are combined, while in other states facilities are required to have both a state or county (air district) permit and a federal permit. As part of annual emission inventory reporting requirements, many states already require reporting of HAP and VOC because of their state implementation plan (SIP). 14.10.2.2.2 Clean Water Act (CWA; 33 U.S. Code 1251 et seq.) The CWA is the major law protecting the “chemical, physical and biological integrity of the nation’s waters.” Under it, the U.S. EPA establishes water-quality criteria used to develop water quality standards, technology-based effluent limitation guidelines, and pretreatment standards and has established a national permit program [National Pollution Discharge Elimination System (NPDES) permits; 40 CFR 122] to regulate the discharge of pollutants. The states have responsibility to develop water-quality management programs. For extraction solvents vegetable oil extracting facilities are covered by basic discharge effluent limitations [direct discharges to receiving waters or indirect discharges to publicly owned treatment works (POTW)], and stormwater regulations.15 The amount of solvent in effluent discharges and in stormwater (for those covered) needs to be determined and possibly monitored as part of an NPDES permit and as part of the visual examination or testing of stormwater quality. 14.10.2.2.3 Resource Conservation and Recovery Act (RCRA; 42 U.S.Code 6901 et seq.) RCRA subtitle C (40 CFR 261) is a federal “cradle-to-grave” system to manage hazardous waste (including provisions for cleaning up releases and setting statutory and regulatory requirements). Subtitle D covers nonhazardous wastes. Materials or items are hazardous wastes if and when they are discarded or intended to be discarded. The act requires generators, transporters, and disposers to maintain written records of waste transfers, and requires the U.S. EPA to establish standards, procedures, and permit requirements for disposal. The act also requires states to have solid waste management plans, prohibits open dumping, and requires the EPA to establish criteria for sanitary landfills. EPA under RCRA also regulates underground storage tanks that store or have stored petroleum or hazardous substances. Hazardous wastes are either listed wastes (40 CFR 261.30-.33) or characteristic wastes (40 CFR 261.21-.24). The U.S. EPA defines four characteristics for hazardous waste: ignitability (40 CFR 260.21); corrosivity (40 CFR 260.22); reactivity (40 CFR 260.23); and toxicity (40 CFR 260.24). Any waste that exhibits one or more of these characteristics is classified as hazardous under RCRA. The ignitability definition includes a liquid that has a flash point less than 60oC (140oF); the EPA included ignitability to identify wastes that could cause fires during transport, storage, or disposal (e.g., used solvents). All of the solvents in Table 14.10.5 have flash points less than 60oC, so all could be a RCRA ignitability waste.
14.10 Food industry
933
14.10.2.2.4 Emergency Planning and Community Right-to-Know Act (EPCRA; 42 U.S. Code 11001 et seq.) Enacted as Title III of the 1986 Superfund Amendments and Reauthorization Act (“SARA”), the Act mandates the EPA to monitor and protect communities regarding releases of chemicals into the environment. It requires states to establish emergency planning districts with local committees to devise plans for preventing and responding to chemical spills and releases. [“Superfund” is the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA) of 1980 that gives the U.S. EPA authority to force those responsible for hazardous waste sites or other releases of hazardous substances, pollutants, and contaminants to conduct cleanup or other effective response actions.] Section 304 (40 CFR 355.40): Facilities are subject to state and local reporting for accidental releases, in quantities equal to or greater than their reportable quantities (RQ), of extremely hazardous substances (EHS) or CERCLA hazardous substances (40 CFR 302, Table 302.4) under Section 304. n-Hexane, cyclohexane, acetone, and some of the other solvents discussed are CERCLA hazardous substances and have CERCLA RQ for spills (Table 14.10.5). Section 311, 312 (40 CFR 370.20-.21): Business must make MSDSs, for chemicals that are required to have an MSDS, available to state and local officials. Since all of the solvents discussed require MSDSs under the OSHA HCS, all are covered by these requirements. Section 313 (40 CFR 372), Toxic Release Inventory (TRI): Businesses are required to file annual reports with federal and state authorities of releases to air, water, and land above a certain threshold for chemicals on the TRI/Section 313 list (40 CFR 372.65) by July 1 each year for the previous year’s releases.21 TRI requirements are triggered if a facility is involved in manufacturing with 10 or more full-time employees, manufactures, processes, or otherwise uses with one or more listed substance(s) in a quantity above the statutory reporting threshold of 25,000 lbs./yr (manufactured or processed) or 10,000 lbs./yr (otherwise used). Beginning with the 1991 reporting year, such facilities also must report pollution prevention and recycling data for such chemicals pursuant to Section 6607 of the Pollution Prevention Act (42 U.S. Code 13106). n-Hexane was added to the TRI list in 1994 with reporting for 1995 emissions.19 The other solvents discussed are not on the TRI list. The EPA can add new chemicals to or delete chemicals from the TRI list as it deemed necessary and any person may petition the EPA to add chemicals or delete chemicals from the list. 14.10.2.2.5 Toxic Substances Control Act (TSCA; 15 U.S. Code 2601 et seq.) If a chemical’s manufacture, processing, distribution, use, or disposal would create unreasonable risks, the U.S. EPA, under the TSCA, can regulate it, ban it, or require additional testing. TSCA mandates the U.S. EPA to monitor and control the use of toxic substances by requiring the Agency to review the health and environmental effects of new chemicals [referred to as “Premanufacturing Notice” or “PMN”; Section 5(a)(1) of TSCA] and chemicals already in commerce. The U.S. EPA also has Significant New Use Rules (SNUR) under Section 5(a)(2) of TSCA which provides a way for the U.S. EPA to restrict uses of a chemical substance already in commerce that are proposed for new uses. All of the solvents discussed are already commercially available, so a PMN would not apply; some could be
934
Phillip J. Wakelyn, Peter J. Wan
subjected to SNUR (40 CFR 721, subpart A), since some are not presently being used as extraction solvents in large quantities. Under Section 4(a) of TSCA, the U.S. EPA can require testing of a chemical substance or mixture to develop data relevant for assessing the risks to health and the environment. Section 8(d) of TSCA requires that lists of health and safety studies conducted or initiated with respect to a substance or mixture be submitted to the U.S. EPA. All new toxicological data of the effects of a chemical not previously mentioned must be reported immediately if the data reasonably supports the conclusion that such substance or mixture presents a substantial risk of injury to health or the environment [Section 8(e) of TSCA]. Testing (Section 4 test rule) was required for several of the solvents earlier (e.g., commercial hexane for which new toxicological information was reported to the U.S. EPA since 1992),22 and any new toxicological information will have to be reported to the U.S. EPA under Section 8(e) and 8(d). 14.10.2.3 Food safety In the U.S. the use of a solvent to extract oil, that is a human food product or used in a food product, from oilseeds and biological materials falls under the rules and regulatory jurisdiction of the U.S. Food and Drug Administration (FDA), which regulates all aspects of food, including food ingredients and labeling in the U.S. In order to be legally used as an oilseed extraction solvent in the U.S., a substance must have been subject to an approval by the U.S. FDA or the U.S. Department of Agriculture (USDA) during 1938-1958 for this use (“prior sanction”); be generally recognized as safe (GRAS) for this use; or be used in accordance with food additive regulations promulgated by the U.S. FDA. Many prior sanctions and GRAS determinations are not codified in the U.S. FDA regulations. However, extracting solvents used in food manufacturing, such as n-hexane, have been labeled as a food additive, solvent, defoaming agent, component of a secondary food and color additives, minor constituent, or incidental additives (i.e., “additives that are present in a food at significant levels and do not have any technical or functional effect in that food”) depending on the application. Incidental additives can be “processing aids,” (i.e., “substances that are added to a food during processing but removed from the food before it is packaged”). Most food-processing substances, including solvents, can be regarded as “incidental additives” and thus are exempt from label declaration in the finished food product. Even if exempt from label declaration, all extraction solvents must be used in accordance with the U.S. FDA good manufacturing practices (GMP; 21 CFR 100). In the U.S., the Flavor and Extract Manufacturers’ Association (FEMA) has conducted a program since 1958 using a panel of expert pharmacologists and toxicologists to determine substances that are GRAS. This panel uses all available data, including experience based on common uses in food. This safety assessment program (“FEMA GRAS”) is widely accepted and considered an industry/government partnership with the U.S. FDA.23 A number of papers published in Food Technology since 196124,25 list the substances that the panel has determined to be GRAS and the average maximum levels in parts per million (ppm) at which each has been reported to be GRAS for different categories of food. The U.S. FDA has not incorporated these substances in their regulations but does recognize the findings of the Expert Panel of FEMA as GRAS substances. Since vegetable oil and other human food grade oils undergo deodorization (steam distillation) and other purification processes (i.e., refining and bleaching) as part of the manufacturing process prior to being used as a food product, they should not contain any of
14.10 Food industry
935
the extraction solvent, if proper manufacturing practices are followed. (see Section 14.10.3.3 Processing crude oil, for more details.) Refining removes free fatty acids and other non-oil compounds (e.g., phospholipids, color, and trace metals); bleaching with acid-activated bleaching earth or clay (e.g., bentonite), removes color-producing substances and residual soaps; and deodorization, the last major processing step in edible oils refining removes volatile compounds (undesirable ingredients occurring in natural oils and those that may be imparted by prior unit processes or even storage, many of which are associated with undesirable flavors and odors).26,27 Most commercial deodorizers operate at a temperature of 245-275oC (475-525oF) under a negative pressure of 2-10 mm Hg.26,27 It has been reported that no hexane residue remains in the finished oil after processing due to its high volatility.28 In addition, animal-feeding studies with expeller and solvent-extracted meals have not indicated any adverse health affects related to the extraction solvent.29 Hexane has been used since the 1940’s as an Figure 14.10.1a. Flow diagram of oilseed ex- oilseed-extraction solvent on the determination that traction process from seed to crude oil and it is GRAS and it may also be subject to a prior sancmeal. tion. However, like many other food-processing substances, there is no U.S. FDA regulation specifically listing hexane as GRAS or prior sanctioned. GRAS status may be determined by a company (“GRAS self-determination”), an industry, an independent professional scientific organization (e.g., FEMA GRAS), or the U.S. FDA. The Federal Food, Drug and Cosmetic Act (FFDCA; 21 U.S. Code 321 et seq.) does not provide for the U.S. FDA to approve all ingredients used in food, and the U.S. FDA explicitly recognizes that its published GRAS list is not meant to be a complete listing of all substances that are in fact GRAS food substances. Although there is no requirement to inform the U.S. FDA of a GRAS self-determination or to request FDA review or approval on the matter, the U.S. FDA has established a voluntary GRAS affirmation program under which such advice will be provided by the agency. Solvents that do not have prior sanction, a GRAS determination, or a tolerance set, probably should be evaluated for compliance under food safety requirements, if a facility is considering changing its extracting solvent or using a solvent for the extraction of the various biological materials for specialty markets. 14.10.3 THE SOLVENT EXTRACTION PROCESS Three types of processing systems are used to extract oil from oil-bearing materials: expeller pressing, prepress solvent extraction, and direct solvent extraction. Only prepress solvent extraction and direct solvent extraction, which remove the oil from the conditioned, prepared seed with an organic solvent, will be discussed here1,27 (see Figure 14.10.1). Oil-bearing materials have to be prepared for extraction to separate the crude oil from the
936
Phillip J. Wakelyn, Peter J. Wan
Figure 14.10.1b. Flow diagram of oilseed extraction process. Overview of extraction operation and identification of emission sources.
meal. Careful control of moisture and temperature during processing must be exercised to maintain the quality of the protein in the meal and to minimize the damage to the oil. Crude oils are refined by conditioning with phosphoric acid and treating with sodium hydroxide (alkali-refining) (see Figure 14.10.2). Refined oil is bleached with activated clay to remove color pigments. Bleached oils are then deodorized by steam distillation. The refined, bleached, and deodorized oil (RBD oil) is used to produce finished products, e.g., salad and cooking oils, shortening and margarine. Some of the finished products also require the oil to be hydrogenated, which changes the consistency of the oil, and increases stability to oxidation, which extends the shelf life of the finished products. Also some of the oils are winterized to remove the higher melting constituents, which can be used in confectionary products; the winterized oil is less likely to become cloudy in refrigerated storage. 14.10.3.1 Preparation for extraction Storage: For optimum extraction and quality of oil, the oil-bearing material should be stored so that it remains dry and at relatively low temperature. If it is wet, it should be processed as soon as possible after harvest. Oils in the presence of water can deteriorate rapidly, forming free fatty acids and causing greater refining loss. Seed cleaning: The first step in the commercial processing of oilseeds is “cleaning”, to remove foreign materials, such as sticks, stems, leaves, other seeds, sand, and dirt using dry screeners and a combination of screens and aspiration. Permanent electromagnets are also used for the removal of trash iron objects. Final cleaning of the seed usually is done at the extraction plant just prior to processing.
14.10 Food industry
937
Figure 14.10.2. Flow diagram of edible oil processing.
Dehulling: After cleaning, it may be necessary to remove the seed’s outer seed- coat (hull). The seedcoat contains little or no oil, so its inclusion makes the extraction less efficient. Also, the next processing step is grinding to reduce particle size, and any tough seedcoats would interfere with this process. If the hulls are not removed prior to extraction, they will reduce the total yield of oil by absorbing and retaining oil in the press cake. An acceptable level of hull removal must be determined, depending on the desired protein level of the final meal. Hulls are removed by aspirator and undehulled seeds are removed from the kernels by screening and returned to the huller. Some meats still adhere to the hulls, which are beaten, then screened again to obtain the meat. Grinding, rolling, or flaking: After dehulling, the meats are reduced in size, or “flaked,” to facilitate oil removal. Proper moisture content of the seeds is essential for flaking, and if the moisture level is too low, the seeds are “conditioned,” with water or steam, to raise the moisture to about 11%. For solvent extraction, flakes are commonly not less than 0.203-0.254 mm (0.008-0.010 inch), which can be solvent extracted efficiently with less than 1% residual oil. Thinner flakes tend to disintegrate during the solvent extraction process and reduce the miscella percolation rate. Cooking: Prior to extraction, the flakes are heated. The purpose of cooking the flakes is: (1) cell walls are broken down, allowing the oil to escape; (2) oil viscosity is reduced; (3) moisture content is controlled; (4) protein is coagulated; (5) enzymes are inactivated and microorganisms are killed; and (6) certain phosphatides are fixed in the cake, which helps to minimize subsequent refining losses. Flakes are cooked in stack cookers to over 87.8oC (190oF) in the upper kettle. Flakes with high phosphatide content may benefit from being cooked at slightly lower temperatures to avoid elevating refining losses. The temperature of the flakes is raised to 110-132.2oC (230-270oF) in the lower kettles. The seeds are cooked
938
Phillip J. Wakelyn, Peter J. Wan
for up to 120 min. Overcooking lowers the nutritional quality of the meal and can darken both the oil and meal. Poor-quality seeds with high levels of free fatty acids cannot be cooked for as long a period as high-quality seeds because of darkening. Darker oil requires additional refining to achieve a certain bleach color. Expanders: Sometimes low shear extruders called expanders are used. This equipment has the capability to process both low- and high-oil content materials. The meats are fed into an extruder after dehulling, flaking, and cooking and are heated as they are conveyed by a screw press through the extruder barrel. The meats are under considerable pressure and temperature when they reach the exit of the extruder. The change in pressure as the material leaves the extruder causes it to expand and the oil cells are ruptured, releasing the oil, which is rapidly reabsorbed. The expanded “collets” produced are then cooled and extracted with solvent. 14.10.3.2 Oil extraction Prepress solvent extraction: In this process the oil-bearing material are first mildly pressed mechanically by means of a continuous screw press operation to reduce the oil by half to two-thirds of its original level before solvent extraction to remove the remaining oil in the pre-pressed cake. Pressing follow by solvent extraction is more commonly used when high oil content materials (e.g., canola/rapeseed, flaxseed, corn germ) are processed. Direct solvent extraction: This process involves the use of a nonpolar solvent, usually hexane, to dissolve the oil without removing proteins and other non-oil soluble compounds. Solvent extraction yields about 11.5% more oil than does the screw press method, and less oil remains in the meal. The cooked flakes or collets (if expanders are used) are mixed with hexane in a batch or continuous operation. The hexane vapor pressure limits the practical operating temperature of the extraction and its contents to about 50-55oC. The resulting miscella (oil-solvent mixture) and the marc (solvent laiden collets) are heated to evaporate the solvent, which is collected and reused. The oil is freed from the miscella, by using a series of stills, stripping columns, and associated condensers. The hexane-free oil (i.e., crude oil) is cooled and filtered before leaving the solvent-extraction plant for storage or further treatment. This is the crude oil normally traded in the commodity market. Occasional overheating of the oil-solvent miscella will cause irreversible color changes in the oil. 14.10.3.3 Processing crude oil Most crude edible oils, obtained from oil-bearing materials, consist primarily of triglycerides (triacylglycerols). The triglycerides (approximately 95% of the crude oil) are the constituents recovered for use as neutral oil in the manufacture of finished products. The remaining nontriglyceride portion contains variable amounts of other lipophilic compounds, such as free fatty acids (FFA), nonfatty materials generally classified as “gums,” phospholipids (phosphatides), tocopherols, color pigments, trace metals, sterols, meal, oxidized materials, waxes, moisture, and dirt. Most of these minor lipid components are detrimental to finished product color, flavor, and smoking stability, and so must be removed from the neutral oil by a processing/purification process. The object of the processing/purification steps is to remove the objectionable impurities while minimizing possible damage to the neutral oil and tocopherols and loss of oil during such processing. Lecithin and cephalin are common phosphatides found in edible oils. Soybean, canola/rapeseed, corn, and cottonseed are the major oils that contain significant quantities of phosphatides. Alkaline treatment used for FFA reduction is also capable of removing most
14.10 Food industry
939
of the phosphatides from these crude oils. Tocopherols are important minor constituents of vegetable oils, which are natural antioxidant that retard the development of rancidity. Refining, bleaching, and deodorization are the steps that are necessary if the oil is to be used in food applications. Oil that has only gone through these three steps is called “RBD” oil. Figure 14.10.2 illustrates the processing pathways. Refining: Refining involves the removal of nonglyceride materials (phospholipids, color, and trace materials) and FFA. The goal is to produce a high-quality refined oil with the highest yield of purified triglycerides. Refining is by far the most important step in processing. An improperly refined oil will present problems in bleaching and deodorization and reduce quality. Some solvent-extracted crude oils, including soybean or canola/rapeseed, contain approximately 2-3% gums, which are mainly phosphatides and require degumming. The principal phosphatides are lecithin and cephalin. Gums can cause problems through higher then necessary refining losses, or by settling out in storage tanks. The degumming operation exploits the affinity of most phosphatides for water, converting them to hydrated gums that are insoluble in oil and readily separated by centrifugal action. Lecithin can be recovered and concentrated from the gums in a separate solvent extraction process, usually with acetone. Either water-degummed oil or crude oil can be treated with sodium hydroxide solution to saponify free fatty acids that are subsequently removed as soapstock by a primary refining centrifuge. Conventional alkali refining is by far the most widespread method of edible oil refining. The success of the alkali refining operation is the coordination of five prime factors: (1) use of the proper amount of reagent (sodium hydroxide), (2) proper mixing, (3) proper temperature control, (4) proper residual contact time, and (5) efficient separation. Oil is alkali-refined by the addition of sodium hydroxide solution at a level sufficient to neutralize the FFA content of the oil. An excess of sodium hydroxide is required to reduce the color of the refined oil and to ensure the completion of the saponification reaction and to remove other trace elements. The amount and strength of the sodium hydroxide solution needed to neutralize the FFA is dependent on the amount of both FFAs and phosphatides present in the crude oil. Water-soluble soaps are formed in the primary reaction between the sodium hydroxide and FFAs. The hydratable phosphatides react with the caustic forming oil-insoluble hydrates. The caustic used in alkali refining is normally diluted to about 8-14% NaOH, although higher concentrations are occasionally used to reduce color. The proper amount of NaOH solution added to the oil will produce an adequately refined oil with the minimum of triglyceride oil loss. The amount of NaOH solution (neutralizing dose plus excess) is determined by experience and adjusted according to laboratory results. After the NaOH solution is injected, it is mixed for 6-10 minutes to ensure thorough contact. The treated oil is then heated to assist in breaking of the emulsion prior to separation of the soapstock from oil in continuous centrifuges. Any soap remaining, after the primary soapstock separation, is removed through continuous hot water washings. In this step, water is added at 10-15% at a temperature sufficient to prevent emulsification, generally 82-90.5ºC (180-195ºF). The oil is again separated from the soapy phase in water wash separators and drier prior to bleaching. Bleaching: The oil is further purified by “bleaching”, which removes color bodies and trace metals as well as entrained soaps, and products of oxidation that are adsorbed onto the surface of bleaching agents or adsorbents. Types of adsorbents most commonly used in-
940
Phillip J. Wakelyn, Peter J. Wan
clude neutral clay, acid activated clay, and activated carbon. The choice of adsorbent will depend on a balance between activity of the adsorbent, oil retention loss, and adsorbent cost. The process is generally carried out via batch or continuous bleaching. Adsorbent is mixed with the refined oil creating a slurry that is agitated to enhance contact between the oil and the adsorbent. This is generally carried out under a vacuum at 90-95ºC (194-203ºF) for 15-30 minutes. Vacuum bleaching offers the advantages of an oil with improved oxidative and flavor stability. Finally the adsorbent is filtered from the oil using pressure leaf filters precoated with diatomaceous earth. Spent clay is steamed for efficient oil recovery. Deodorization: Deodorization, which removes the volatile compounds along with residual FFA, is a critical step in ensuring the purity of any vegetable oil and improves flavor, odor, color, and oxidative stability. Many of the volatile compounds removed are formed by the auto-oxidation of fat, which produces aldehydes, ketones, alcohols, and hydrocarbons that are associated with undesirable flavors and odors. The process also is effective in removing any remaining pesticide residues or metabolites that may be in the oil. Deodorization, which can be conducted as a batch operation in smaller plants or as a continuous or semicontinuous process by larger deodorizing facilities, consists of a steam distillation process in which the oil is heated to 230ºC (446ºF) under a vacuum of 2-10 mm Hg. Steam is sparged through the oil to carry away the volatiles and provide agitation. The odor and flavor compounds, which are more volatile than the triglycerides, are preferentially removed. After deodorization and during the cooling stage, 0.005-0.01% citric acid is generally added to chelate trace metals, which can promote oxidation. Deodorized oils preferably are stored in an inert atmosphere of nitrogen to prevent oxidation. Tocopherols and sterols are also partially removed in the deodorization process. Tocopherols can be recovered from the deodorizer distillate in a separate operation. 14.10.4 REVIEW OF SOLVENTS STUDIED FOR EXTRACTION EFFICIENCY Research on solvents for extraction has been carried out for more than 150 years and has intensified since the first patent was issued to Deiss of France in 1855.1,3,48 In the early effort of selecting an extraction solvent, the availability, operation safety, extraction efficiency, product quality and cost were the major concerns. In recent decades, toxicity, bio-renewability, environmental friendliness have been added to the solvent selection criteria. Among the solvents tested, a majority of the candidate solvents were excluded on the ground of toxicity and safety. Only a handful of solvents are used to any degree. These are acetone, alcohol, hexanes, heptane, and water.1,4-7,49 Water is used in rendering of fat from animal tissues and fish and in coconut processing,49 alcohol for spice and flavorants extraction,5,49 acetone for lecithin separation and purification.4 For commodity oils derived from vegetable sources, only hydrocarbon solvents have been used since 1930’s. Acetone was used by an Italian cottonseed oil mill during the 1970’s.4 Aqueous acetone and acetone-hexane-water azeotrope were studied by the scientists at the Southern Regional Research Center of Agricultural Research Service, USDA during the 1960’s and 1970’s.4 The effort was stopped due to the cost of retrofit required, the difficulties in managing the mixture of solvents with the presence of water and product quality concerns - a strong undesirable odor associated with the acetone extracted meals.4 Ethanol and isopropanol were studied in the 1980’s as a potential replacement of hexane for oil extraction. Both were proven technically feasible but economically unacceptable.5,6 n-Hexane is listed as a HAP under the CAA14,15 (See Section 14.10.2.2.1 CAA) and there are other regulatory requirements. As a way to meet envi-
14.10 Food industry
941
ronmental regulations, a short term option to commercial hexane appears to be hydrocarbons with significantly reduced n-hexane content. 14.10.4.1 Hydrocarbon solvents Extraction of oils has largely relied on mechanical or heat rendering process for centuries.50 Increased demand of productivity to separate oils from oilseeds has been the principal factor driving the changes of oilseed processing from the ancient hydraulic press to a continuous screw press or expeller in early 1900’s.51,52 This operation still left more than 4-5% residual oil in the pressed cake.51 More complete recovery of oil can only be effectively accomplished by solvent extraction.53,54 Solvent extraction of oils had an early beginning. Deiss in France received a patent to extract fat from bone and wool with carbon bisulfide in 1855.53 A year later, Deiss received additional patents covering the extraction of oil bearing seeds. Large scale solvent extraction already was established in Europe in 1870.55 The earliest extractors were unagitated single-unit batch extractors of small capacity and not very efficient.56,57 These extractors were gradually modified by the addition of agitation. They were organized in a battery of ten batch extractors which can be operated in a countercurrent principle. Extractors of this type operated in European plants during the last three decades of the 19th century.57 Further development in solvent extraction technology was relatively slow until early twentieth century. Solvent extraction spread from Europe to various parts of the world including the United States and South America.57 The first extraction plant in the United States was used to recover grease from garbage, bones, cracklings, and other packing-house wastes and to recover residual oil from castor pomace.57 Wesson58 reported his efforts applying solvent extraction to recover cottonseed oil from 1889 till the close of World War I. During the 1930’s solvent extraction was introduced in the United States for the recovery of oil from soybeans and the German equipment of the continuous type was used almost exclusively.57-61 Just prior to World War II the installation of continuous solvent extraction equipment was greatly accelerated and throughout the period of the War new plants were erected in an effort to keep pace with the constantly increasing production of soybeans. All of the later installations have been of American manufacture and in a number of cases of American design.57-61 Solvents used in the early effort to extract grease and oils were diverse. Besides carbon bisulfide used by Deiss,53 chlorinated hydrocarbons, benzene, and alcohols were all being tried. Extracting oil from corn and cottonseed with both aviation gasoline and petroleum distillate was performed in the United States in 1915 and 1917 respectively.62 The hydrocarbon paraffins became the preferred solvents for oilseed extraction during 1930’s through the process of elimination.63-71 Due to the prominent defects of early solvent extraction: dark crude oil, strong solvent odor in meal and high cost associated with solvent loss, low boiling hydrocarbons such as propane and butane were recommended as oil extraction media.72 The flammability of hydrocarbons also prompted much research in 1940’s using chlorinated hydrocarbons as the extraction solvents73,75 before its meal was found unsafe as feed.75-77 For the purpose of improved protein and oil quality75,78 and of processing safety and biorenewable solvents,75,79,80 both ethanol and isopropanol were investigated as the oil extracting solvents. While these alcohols offer various advantages in product quality and process safety and are renewable, they are still not economically feasible to replace hydrocarbons as oilseed extraction solvents.81 Hexane rich solvent became popular for the oilseed industry,54,57,82,83 because it is the most efficient solvent, extracts minimum non-oil
942
Phillip J. Wakelyn, Peter J. Wan
material and is easy to separate from the crude oil and marc.63 A thorough comparison of various hydrocarbon solvents for cottonseed oil extraction on a lab scale basis was reported by Ayers and Dooley in 1948.84 A more recent study by Wan et al.85 used a laboratory scale dynamic percolation extractor which operates at the conditions similar to those applied in the oil mill practice. Plant trials of isohexane and heptane solvents versus hexane in a 300 tons/day cottonseed factory revealed some interesting findings.7 14.10.4.1.1 Nomenclature, structure, composition and properties of hydrocarbons Petroleum and natural gas are the most abundant and affordable sources of hydrocarbon. Sometimes naphtha is used to describe the low boiling liquid petroleum and liquid products of natural gas with a boiling range from 15.6oC (60oF) to 221oC (430oF). This large group of compounds can be structurally classified as aliphatic and aromatic. Aliphatic hydrocarbons include saturated alkanes (paraffins), unsaturated alkenes (olefins) and alkynes (acetylenes), and cycloparaffins (naphthenes). Paraffins can be linear such as n-butane, n-pentane, and n-hexane, and branched such as isobutane, isopentane, isohexane, etc. Example of olefins is ethylene; of cycloparaffins, cyclopentane and cyclohexane; and of an aromatic, benzene.8,86,87 These compounds are derived from natural gas and petroleum oils which normally contain thousands of hydrocarbons with molecular weight ranging from methane to about 50,000 - 100,000 Daltons. Upon refining, the crude petroleum is divided into hydrocarbon gases (methane, ethane, propane and butane), light distillates (naphthas and refined oils), intermediate distillates (gas oil and absorber oil), heavy distillates (technical oils, paraffin wax and lubricating oils), residues (petroleum grease, residual fuel oil still wax, asphalts and coke), and refinery sludges (acid coke, sulphonic acid, heavy fuel oils and sulfuric acid).88 Historically, various fractions of petroleum naphthas, pentane and hexane from the light naphthas, aviation gasoline and benzol from the intermediate naphthas, and aromatic hydrocarbon, benzene, have been tested for oil extraction.62-71 14.10.4.1.2 Performance of selected hydrocarbon solvents Factors affecting extraction: There is little theoretical basis to be followed for the extraction of oilseeds.66,89-91 The study of the extraction of oilseeds is complicated by the fact that the total extractible material is variable in quantity and composition.66,89 Composition of the early extracted material is nearly pure triglycerides. As the extraction progresses, increasing amount of non-glyceride material will be extracted.66,89 IT is believed that the majority of the oil from oilseed flakes is easily and readily extracted.66,90 While the thickness of flakes affects extraction rate, the concentration of miscella below 20% does not greatly increase the amount of time to reduce the residual oil in flakes to 1%.89 Good91 summarized much of the early effort in soybean extraction: (1) The first oil extracted is superior in quality to the last small fraction; (2) While other solvents have been used in the past, hexane has become the primary solvent due to a combination of properties; (3) Flake thickness is the most important factor in achieving good extraction results; (4) Higher extractor temperatures up to nearly the boiling point, improve extraction results; (5) Moisture control is important throughout the extraction process; (6) Heat treatment affects the total extractibles; and (7) The soaking theory of extraction indicates that weak miscellas are very effective in helping to achieve good extraction results. Particle size which relates to the surface area available for extraction and is obviously one of the most important factors for extraction study. Coats and Wingard92 noticed that par-
14.10 Food industry
943
ticle size was more influential when the seed grit was being extracted. When oilseed flakes were being extracted, the flake thickness would be a more important factor instead of size of the flakes. Moisture content in oilseed can affect the extraction results.93-95 Optimum moisture content of cottonseed meats for extraction was first reported by Reuther et at.94 to be from 9 to 10%. Work by Arnold and Patel95 indicated 7 to 10% to be the optimum moisture for cottonseed flakes and very little variation in extraction rate for soybean with moisture content between 8 and 12%. Wingard and Phillips96 developed a mathematical model to describe the effect of temperature on extraction rate using a percolation extractor as follows:
(
)
log(time,min) = n log Temp., oF + log k
or
time = k (Temp.)
n
[14.10.1]
where time is defined as the number of minutes required to reach 1% residual oil in the oilseed flakes. For all practical purpose, they concluded that the time in minutes required to reduce the oilseed to 1% residual oil content on a dry basis varied inversely with the square of the extraction temperature in degrees Fahrenheit. Evaluation methods: Except for the pilot plant batch or counter-current extraction described by various labs,54,74,78 most of the solvent extraction evaluation work found in the literature was done in one or several of the lab scale devices. The percolation batch-extraction apparatus of the Soxhlet type has often been used to evaluate the rate of extraction of hydrocarbon solvents such as the one described by Bull and Hopper.89 Wingard and Shand97 described a percolation type of extractor and a co-current batch extractor and claimed to be useful to study the factors influencing equipment design and plant operation as well as fundamental studies contributing to a general understanding of extraction. Wan, et al. modified the design of percolation type extractor to closely simulate a single stage counter current miscella extraction conditions as practiced in the factory.85 Co-current batch extractor with numerous variations was also frequently applied for the extraction properties of selected solvents which were often operated at room temperature.75,97 Soxhlet extraction84,85 and Soxtec System HT6 (Perstorp Analytical, Herndon, VA) were also frequently used to evaluate solvents.98 Soxhlet extractor allows vaporized and condensed pure solvent to percolate through oilseed sample. The temperature of the condensed solvent is normally lower than its boiling point. Depending upon the cooling efficiency of the condenser and the room temperature, the temperature of the condensed solvent and the temperature of the extracting solvent in the extractor largely varied from lab to lab. This extraction temperature variability was minimized with the Soxtec method by refluxing the oilseed sample in the boiling solvent for 15 minutes followed by Soxhlet type of rinsing for 35 minutes. In theory the Soxtec method is more efficient and better reproduced. However, the Soxtec method only utilized a 3 g oilseed sample. The heterogeneity of an oilseed sample could be a significant source of variation. Flakes of oilseeds were most frequently used for the solvent extraction studies. Sometimes, ground oilseed kernels through a specified sieve size was used.98 Residual oil content in the extracted flakes after a certain specified extraction condition or oil content in miscella (mixture of oil and solvent) was examined and the percentage of total oil extracted was calculated.89-97 The total extractable oil of flakes was determined by four hours Soxhlet extraction. Wan et al.85 used a precision densitometer to determine the miscella concentration (percent of oil in miscella by weight) after a given time of extraction from which the per-
944
Phillip J. Wakelyn, Peter J. Wan
centage of oil extracted from cottonseed flakes was calculated. From these data, Wan et al.85 was also able to estimate the initial rate of extraction and final extraction capacity for each solvent as fresh and at selected initial miscella concentrations up to 30%. Bull and Hopper89 conducted extraction of soybean flakes in a stainless steel batch-extraction apparatus of the Soxhlet type with petroleum solvents, Skellysolve F (boiling range, 35 to 58oC) at 28oC and Skellysolve B (boiling range, 63 to 70oC) at 40oC. The extraction was carried out to permit the miscella obtained by each flooding of the flakes with solvent to be recovered separately. Their results showed that iodine number decreased and refractive index increased slightly with the extraction time which implied that more saturated fat was extracted during later stage of the extraction. Oils extracted during the later stages of the extraction were found to contain greater amounts of unsaponifiable matter and were rich in phosphatides, as high as 18% of the last fraction. Skellysolve B which is a hexane rich solvent demonstrated a much faster initial rate of extraction than that of Skellysolve F which is a pentane rich solvent and therefore, it took longer to complete the extraction for Skellysolve F. The fatty acid profile of each fraction showed a slight increase of saturated and a slight decrease of unsaturated fatty acid in the later fractions. Arnold and Choudhury82 reported results derived from a lab scale extraction of soybean and cottonseed flakes in a tubular percolation extractor at 135-140oF with pure, high purity and commercial hexane, and reagent grade benzene. They claimed that pure hexane extracted soybean slower than high purity and commercial hexane. During the first 60 minutes of extraction, benzene extracted more oil than the hexanes. However, at the end of 80 minutes, benzene extracted only slightly more than pure hexane but definitely less than the commercial hexanes. Similar results were obtained for the four solvents when cottonseed flakes were extracted. A laboratory extraction study of cottonseed flakes using various hydrocarbon solvents was reported by Ayers and Dooley.84 Soxhlet extractor and Waring blender were used for these experiments. Among the petroleum hydrocarbon solvents tested were branched, normal and cyclo-paraffins as well as aromatic hydrocarbons with various degrees of purity. They were: pure grade (99 mole percent purity) n-pentane, isopentane, cyclohexane, benzene, and n-heptane; technical grade (95 mole percent purity) neohexane, diisopropyl, 2-methylpentane, 3-methylpentane, n-hexane, and methylcyclopentane; technical grade (90 mole percent purity) cyclopentane; and commercial grade n-heptane, isohexanes, n-hexane, isoheptane and n-heptane. To assess the performance of these solvents, they used the following empirical formula: Quality-Efficiency Rating = 0.4 (Oil Yield Factor) + 0.4 (Refining Loss Factor) + 0.2 (Refined and Bleached Oil Color Factor)
[14.10.2]
When comparing the oil yield factor alone, 3-methylpentane was rated the best. When comparing the solvents based on the empirical Quality-Efficiency Rating formula, they concluded that methylpentanes (3- and 2-methylpentane) were superior extraction solvents for cottonseed oil. The normal paraffins, highly-branched isohexanes, cycloparaffins, and aromatics were progressively rated as less efficient than methylpentanes. Therefore, they recommended a tailor-made solvent for the extraction of cottonseed should exclude aromatic hydrocarbons, have low limits on cycloparaffin content, and consist largely of normal and isoparaffin hydrocarbons.
14.10 Food industry
945
A more recent study by Wan et al.85 using a laboratory scale dynamic percolation type of extractor (Figure 14.10.3) operated at the following conditions such as, temperature (5oC below the boiling point of each solvent) and miscella flow rate (9 gal/min/ft2), similar to those applied in the oil mill practice. Commercial grade hexane, heptane, isohexane, neohexane, cyclohexane, and cyclopentane were used to extract cottonseed flakes which had 5.8% moisture and 31.4% oil. When these solvents were tested near their boiling points, hexane apparently extracted cottonseed oil at a higher initial rate, > 94% oil extracted after 2 minutes, than all other solvents. Both heptane and hexane were able to extract more oil at the end of 10 minutes of extraction. Isohexane demonstrated to have adequate initial extraction Figure 14.10.3. Schematic of bench-scale dynamic percolation exrate (80% oil extracted after 2 tractor. minutes) and extraction capacity (93% oil extracted after 10 minutes of extraction) but is noticeably less effective than hexane. Similar to findings by Ayers and Dooley,84 results from the study by Wan et al.85 also demonstrates that neohexane, cyclohexane and cyclopentane performed distinctly less efficiently than hexane, heptane and isohexane. Conkerton et.al.98 tested commercial heptane versus hexane in a Soxtec extractor. Under this extraction condition, heptane actually extracted more oil than hexane from ground cottonseed kernel passed through a 20 mesh screen. The oil and meal quality were not appreciably affected by the higher temperature extraction of heptane. Plant scale results: Although hydrocarbon solvents have been used for oilseed extraction since the 1930’s, very little in plant operating data are available. During the spring of 1994, Wan et al.7 conducted plant trials with commercial heptane and isohexane at a 300 tons/day cottonseed crushing plant which routinely used hexane as the extraction solvent. Test results indicated that heptane performed well as an extraction solvent. However, it required extra energy and time to recover and consequently reduced the throughput rate of cottonseed being processed. Isohexane on the other hand was termed as an “easier” solvent by the plant engineers than hexane to operate. The plant also experienced a 40% steam savings and better than 20% throughput increase when it was operating with isohexane.7 This encouraging result prompted a second plant trial with commercial isohexane.45 The second plant trial was carried out at a cottonseed oil mill with a relatively new extraction and
946
Phillip J. Wakelyn, Peter J. Wan
miscella refining facility which was constructed in 1988 with a designed capacity of 500 tons/day but operated at only 270 tons/day due to limited delinting capacity. After week long testing with commercial isohexane, this plant experienced more than 20% natural gas usage and easily increased the throughput rate by close to 10% when compared with commercial hexane. This energy savings with commercial isohexane over commercial hexane may be largely attributed to the difference in the amount of water present in their corresponding azeotropes. Isohexane requires an additional step - isomerization to manufacture and will always be priced higher than hexane. But based on the two cottonseed oil mill trials, isohexane can be a cost efficient solvent.99 One additional benefit, the shorter residence time of the extracted cottonseed marc in the desolventizer/toaster because of the lower boiling range of isohexane will likely preserve more vegetable protein in the final meals which has been observed by both plants during the tests.7,45 The benefit in improved quality of oils were not obvious in both plant trials but might be realized with extended trials. Further evaluation of hydrocarbon solvents: As indicated in the study conducted by Wan et al.,85 the commercial cyclic hydrocarbons are the least effective extraction solvent than the branched and linear hydrocarbons. The comparison of extraction efficiency of pure isomers of hexane was conducted with the same single stage extractor as displayed in Figure 14.10.3. This was done to identify any unique structure-function characteristics of these pure components of commercial hexane and provide some guidance to the selection and tailored formulation for future commercial isohexane for the oilseed extraction industry. The extraction results for various pure isomers of six carbon paraffins using the single stage extractor indicated the following: (a) cyclohexane is noticeably less efficient in extracting cottonseed flake than all the other isomers; (b) slightly branched isomers, such as, 2-methyl and 3-methyl pentane, and methyl-cyclopentane are very slightly less efficient than n-hexane; and (c) highly branched isomers, 2,2-dimethyl and 2,3-dimethyl butane, are slightly less efficient than slightly branched isomers in extraction (Unpublished data). 14.10.5 FUTURE TRENDS In the future there most likely will be new demands for highly specialized extraction solvents as newly domesticated species that make useful novel oils30 and other products and new or altered biological products with enhanced nutritional and industrial properties will be developed through conventional breeding and genetic engineering for use as “functional foods”31 (e.g., phytosterols to achieve cholesterol lowering); as oils with altered lipid profiles32 (e.g., for lower saturated fat) or with more vitamin E; new drugs/nutraceuticals, industrial chemicals (e.g., fatty acids for lubricants, as cosmetics, coatings, detergents, surfactants, flavors, polymers, etc.); as sources for specialty chemicals; as value added products; etc.31-38 Genetically enhanced (GE) /biotech crops make up a growing share of the agricultural output.39 Biotechnology is the most powerful tool ever put in the hands of agricultural scientists. The ability to breed desirable traits or eliminate problematic ones can yield potentially spectacular benefits, such as various chemicals of importance including improved fats and oils, and vaccines and medicine, improved nutrition (e.g., in casaba, oilseeds, rice, sweet potatoes), and improved yields with the use of less agricultural chemicals. GE/biotech crops could be increasingly developed as biofactories for a wide range of products, including nutrients pharmaceuticals, and plastics. There is much promise for being able to produce products that would protect millions from disease, starvation, and death. However biotechnology and GM crops have become very controversial, and have run into serious
14.10 Food industry
947
problems in Europe, particularly in the UK.40 Europe,41 Japan, Korea, and Australia/New Zealand already have some restrictions and require some labeling. The U.S. is reviewing the issue.42 Thus even though this technology has great promise for increased use of new and existing solvents for extraction of products from diverse biological materials, there are also many potential problems because of misperceptions and misinformation. In the U.S. EPA is developing Maximum Available Control Technology (MACT) standards for vegetable oil processing that are likely to be finalized in 2001 (see Section 14.10.2.2.1, Hazardous Air Pollutants), with enforcement three years after promulgation.16,17 Commercial hexane, which is a HAP and a VOC, is presently the solvent used.16,17 To meet new and existing CAA requirements it is likely that extraction facilities will become much more efficient chemical engineering operations with upgraded equipment,43 more computerized monitoring and control for better quality management,44 and better environmental management/stewardship.19 In addition it is possible that alternate solvents (e.g., isohexane3,7,45) or lower n-hexane content commercial hexane (30-50% vs. 64%) will be used to meet these regulations. It is also possible that solvents like acetone, which is not a HAP or VOC and is not on the TRI list, will be more strongly investigated.16,17,46 In Europe trans fatty acid labeling of retail foods is required and in the U.S. FDA has proposed to label trans fats as saturated fat on the nutrition labeling panel required on packaged food sold at retail (64 FR 62,764; Nov. 17, 1999). This regulation if, promulgated as proposed, will result in reformulation of many products that could affect the vegetable oil producing and extracting industries. It is clear that the future has much uncertainty, while at the same time it offers much promise. It appears that there will be many potential changes that will put new demands on extraction solvents. Solvents that are more environmentally friendly, are nontoxic to plants, workers, and consumers, have specialized properties, have high solvent power at low temperatures (are easy to desolventize), etc., may have wider potential use in conventional extraction as well as specialized niche markets. REFERENCES 1 2 3 4 5 6 7 8
9
10
11 12
Technology and Solvents for Extracting Oilseeds and Non-petroleum Oils, P.J. Wan and P.J. Wakelyn, Eds., AOCS Press, Champagne, IL, 1997. M.A. Williams and R.J. Hron, Bailey’s Industrial Oils and Fat Products, 5th edn., Vol. 4: Edible Oil and Fat Products: Processing Technology, Y.H Hui,., Ed., John Wiley and Sons, Inc., 1996, p. 119. P.J. Wan, Hydrocarbon Solvents, in Technology and Solvents for Extracting Oilseeds and Non-petroleum Oils, P.J. Wan and P.J. Wakelyn, Eds., AOCS Press, Champaign, IL, 1997, p.170-185. R.J. Hron, Acetone, in Ibid, p.186-191. R.J. Hron, Ethanol, in Ibid, p.192-197. E.W. Lucas and E. Hernandez, Isopropyl Alcohol, in Ibid, p. 199-266. P.J. Wan, R.J. Hron, M.K. Dowd, M.S. Kuk, and E.J. Conkerton, J. Am. Oil Chem. Soc., 72, 661 (1995). Occupational Health and Safety Administration Field Operations Manual, Chapter IV: Violations, C. Health Standards Violations, (OSHA Instruction 2.45B CH-4, Dec.13, 1993), The Bureau of National Affairs, Washington, DC, 1994, pp. 77:2513-18. Occupational Health and Safety Administration Technical Manual, Section I- Sampling, Measurement Methods, and Instruments, Chapter 1 - Personal Sampling for Air Contaminants, Appendix I:1-6. Sampling and Analytical Errors (SAEs) (Issued by OSHA Instruction TED 1.15, September 22, 1995; amended by OSHA Instruction TED 1.15 CH-1, May 24, 1996). 1997 TLVs and BEIs, Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices, The American Conference of Governmental Industrial Hygienists, Cincinnati, OH, 1997, pp. 12-40. J.B Galvin, C.J. Kirwin, D.W. Kelly, INFORM, 6(8), 951 (1995). H.H. Schaumberg and P.S. Spencer, Brain, 99, 183 (1976).
948
13
14 15 16 17 18
19
20 21
22 23 24 25 26 27 28 29
30 31 32 33 34
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
Phillip J. Wakelyn, Peter J. Wan
J.B. Galvin, Toxicity Data for Extraction Solvents Other Than Isohexane/Hexane Isomers, in Technology and Solvents for Extracting Oilseeds and Nonpetroleum oils, P.J.Wan and P.J.Wakelyn, Eds., AOCS Press, Champaign, IL, 1997, p. 75-85. P.J. Wakelyn, Cotton Gin and Oil Mills Press, 92(17), 12 (1991). P.J. Wakelyn and L.A. Forster, Jr., Oil Mill Gaz., 99(12), 21 (1994). P.J. Wan and P.J. Wakelyn, INFORM, 9(12),1155 (1998). P.J. Wan and P.J. Wakelyn, Oil Mill Gaz., 104(6), 15 (1998). U.S. EPA, Study of Volatile Organic Compound Emissions from Consumer and Commercial Products, U.S. Environmental Protection Agency, Office of Quality Planning and Standards, Research Triangle Park, NC 27711, EPA - 453/R-94-066-A, March 1995. P.J. Wakelyn and P.K. Adair, Assessment of Risk and Environmental Management, in Emerging Technologies, Current Practices, Quality Control, Technology Transfer, and Environmental Issues, Vol. 1, Proc. of the World Conference on Oilseed and Edible Oil Processing, S.S. Koseoglu, K.C. Rhie, and R.F. Wilson, Eds. AOCS Press, Champaign, IL, 1998, pp. 305-312 n-Hexane, U.S. EPA Integrated Risk Information System (IRIS) Substance File, U.S. EPA, 1999 (www.epa.gov/ngispgm3/IRIS/subst/0486.htm). U.S. EPA, Toxic Chemical Release Inventory Reporting Form R and Instructions (Revised 1995 Version), U.S. EPA, Office of Pollution Prevention and Toxics, Washington, DC, EPA 745-K-96-001, March 1996, Table II, p. II-1 J.K. Dunnick, Toxicity Studies of n-Hexane in F344/N Rats and B6C3F1 Mice, National Toxicology Programs, U.S. Dept. of Health and Human Services, NTP TOX 2, NIH Publication No. 91-3121, 1991. A. H. Allen, GRAS Self-Determination: Staying Out of the Regulatory Soup, Food Product Design, (April 1996 supplement to Food Product Design, 6 pages) (1996). B.L. Oser and R.A. Ford, Food Technol., 27(1), 64 (1973). R.L. Hall and B.L. Oser, Food Technol., 19, 151 (1965). A.M. Galvin, in Introduction to Fats and Oils Technology, P.J. Wan, Ed., AOCS Press, Champagne, IL, 1991,pp. 137-164. L.A. Jones and C.C. King, in Bailey’s Industrial Oil and Fats Products, 5th edn., Vol. 2 Edible Oil and Fat Products: Oils and Oil Seeds, Y.H. Hui, Ed., John Wiley and Sons, Inc., 1996, pp. 177-181. H.W. Lawson, Standards for Fat and Oils, The AVI Publishing Co., Inc., Westport, CT, 1985, p. 34. S.W. Kuhlmann, M.C. Calhoun, J.E. Huston and B.C. Baldwin, Jr., Total (+)- and (-)- Gossypol in Plasma and Liver of Lambs Fed Cottonseed Meal Processed by Three Methods, J. Anim. Sci., 72 (suppl. 1), 145 (1994). D.J. Murphy, INFORM, 11(1), 112 (2000) M.A. Ryan, Today’s Chemist at Work, 8(9), 59 (1999). B.F. Haumann, INFORM, 8(10), 1001 (1997). B. Flickinger and E. Hines, Food Quality, 6(7), 18 (1999). C.T. Hou, Value Added Products from Oils and Fats through Bio-processes, Int. Symp. On New Approaches to Functional Cereals and Oils, Beijing, China, Nov. 9-14, 1997, Chinese Cereals and Oils Association, Beijing, China, 1997, p. 669. D.J. Kyle, New Specialty Oils: Development of a DHA-rich Nutraceutical Product, Ibid, p. 681. J.K. Daum, Modified Fatty Acid Profiles in Canadian Oilseeds, Ibid, pp. 659-668. C.M. Henry, Chemical Eng. News, 77(48), 42 (1999). R. Ohlson, INFORM, 10(7), 722 (1999). G.J. Persley and J.N. Siedow, Applications of Biotechnology to Crops: Benefits and Risks, CAST Issue Paper No. 12, Council for Agricultural Science and Technology, December 1999. M. Heylin, Chem. Eng. News, 77(49), 73 (1999). Anom., Official J. European Communities, 43(L6), 13 (2000). B. Hileman, Chem. Eng. News, 77(50), 31 (1999). P. Delamater, Oil Mill Gaz., 104(11), 34 (1999). P.J. Wakelyn, P.K. Adair, and S.R. Gregory, Oil Mill Gaz., 103(6), 23 (1997). M. Horsman, Oil Mill Gaz., 105(8), 20 (2000). R.J. Hron, P.J. Wan, and P.J. Wakelyn, INFORM, In Press, (2000). J.A. Dean, Dean’s Handbook of Chemistry, Thirteenth Edition, McGraw-Hill, Inc., New York, NY, 1985. R.D. Hagenmaier, Aqueous Processing, in Technology and Solvents for Extracting Oilseeds and Non-petroleum Oils, P.J. Wan and P.J. Wakelyn, Eds., AOCS Press, Champaign, IL, 1997, p.311-322. L. A. Johnson, Theoretical, Comparative, and Historical Analyses of Alternative Technologies for Oilseed Extraction, Ibid, p.4-47.
14.10 Food industry
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
949
D.K. Bredeson, J. Am. Oil Chem. Soc., 60, 163A (1983). Cottonseed and Cottonseed Products, A.E. Bailey, Ed., Interscience Publishers Inc., New York, pp. 5, 615-643,(1948). V.D. Anderson, U.S. Patent 647,354 (1900). Anonymous. J. Am. Oil Chem. Soc., 54, 202A (1977). J. Pominski, L.J. Molaison, A.J. Crovetto, R.D. Westbrook, E.L.D’Aquin, and W.F. Guilbeau, Oil Mill Gaz., 51(12), 33 (1947). O.K. Hildebrandt, Fette Seifen Anstrichm, 46, 350 (1939). M. Bonotto, Oil & Soap, 14, 310 (1937). K.S. Markley, Oil Mill Gaz., 51(7), 27 (1947). D. Wesson, Oil & Soap, 10, 151 (1933). E. Bernardini, J. Am. Oil Chem. Soc., 53, 275 (1976). K.W. Becker, ibid., 55, 754 (1978). K.W. Becker, Oil Mill Gaz., 84, 20 (1980). W.E. Meyerweissflog, Oil & Soap, 14, 10 (1937). W.H. Goss, ibid, 23, 348 (1946). W.H. Goss, J. Am. Oil Chem. Soc., 29, 253 (1952). R.P. Hutchins, ibid, 54, 202A (1977). G. Karnofsky, ibid, 26, 564 (1949). A.E. MacGee, Oil and Soap, 14, 322 (1937). A.E. MacGee, ibid, 14, 324 (1937). A.E. MacGee, J. Am. Oil Chem. Soc., 26, 176 (1949). A.E. MacGee, Oil Mill Gaz., 52, 17,35 (1947). A.E. MacGee, ibid, 67, 22 (1963). H. Rosenthal, and H.P. Trevithick, Oil and Soap, 11, 133 (1934). I.J. Duncan, J. Am. Oil Chem. Soc., 25, 277 (1948). O.R. Sweeney, and L.K. Arnold, ibid, 26, 697 (1949). A.C. Beckel, P.A., Belter, and A.K. Smith, ibid, 25, 7 (1948). L.L. McKinney, F.B. Weakley, R.E. Campbell, A.C. Eldridge, J.C. Cowan, J.C. Picken, and N.L. Jacobson, ibid, 34, 461 (1957). T.A. Seto, M.O. Shutze, V. Perman, F.W. Bates, and J.M. Saulter, Agric. Food Chem., 6, 49 (1958). F.K. Rao, and L.K. Arnold, J. Am. Oil Chem. Soc., 35, 277 (1958). E. W. Lusas, L.R. Watkins and K.C. Rhee, in Edible Fats and Oils Processing: Basic Principles and Modern Practices, D. R. Erickson, Ed., AOCS Press, IL, p. 56, 1990. R.J. Hron, Sr., S. P. Koltun and A. V. Graci, J. Am. Oil Chem. Soc., 59(9), 674A (1982). R.J. Hron, Sr., M.S. Kuk, G. Abraham and P. J. Wan, ibid, 71(4), 417 (1994). L.K. Arnold and R.B.R. Choudhury, ibid, 37, 458 (1960). K.S. Olson, Oil Mill Gaz., 85, 20 (1980). A.L. Ayers and J.J. Dooley, J. Am. Oil Chem. Soc., 25, 372 (1948). P. J. Wan, D. R. Pakarinen, R. J. Hron, Sr., and E. J. Conkerton, ibid, 72(6), 653 (1995). M.P. Doss, Physical Constants of the Principal Hydrocarbons, The Texas Company, Third Edition, New York. (1942). L.A. Johnson and E. W. Lusas, J. Am. Oil Chem. Soc., 60(2), 229 (1983). McGraw-Hill Encyclopedia of Science and Technology, 10, 71 (1960). W.C. Bull and T.H. Hopper, Oil and Soap, 18, 219 (1941). H.B. Coats and G. Karnofsky, J. Am. Oil Chem. Soc., 27, 51 (1950). R.D. Good, Oil Mill Gaz., 75, 14 (1970). H.B. Coats and M.R. Wingard, J. Am. Oil Chem. Soc., 27, 93 (1950). W.C. Bull, Oil and Soap, 20, 94 (1943). C.G. Reuther, Jr., R.D. Westbrook, W.H. Hoffman, H.L.E. Vix and E.A. Gastrock, J. Am. Oil Chem. Soc., 28, 146 (1951). L.K. Arnold and D.J. Patel, ibid, 30, 216 (1953). M.R. Wingard and R.C. Phillips, ibid, 28, 149 (1951). M.R. Wingard and W.C. Shand, ibid, 26, 422 (1949). E.J. Conkerton, P.J. Wan and O.A. Richard, ibid, 72, 963 (1995). P.J. Wan, INFORM, 7, 624 (1996).
950
George Wypych
14.11 GROUND TRANSPORTATION George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The ground transportation industry in the USA is dominated by truck freight (78.6%). Other methods of transportation include: rail (7.9%), water (5.2%), air (4%), pipeline (2.2%), and other (2.1%). Solvents are used and solvent wastes and emissions are generated during refurbishing and maintenance. Rail car refurbishing involves stripping and painting. Paint is usually removed by mechanical means (steel grit blast system) but solvents are occasionally used. Solid wastes are generated from latex paint wastes but hazardous wastes are also generated from solvent-based paints and thinners. Parts cleaning is mostly done using mineral spirits. Waste solvents are sent off-site for reclamation. Truck maintenance work usually requires a parts washer which may involve either a heated or ambient temperature solvent, hot tank, or a spray washer. In the solvent tank washer, solvent (usually mineral spirits, petroleum distillates, and naphtha) is recirculated from solvent tank. Spent solvent is usually replaced monthly. Carburetor cleaning compounds contain dichloromethane. Tanker cleaning often involves a solvent spray. The ground transportation industry employs a large number of people (more than 2 million in the USA). It is one of the less polluting industries. It generates 0.3% of VOC released by all major industries combined (about half of that of the aerospace industry). Most solvents used are of low toxicity. Good system of collection and reclamation of solvent wastes is done effectively and this is the major reason for the relatively good performance of the industry.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Ground Transportation Industry. Trucking, Railroad, and Pipeline. US Environmental Protection Agency, 1997. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.12 INORGANIC CHEMICAL INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
This industry has two major sectors: inorganic chemicals and chlor-alkali. Inorganic chemicals are often of mineral origin processed to basic chemicals such as acids, alkalies, salts, oxidizing agents, halogens, etc. The chlor-alkali sector manufactures chlorine, caustic soda, soda ash, sodium bicarbonate, potassium hydroxide and potassium carbonate. The major processes in this industry do not use solvents but there are many specialized auxiliary processes which use solvents. Tables 14.12.1 and 14.12.2 give information on the reported solvent releases and transfers from inorganic chemical industry.
14.13 Iron and steel industry
951
The tables show that the industry, which operates almost 1,500 plants and employs over 110,000 people, has minimal impact on global emission of VOCs. Consequently, the industry does not have any major initiative to deal with solvent emissions or wastes. Future safety improvements concentrate on non-solvent issues.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Inorganic Chemical Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
Table 14.12.1. Reported solvent releases from the inorganic chemical plants in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
33,000
hexane
2,000
carbon tetrachloride
3,000
methanol
574,000
chloromethane
2,600
methyl ethyl ketone
460
dichloromethane
12,500
N-methyl-2-pyrrolidone
180
ethyl benzene
110
toluene
12,000
ethylene glycol
1,800
xylene
1,500
Table 14.12.2. Reported solvent transfers from the inorganic chemical plants in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
780
methyl ethyl ketone
9,000
carbon tetrachloride
6,400
N-methyl-2-pyrrolidone
8,700
dichloromethane
5,000
toluene
6,000
ethylene glycol
12,000
xylene
96,000
14.13 IRON AND STEEL INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
With almost 1,400 plants, the US iron and steel industry is very diverse industry having total sales of $100 billion and over 400,000 employees. Figure 14.13.1 is a schematic diagram of the iron and steel making process. Only one stage − finishing − employs solvents. The finishing stage includes processes to remove mill scale, rust, oxides, oil, grease and soil prior
952
George Wypych
Figure 14.13.1. Schematic diagram of operations in the iron and steel manufacturing process. [Reproduced from EPA Office of Compliance Sector Notebook Project. Reference 1.]
Table 14.13.1. Reported solvent releases from the iron and steel plants in 1995 [Data from Ref. 2] Solvent benzene
Amount, kg/year 321,000
Solvent N-methyl-2-pyrrolidone
Amount, kg/year 3,600
n-butyl alcohol
26,000
polycyclic aromatic compounds
2,400
cresol
1,800
tetrachloroethylene
91,000
dichloromethane
318,000
1,2,4-trimethylbenzene
17,000
ethylbenzene
5,000
trichloroethylene
620,000
methanol
241,000
toluene
261,000
methyl ethyl ketone
358,000
xylene
168,000
to coating. Methods used include solvent cleaning, pressurized water or air blasting, cleaning with abrasives, and alkaline or acid pickling. Tables 14.13.1 and 14.13.2 give information on the reported solvent releases and transfers from the iron and steel industries. Not all the solvents listed in the tables are used in processing. Some are by-products of coke manufacture from coal. Benzene and polycyclic aromatics compounds are by-products. Strong solvents such as methyl ethyl ketone, toluene, xylene, and trichloroethylene are typical of those used in cleaning processes. There is no program formulated by the industry to reduce amounts of solvents used.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Iron and Steel Industry. Trucking, Railroad, and Pipeline. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.14 Lumber and wood products
953
Table 14.13.2. Reported solvent transfers from the iron and steel plants in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
3,000
N-methyl-2-pyrrolidone
11,500
n-butyl alcohol
1,400
polycyclic aromatic compounds
3,820,000
cresol
12
tetrachloroethylene
20,000
dichloromethane
14,500
1,2,4-trimethylbenzene
3,600
ethylene glycol
197,000
trichloroethylene
165,000
ethylbenzene
550
toluene
11,500
methanol
25
xylene
14,000
methyl ethyl ketone
66,000
14.14 LUMBER AND WOOD PRODUCTS - WOOD PRESERVATION TREATMENT: SIGNIFICANCE OF SOLVENTS Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Tòbingen, Tòbingen, Germany
Gerhard Volland Otto-Graf-Institut, Universit¬t Stuttgart, Stuttgart, Germany
14.14.1 GENERAL ASPECTS Wood preservation is based on various fundamental principles, e.g., construction aspects such as exposure to humidity, selection of different types of wood products according to their durability, and chemistry of wood preservatives. Important groups of chemical wood preservatives are water-soluble and solvent-based substances.1 The main requirements of chemical wood preservatives are:1 • Stability, especially chemical stability. • Resistance to environmental conditions, e.g., light or heat. • Penetration into the wood products. • Effectiveness against wood attacking agents (e.g., insects, fungi, bacteria). • Compatibility with other construction components, e.g., paints, adhesives, and fasteners. • Construction aspects, e.g., corrosion. • Minimal environmental impact, e.g., minimum emissions or minimum environmental pollution. • Ability to work with a range of wood products. • Case applications, e.g. fundamental differences of indoor and outdoor coatings. • Having favorable visual aspects, e.g., surface properties, color, uniformity, influence on grain pattern, etc. All requirements cannot be fulfilled completely by the various wood preservatives. Therefore wood preservatives should be selected according to the particular case.
954
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
14.14.2 ROLE OF SOLVENTS 14.14.2.1 Occurrence Various solvents are added to wood preservatives. Only a limited number of wood preservatives are authorized by governmental agencies, e.g., in Germany “Institut für Bautechnik (DIBt)”.2 But there is a large grey market for wood preservatives other than the authorized substances. As a result, a large variety of solvents can occur in wood preservatives. Different systems of classification are used worldwide. In Germany, authorized wood preservative substances are published in an index of wood preservatives (“Holzschutzmittelverzeichnis”)2 which is elaborated by the DIBt and the central German environmental authority (“Umweltbundesamt”). A systematic survey of wood preservatives is shown in Table 14.14.1, including wood preservatives containing solvents. Solvents are normally found in wet systems of wood preservatives. Commonly used solvents are substances which are applied in connection with normally used binders (aldehyde resins, acrylates and polyurethanes). Water or the appropriate solvents are added to binders. Table 14.14.1. Systems of wood preservatives Purpose and base
Terms “CF-salts” “CFA-salts”
Water-soluble agents as preventive treatment against fungi and insects
“SF-salts” “HF-salts” “B-salts” “Single CK-salts” “CKA-salts” “CKB-salts” “CKF-salts” “CFB-salts” collective group
Oily agents as preventive treatment against fungi and insects
tar oil preparations preparations containing solvents pigment-free preparations containing binders and solvents preparations with stained pigments containing solvents special preparations only used in stationary installations preparations containing coal tar oil
Preparations used for special applications
Active components chromium and fluorine compounds alkali fluorides, alkali arsenate, and bichromate (no longer permitted) silicofluorides hydrogen fluorides inorganic boron compounds copper salts, bichromate copper salts, bichromate with arsenic compounds copper salts, bichromate with boron compounds copper salts, bichromate with fluorine compounds chromium, boron and fluorine compounds other compounds, e.g., bis(N-cyclohexyldiazeniumdioxyl)-copper distillates of bituminous coal tar (carbolineum) organic fungicides and insecticides organic fungicides and insecticides organic fungicides and insecticides organic fungicides and insecticides organic agents, special distillates containing coal tar oil, solvents and pigments
pastes wood preservatives used in particle board in manufacturing plants agents used as preventive treatment against insects contain organic insecticides
14.15 Medical applications
955
Solventborne wood preservatives contain mainly nonpolar, organic solvents apart from other substances such as fungicides and insecticides.1 These solvents are classified as VOCs. 14.14.2.2 Technical and environmental aspects Solvent-based wood preservatives show several advantages, especially in their application and technical effectiveness.1 They can be applied repeatedly and do not alter the structure of the wood products. Application is faster and the characteristics of the final product are improved, e.g., visual appearance of surface. Nevertheless, there are some disadvantages, especially environmental ones. Most solvents are released quickly (VOCs) and can cause severe environmental effects. This is especially true if toxic solvents are employed. Emissions of solvents from wood products are described under various conditions, e.g. indoor air emissions from furniture or emissions in test chambers.3,4 Solvents can be emitted as primary or reactive products of the wood product or the coating system; solvents can also be investigated as secondary emission products.3,5 The emission characteristics depend on solvent properties and surrounding conditions, e.g., air velocity and air exchange rate.6 In the indoor air, solvents from wood products follow various pathways. Examples of interactions are possible reactions of solvents (e.g., styrene) with air components (e.g., hydroxy radicals),3 transport into and through indoor materials7 or sorption processes.5 The emitted solvents can be reduced by ventilation processes or they may be absorbed by organisms. For humans, absorption of the wood preservatives or ingredients (e.g., solvents) can cause various toxic effects. It is often difficult to pinpoint the causative agents (see Chapter 20). REFERENCES 1 2 3 4 5 6 7
Ullmann`s Encyclopedia of Industrial Chemistry, 1998. DIBt (Deutsches Institut für Bautechnik). Holzschutzmittelverzeichnis. Index of wood preservatives (1999). T. Salthammer, A. Schwarz, F. Fuhrmann, Atmospher. Environ., 33, 75 (1999). T. Salthammer, Atmospher. Environ., 7, 189 (1997). M. Wensing, H.J. Moriske, T. Salthammer, Gefahrstoffe Reinhaltung der Luft, 58, 463 (1998). E. Uhde, A. Borgschulte, T. Salthammer, Atmospher. Environ., 32, 773 (1998). R. Meininghaus, T. Salthammer, H. Knoppel, Atmospher. Environ., 33, 2395 (1999).
14.15 MEDICAL APPLICATIONS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Industries manufacturing medical devices use a wide variety of technological processes which most likely take advantage of most of the available solvents. The range of solvent use is so wide that a complete description of each solvent and its application is not possible in this book. It is questionable if such analysis is possible given that many processes are guarded by trade secrets where there is no patent disclosure. Some examples are given,
956
George Wypych
more to show that, although solvents do contribute to pollution, they also help to produce materials which are needed for health and well being. Polyurethanes are materials which have the required properties and biocompatilibity which makes them good candidate for use in medical devices. Common applications include pacemaker leads, peripheral and central catheters, feeding tubes, balloons, condoms, surgical gloves, instrument and appliance covers, wound dressings, and many other.1-4 Several methods are used to process polyurethanes. These include injection molding, extrusion, and solution processing. In solution processing film casting and dip molding are the most frequent techniques. Dimethylacetamine, tetrahydrofuran, dichloromethane, methyl ethyl ketone, N,N-dimethylformamide, N-methylpyrrolidone, cyclopentanone, cyclohexanone, dioxane, and chloroform are the most commonly used solvents. Most of these are hazardous but used because they contribute to highly transparent product which is very desirable in medical devices. Transparent materials can only be made from transparent solutions.1 These solvents can dissolve polymers well and form clear solutions. Ease of solvent removal from the material is very important in formulation design. Obviously, no traces of solvents should remain in the medical devices since even trace amounts may interfere with the treatment and the patient’s health. An inappropriate solvent selection may cause the formation of crust as the solvent escapes. This leads to material discontinuity (e.g., pinholes) which renders the product inferior. This brings a discussion of solvent evaporation, the rheological properties of formulation, and formation of multilayer materials. Good solvents can be used in lower concentration but they result in viscous solutions which, in dip coating, form thick films which have the potential of blistering on evaporation. If the solution is diluted, film continuity suffers which increases the number of pinholes. Rapid evaporation causes a formation of a crust of gelled solidified polymer which makes solvent removal more difficult and damages the integrity of the layer. Also, material does not have time to adjust and leveling suffers. On the other hand slow evaporation may cause dissolution of the layer below the coating in a multilayered products and bubbling between the layers. The selection of solvents for dip coating is usually a complex process ultimately requiring multicomponent solvent mixtures which include a good solvent, a poor solvent, and a solvent of lower boiling point (sometime called “blush resistor”) to balance viscosity and rate of evaporation.2 In wound dressings, the solvents selected affect the material microstructure which controls the evaporation of exuded body fluids but prevents bacteria and pathogens from entering the wound.3 In infection-resistant medical devices, the antimicrobial agent must be uniformly distributed over all areas of the medical device which may come into contact with a patient. Otherwise there is a risk of infection.4 Not all solvents dissolve antimicrobial agents and swell surface of medical device. Cleaning of penetrable septa, tubing systems, and infusion and dialysis systems is another application in which solvents are used. The solvents which are suitable for elastomer cleaning are dichloromethane, perchloroethylene, halogenated hydrocarbons, and freons.5 This cleaning method extracts undesirable organic materials from medical devices which might otherwise be extracted by body fluids. Heat treatment of catheters followed by washing with a polar solvent increase its surface lubricity. Catheter with poor surface lubricity often causes frictional pain upon its insertion into the body cavity and damages the mucosal tissue resulting in cross infection.6 Film dressings contain two types of solvents: solvents to
14.16 Metal casting
957
dissolve the polymer and propellant solvents. These must be selected to achieve technological goals related to solubility and compatibility.7 These examples show that the many technological considerations place constraints on in solvent selection. Solvent replacement in complex products and technological processes is a long-term, expensive proposition which usually results in a need for complete reformulation of the material with failure to achieve the objective a very possible outcome.
REFERENCES 1 2 3 4 5 6 7
A J Walder, Plast. Eng., 54, No.4, 29-31 (1998). M T Shah, US Patent 5,571,567, Polygenex International, Inc., 1996. J Delgado, R J Goetz, S F Silver, D H Lucast, US Patent 5,614,310, 3M, 1997. S Modak, L Sampath, US Patent 5,567,495, Columbia University, 1996. S H Smith, J M Brugger, H W Frey, US Patent 5,639,810, COBE Laboratories, Inc., 1997. L Mao, Y Hu, D Piao, US Patent 5,688,459, China Rehabilitation Research Center, 1997. A J Tipton, S M Fujita, R L Dunn, US Patent 5,792,469, Atrix Laboratories, Inc., 1998.
14.16 METAL CASTING George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The metal casting industry has 3,100 facilities in the USA and employs 250,000 people. Most plants are small and technological processes are very diverse. The processes do share common phases, including pattern making, mold and core preparation, furnace charge and metal melting, mold charging, cooling and finishing. Most steps use solvents. In the mold-making process, a many chemical binding systems are used, some of which contain methanol, benzene, toluene, and cresol. The metal is most often recycled and it typically requires cleaning before it is charged to the furnace. This is accomplished either by precombustion or solvent cleaning. In die casting operations, solvent-based or water-based lubricants are used. Die casters also use die fluxes which contain solvents. Some solvent replacement additives in water-based lubricants contain hazardous solvents. Finishing operations involve casting cleaning to remove scale, rust, oxides, oil, grease, and dirt. Solvents are typically chlorinated solvents, naphtha, toluene, and methanol. Cleaning can also be done by emulsifiers, abrasives, alkaline agents, and acid pickling. The cleaning operation is usually followed by painting which frequently involves solvent-based paints and thinners. Tables 14.16.1 and 14.16.2 contain information on the reported solvent releases and transfers from metal casting industry. The data show that solvent use is not excessive relative to other industries. The industry plans to further improve its environmental record by developing environmentally improved materials which meet regulations. The solvent cleaning and die lubrication are processes under study.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Metal Casting Industry. US Environmental Protection Agency, 1998. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
958
George Wypych
Table 14.16.1. Reported solvent releases from the metal casting industry in 1995 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
110,000
methyl ethyl ketone
22,000
n-butyl alcohol
15,000
methyl isobutyl ketone
22,000
cresol
20,000
N-methyl-2-pyrrolidone
41,000
dichloromethane
50,000
tetrachloroethylene
13,000
ethylbenzene
10,500
1,1,1-trichloroethane
111,000
ethylene glycol
64,000
trichloroethylene
75,000
hexachloroethane
16,000
toluene
233,000
methanol
5,860,000
xylene
388,000
Table 14.16.2. Reported solvent transfers from the metal casting industry in 1995 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
115
N-methyl-2-pyrrolidone
22,000
ethylbenzene
340
1,1,1-trichloroethane
500
ethylene glycol
50,000
trichloroethylene
1,000
methanol
10,000
toluene
4,000
methyl ethyl ketone
8,000
xylene
82,000
14.17 MOTOR VEHICLE ASSEMBLY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Automotive industry in US alone employs 6.7 million person and it is a large contributor to the gross national product. It uses large quantities of solvents and it is perceived to contribute to pollution by solvents and other materials.1-6 Solvents are used in a variety of cleaning, preparation, and painting operations. Automotive finishing process may be divided into four main categories: anti-corrosion operations (cleaning, phosphate treatment, and chromic acid treatment), priming operations (electrodeposition of primer, anti-chip coating application, and primer application), joint sealant application, and other finishing operations (color coat, clear coat). These main operations employ many materials which contain solvents.
14.17 Motor vehicle assembly
959
The cleaning process involves acid/alkaline and solvent cleaning. Typical solvents involved are acetone, xylene, toluene, and 1,1,1-trichloroethylene. The primer bath is water-based but usually some organic solvents are present (5-10%). These solvents are the same as those listed above. After the application of primer, the car body is baked and then undergoes waterproofing with an application of polyvinylchloride sealant which contains a small amount of solvents. Following waterproofing, the automotive body proceeds to the anti-chip booth, where urethane or epoxy solvent-coating systems are applied. This process is followed by application of primer-surfacer coating which is either a polyester or an epoxy ester in a solvent system. The primer-surfacer coating is applied by spraying and provides a durable finish which can be sanded. After the sanding step, the primary color coating is applied also by spraying. These primary color formulations contain about twice as much solvent as the primer-surfacer coating. Solvents are flashed-off (no heating) and a clear coat is applied. Then the entire car body is baked for about 30 min. Solvents used include butanol, isobutanol, methanol, heptane, mineral spirits, butyl acetate ethyl acetate, hexyl acetate, methyl ethyl ketone, acetone, methyl amyl ketone, toluene, and xylene. Several finishing operations also employ solvents. After baking, a sound-deadener is applied to certain areas of the underbody. It is a solvent based material with a tar-like consistency. A trim is applied with adhesives which contain solvents (see section on adhesives and sealants). After the installation of trim and after the engine is installed, car undergoes an inspection. Some repainting is required in about 2% of the production. If damage is minor then repainting is done by a hand operated spray gun. If the damage is substantial a new body is installed. Equipment cleaning solvents are also used. Spraying equipment is cleaned with a “purge solvent” which may consist of a mixture of dimethylbenzene, 4-methyl-2-pentanone, butyl acetate, naphtha, ethyl benzene, 2-butanone, toluene, and 1-butanol. Tables 14.17.1 and 14.17.2 contain information on the reported solvent releases and transfers from the motor vehicle assembly industry. The data show that solvent use is very large compared with all industries covered so far in our discussion except for the steel and iron industry. The motor vehicle assembly industry is the sixth largest producer of VOC and also the sixth largest industry in reported emissions and transfers. The data in Tables 14.17.1 and 14.17.2 are data from 1995 the most recent available. The automotive industry and associated paint companies conduct extensive work on replacement of VOC containing paint systems. These efforts are mainly directed to water-based systems and powder coatings. Until recently, water-based systems were preferred but now attention is shifting to powder coatings which eliminate VOC. There is no status quo. Changes are dynamic and kept protected by trade secrets which makes it difficult to comment on specific progress. Solvent use by the European industry5 is that the production of one car requires an average of 10 kg of solvents. Solvents use is not the only problem the industry is facing. 16% of the total energy used in car production is required by painting and finishing operations. Both energy conservation and reduction is solvent consumption must be pursued to meet environmental objectives. Not only can these issues be addressed through material reformulation but the design of equipment used in applying and drying the coating can also reduce emission and save energy. A new trend is apparent as plastics are introduced to automotive production. Plastic parts must also be painted. Paint systems are difficult to select. Chlorinated polyolefins provide good adhesion of paints and reduce VOC but are also under scrutiny because of pres-
960
George Wypych
Table 14.17.1. Reported solvent releases from the motor vehicle assembly industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
13,000
methyl ethyl ketone
2,320,000
n-butyl alcohol
2,260,000
methyl isobutyl ketone
3,060,000
sec-butyl alcohol
86,000
N-methyl-2-pyrrolidone
193,000
tert-butyl alcohol
4,200
methyl tert-butyl ether
32,000
cyclohexane
35,000
tetrachloroethylene
140,000
dichloromethane
380,000
1,1,1-trichloroethane
730,000
ethylbenzene
1,370,000
trichloroethylene
1,300,000
ethylene glycol
180,000
1,2,4-trimethylbenzene
1,120,000
isopropyl alcohol
9,000
toluene
2,610,000
hexane
95,000
xylene
10,800,000
methanol
1,550,000
m-xylene
25,000
Table 14.17.2. Reported solvent transfers from the motor vehicle assembly industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
3,400
methyl ethyl ketone
2,100,000
n-butyl alcohol
1,030,000
methyl isobutyl ketone
4,700,000
sec-butyl alcohol
9,000
N-methyl-2-pyrrolidone
330,000
tert-butyl alcohol
1,000
methyl tert-butyl ether
2,300
cyclohexane
670
tetrachloroethylene
49,000
dichloromethane
450,000
1,1,1-trichloroethane
140,000
ethylbenzene
1,740,000
trichloroethylene
480,000
ethylene glycol
605,000
1,2,4-trimethylbenzene
330,000
isopropyl alcohol
2,000
toluene
2,020,000
hexane
25,000
xylene
9,200,000
methanol
760,000
m-xylene
2,100
ence of chlorine. Powder coatings are available7 but they require a high energy input. These problems are apparent but the solution to them will take several years to implement due, in large part, to the long term testing needed to confirm coating performance (up to 5 years in Florida).
14.18 Organic chemical industry
961
Table 14.18.1. Reported solvent releases from the organic chemical industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
allyl alcohol
31,000
ethylbenzene
370,000
benzene
690,000
ethylene glycol
6,050,000
n-butyl alcohol
850,000
hexane
600,000
sec-butyl alcohol
63,000
isopropyl alcohol
150
tert-butyl alcohol
430,000
methanol
8,750,000
carbon disulfide
85,000
methyl ethyl ketone
260,000
carbon tetrachloride
10,000
methyl isobutyl ketone
520,000
chlorobenzene
27,000
N-methyl-2-pyrrolidone
350,000
chloroform
7,000
methyl tert-butyl ether
64,000
cresol
280,000
pyridine
120,000
m-cresol
320,000
tetrachloroethylene
20,000
o-cresol
270,000
toluene
1,040,000
p-cresol
162,000
1,2,4-trichlorobenzene
41,000
cyclohexane
450,000
1,1,1-trichloroethane
130,000
cyclohexanol
1,100,000
trichloroethylene
18,000
dichloroethane
120,000
xylene
350,000
1,2-dichloroethylene
70
m-xylene
59,000
dichloromethane
310,000
o-xylene
34,000
N,N-dimethylformamide
25,000
p-xylene
660,000
1,4-dioxane
12,000
REFERENCES 1 2 3 4 5 6 7
EPA Office of Compliance Sector Notebook Project. Profile of the Motor Vehicle Assembly Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. G. Wypych, Ed., Weathering of Plastics. Testing to Mirror Real Life Performance, Plastics Design Library, Society of Plastics Engineers, New York, 1999. M Harsch, M Finkbeiner, D Piwowarczyk, K Saur, P Eyerer, Automotive Eng., 107, No.2, 211-4 (1999). C A Kondos, C F Kahle, Automotive Eng., 107, No.1, 99-101 (1999). D C Shepard, J. Coat. Technol., 68, No.857, 99-102 (1996). T Hosomi, T Umemura, T Takata, Y Mori, US Patent 5,717,055, Mitsubishi Gas Chemical Company, Ltd., 1998.
962
George Wypych
Table 14.18.2. Reported solvent transfers from the organic chemical industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
allyl alcohol
210,000
ethylbenzene
980,000
benzene
420,000
ethylene glycol
6,800,000
n-butyl alcohol
1,500,000
hexane
770,000
sec-butyl alcohol
1,700,000
isopropyl alcohol
85,000
tert-butyl alcohol
12,500,000
methanol
23,000,000
carbon disulfide
96,000
methyl ethyl ketone
800,000
carbon tetrachloride
12,000
methyl isobutyl ketone
390,000
chlorobenzene
130,000
N-methyl-2-pyrrolidone
110,000
chloroform
92,000
methyl tert-butyl ether
210,000
cresol
430,000
pyridine
33,000
m-cresol
720,000
tetrachloroethylene
138,000
o-cresol
57,000
toluene
4,400,000
p-cresol
870,000
1,2,4-trichlorobenzene
8,000
cyclohexane
900,000
1,1,1-trichloroethane
290,000
cyclohexanol
3,700
trichloroethylene
42,000
dichloroethane
230,000
xylene
4,000,000
1,2-dichloroethylene
1,000
m-xylene
51,000
dichloromethane
870,000
o-xylene
460,000
N,N-dimethylformamide
370,000
p-xylene
1,700
14.18 ORGANIC CHEMICAL INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The chemical industry operates about 1000 plants in the USA with 53 companies producing 50% of the total output of $65 billion in sales and employing 125,000 people. There are point source solvent emissions (e.g., laboratory hoods, distillation units, reactors, storage tanks, vents, etc.), fugitive emissions (e.g., pump valves, flanges, sample collectors, seals, relief devices, tanks), and secondary emissions (waste water treatment units, cooling towers, spills). Organic liquid wastes containing solvent are generated from processes such as equipment washing, surplus chemicals, product purification, product reaction, housekeeping, etc.
14.19.1 Architectural surface coatings
963
Tables 14.18.1 and 14.18.2 give the reported solvent releases and transfers from the organic chemical industry. Large quantities of solvents are involved. The organic chemical industry produced the second largest quantity of VOC and the second largest releases and transfers. The industry is actively working to reduce solvent use because of the high costs (waste treatment, fines, liabilities, etc). There are many efforts under way to reduce environmental emissions and improve safe practices. The initiatives include process modifications such as a reduction in non-reactive materials (e.g., solvents) to improve process efficiency, a reduction in the concentration of chemicals in aqueous solution, and improved R&D and process engineering. Equipment modifications are planned to reduce leaks, prevent equipment breakdown, and improve the efficiency of emission control devices.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Organic Chemical Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.19 PAINTS AND COATINGS 14.19.1 ARCHITECTURAL SURFACE COATINGS AND SOLVENTS
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Tòbingen, Tòbingen, Germany
Gerhard Volland Otto-Graf-Institut, Universit¬t Stuttgart, Stuttgart, Germany
14.19.1.1 General aspects Coating materials and coating techniques can be distinguished and systematized in various ways. The fundamental principles of common coating systems are:1 • Physical drying. A solid surface film is formed after the evaporation of water or organic solvents. • Physico-chemical drying/curing. Polycondensation or polyaddition are combined with evaporation of organic solvents. • Chemical curing. Solvents, e.g., styrene or acrylic monomers, react with the curing system. The actual effects depend on the surrounding conditions and the ingredients of the coating system, e.g., solvents. Solvents contribute many essential properties to coating systems. Solvents can improve technical factors such as application or surface properties. Solvents also bring negative qualities to coating materials, especially with respect to environmental conditions (e.g., toxic effects of emitted organic solvents). 14.19.1.2 Technical aspects and properties of coating materials Application techniques for coatings can be considered in various ways. The stability and durability of coating is essential. Coatings that have normal wear and tear requirements are
964
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
mainly based on oils and aldehyde resins. Higher durability or stability can be achieved by the use of one of the following one- or two-component systems. One-component: • Bituminous materials • Chlorinated rubber • Polyvinyl chloride • Polyacrylic resin • Polyethene • Saturated polyester • Polyamide Two-component: • Epoxy resin • Polyurethane • Mixtures of reactive resins and tar A survey of the performance of different coating materials together with an assessment of various environmental factors is given in Table 14.19.1.1. Table 14.19.1.1. Environmental performance of some coating materials
Abbreviation
Weathering response
Acid atmosphere
Humidity
Under water
acid
alkali
Solvent
Temperature <60oC
abrasion
scratch
Mechanical
Material
Chemical stress
plant oil (linseed oil)
OEL
+y,s,f
+/-
+/-
-
-
-s
-
+
-
-
alkyd resin
AK
+
+/-
+/-
-
-
-s
-
+
-
-
B
+/-e,f
+
+
+
+/-
+
-
+/-
-
-
chlorinated rubber
RUC
+/-b,c,f
+
+
+
+
+
-
+
+/-
+/-
PVC-soft
PVC
+/-e,f
+
+/-
+/-
+
+/-
-
+
+/-
+/-
Polyacrylic resin
AY
+
+
+
+
+
+
-
+
+
+/-
polyethene
PE
+/-e
+
+
+
+
+
+
+
+
+
polyester (saturated)
SP
+
+
+/-
+/-
+
+/-(s)
+
+
+
+
epoxy resin
EP
+/-c,g
+
+
+
+/-
+
+
+
+
+
polyurethane
PU
+
+
+/-
+/-
+
+/-
+/-
+
+
+
epoxy resin tar
EP-T
+/-c,e
+
+
+
+/-
+
+/-
+
+
+
polyisobutylene
PLB
+/-
+
+
+
+
+
-
+
+
+
bitumen
+ = suitable, +/- = limited suitability, - = unsuitable, () = less distinct; b = bleaching, c = chalking, e = embrittlement, f = acid fragments, g = loss of gloss, s = saponification, y = yellowing
14.19.1 Architectural surface coatings
965
The following are examples of architectural surface coatings. 14.19.1.2.1 Concrete coating materials According to DIN 1045 concrete must be protected against aggressive substances if the chemical attack is severe and long-term. These are the requirements: • Good adhesion • Waterproof and resistant to aggressive substances and resistant to the alkalinity of the concrete • Deformable To realize these requirements, the following technical solutions can be applied: • The use of paint coatings based on duroplast or thermoplastic substances • Surface treatment - impregnation Protective coatings based on polymers are used in construction with and without fillers and with or without reinforcing materials (fibers). The following techniques are generally used: • Film forming paint coatings (brushing, rolling, spraying). • Coating (filling, pouring). During application, the coating materials are normally liquid and subsequently harden by evaporation of solvents or as a result of chemical reactions. The common coating systems for concrete are listed in Table 14.19.1.2. Synthetic resins (e.g., chlorinated rubber, styrene resins, acrylic resins) contain normally 40-60% solvents and form a thin film. Several coats must be applied. Reactive resins may require little or no solvent. Table 14.19.1.2. Survey of often used coating systems for concrete Film-forming agents
Hotmelts
Film-forming agents dissolved
Film-forming agents emulsified
Bituminous substances
+
+
+
Synthetic resin
+
+
Reactive resins
(+)
(+)
14.19.1.2.2 Coating materials for metals Wet coating processes are detailed in Table 14.19.1.3.
Reaction resin liquid
+
966
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
Table 14.19.1.3. Wet coatings - coating materials for metals Abbreviation oil Natural substances
Plastic materials
Application
State
OEL
Curing method oxidizing
binders for wet coating
bitumen
B
physical
tar
T
physical
alkyd resin
AK
oxidizing
polyurethane
PU
epoxy resin
EP
liquid binders for wet coating
chemical chemical
acrylic resin
AY
physical
vinyl resin
PVC
physical
chlorinated
RUC
physical
rubber
The chemical properties of binders are not affected by the drying process if coatings are of physically drying type. After solvent has evaporated, the polymer molecules become intermeshed, thus producing the desired coating properties. In contrast to physical drying, binders based on reactive resins such as epoxy resins, polyurethanes or polyesters consist of two components (liquid resin and curing agent) which are either mixed shortly before the coat is applied, or, in the case of a one component system, one which is applied as slow reacting mixture. The setting reaction occurs at the surface of the coated material. The final products are normally more resistant and more compact than products based on physically drying binders. Pretreatment of substrate is more critical for applications where chemically hardening products are applied. Coatings can be more or less permeable to water vapor and oxygen. Damage to the metal substrate can occur if water and oxygen reach the reactive surface simultaneously. This is normally impossible if the coatings adhere well and the coated surface is continuous. The adhesion of the coating also prevents the penetration by harmful substances via diffusion processes. Adhesion is enhanced by adsorption and chemical bonds. 14.19.1.2.3 Other coating materials Other aspects of solvents contained in paint coatings and varnishes are given in Chapter 18.3. The market offers a wide spectrum of coating systems. Examples of common industrial coating materials are listed in Table 14.19.1.4 and relevant aspects concerning application and environmental or health risks are also included.
14.19.1 Architectural surface coatings
967
Table 14.19.1.4. Examples of coating materials containing hazardous ingredients, especially solvents3 Product group/ compounds
Application
Hazardous ingredients, especially solvents
Relevant health and environmental risks
corrosion protection/primer, application by brush, spray, airless-spraying
xylene 2,5-25%, ethylbenzene 2,5-10%, solvent naphtha 2,5-10%,1-methoxy-2propanol 1-2,5%
irritations (respiratory, skin, eyes), neurological (narcotic) effects, absorption (skin), flammable, explosive mixtures
corrosion protection/coating
solvent naphtha 10-20%, diphenylmethane isocyanate 2,5-10%
sensitization (respiratory) irritation (skin, eyes, gastrointestinal tract), neurological effects (narcotic effects, coordination), absorption (skin), flammable, water polluting
corrosion protection/coatings/rigid system
4-tert-butylphenol 10-25%, M-phenylenebis 2,5-10%, trimethylhexane, 1,6-diamine 10-25%, nonylphenol, 10-25% xylene 10-25%, ethylbenzene 2,5-10%
sensitization (skin, respiratory), irritations (skin, eye, respiratory, gastrointestinal tract), neurological (narcotic) effects (coordination), flammable, water polluting
corrosion protection/coatings/ rigid system
bisphenol A (epichlorohydrin) 25-50% oxirane, mono(C12 -C14-alkyloxy)methyl derivatives 2,5-10%, cyclohexanone 1-2,5%, 2-methylpropane-1-ol 2,5-10%, xylene 10-25%, ethylbenzene 2,5-10%, benzyl alcohol 1-2,5%, 4-methylpentane-2-one 1-2,5%
irritations (skin), sensitization (skin, contact), neurological (narcotic) effects, absorption (skin), flammable, water polluting
modified epoxy resin containing solvent
corrosion protection/top layer
bisphenol A (epichlorohydrin) 50-100%, 3-amino-3,5,5-trimethyl ethylbenzene 10-25%, xylene 10-25%
irritation (skin, eyes, respiratory), sensitization (skin), flammable, explosive gas/air mixtures, hazardous reactions (with acids, oxidizers), water polluting
filled and modified epoxy resin
coatings and corrosion prevention/ pore sealer
bisphenol A (epichlorohydrin) 50-100%, P-tert-butylphenyl1-(2,3-epoxy)propyl-ether 1-2,5%
irritations, sensitization, water polluting
filled and modified epoxy resin
coatings and corrosion prevention/ pore sealer
trimethylhexamethylendiazine1,6-diamine 10-25%, trimethylhexamethylendiamine -1,6-cyanoethylene 50-100%
irritations, sensitization, water polluting
zinc dust coating based on epoxy resin
reactive PUR-polymers containing solvents
modified polyamine containing solvents
modified epoxy resin containing solvents
968
Product group/ compounds
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
Hazardous ingredients, especially solvents
Relevant health and environmental risks
flooring/mortar screen
bisphenol A (epichlorohydrin) 25-50%, benzyl butyl phthalate 2,5-10%, xylene 2,5-10%
irritation (eye, skin, respiratory), sensitization (contact, skin), explosive gas/ air mixtures (with amines, phenols), water polluting
flooring/mortar screen
benzyl alcohol 10-25%, nonylphenol 25-50%, 4,4´-methylenebis(cyclohexylamine) 10-25%, 3,6,9-triazaundecamethylendiamine 10-25%
irritation (eye, skin, respiratory), sensitization (contact, skin), hazardous reactions (with acids, oxidizers), flammable, water polluting
filled and modified epoxy resin
corrosion protection/top layer
naphtha 2,5-10%, 2-methoxy-1-methylethylacetate 2,5-10%, 2-methylpropane-1-ol 1-2,5%, xylene 10-25%, ethylbenzene 2,5-10%
irritation (skin, eyes, respiratory), neurological (narcotic) effects, flammable, explosive, hazardous reactions (with oxidizers), water polluting
copolymer dispersion
walls, especially fungicidal properties, good adhesion in damp environment, resistant against condensed water
isothiazolone 1-2,5%, 3-methoxybutylacetate 1-2,5%
irritations (skin, eyes, gastrointestinal tract), water polluting
solvent naphtha (petroleum) 25-50 %, xylene 2,5-10%, ethylbenzene 1-2,5%
irritations (eyes, skin, respiratory, gastrointestinal tract), neurological (narcotic) effects, flammable, explosive gas/air mixtures, hazardous reactions (with oxidizers), water polluting
filled and modified epoxy resin
modified polyamine
Application
coating of synthetic resin containing solvents
corrosion protection/coating (steel)
bituminous emulsion (phenol-free, anionic)
corrosion protection/coating (steel, drinking water tanks)
naphtha (petroleum) 25-50%
irritations (skin, eyes, respiratory), neurological effects (coordination), flammable, water polluting
modified, filled anthracene oil and polyamine
corrosion protection/top layer
solvent naphtha 2,5-10%, biphenyl-2-ol 2,5-10%, 3-amino-3,5,5-trimethylcyclohexamine 1-2,5%, xylene 1-2,5%
irritations (skin, eyes, respiratory), flammable, water polluting
partially neutralized composition of aminoalcohols
sealants and adhesives/elastic products (floor joints, joints)
2-aminoethanol 2,5-10%
sensitization, irritations, slight water pollution
14.19.2 Recent advances in coalescing solvents
Product group/ compounds
Application
Hazardous ingredients, especially solvents
969
Relevant health and environmental risks
filled, reactive PUR-polymers
sealants and adhesives/elastic products (floor joints, joints)
3-isocyanatemethyl-3,5,5-trimethylcyclohexylisocyanate 0,1-1%, N,N-dibenzylidenepolyoxypropylene diamine 1-2,5%, xylene 2,5-10%, ethylbenzene 1-2,5%
solvent based composition, based on PVC
coatings and corrosion prevention/impregnation, sealer
solvent naphtha 50-100%, 1-methoxy-2-propanol 2,5-10%,
irritations, flammable, neurological effects
acrylate dispersion, water dispersed adhesion, promotor
coatings and corrosion protection/ rigid coat (for concrete and dense mineral substrates)
water-borne so-called solvent-free systems
irritations (long-term contact), slightly water polluting
sensitization, irritations, hazardous reactions (with amines, alcohols)
References 1 2 3
Ullmann´s Encyclopedia of Industrial Chemistry,1998. DIN 1045 (Deutsches Institut für Normung), Concrete and Reinforced Concrete. Beton und Stahlbeton, Beuth Verlag, Berlin (1988), part 1 (1997), part 2 (1999). Sika TechnoBauCD, Sika Chemie GmbH, Stuttgart (2000).
14.19.2 RECENT ADVANCES IN COALESCING SOLVENTS FOR WATERBORNE COATINGS
David Randall Chemoxy International pcl, Cleveland, United Kingdom
14.19.2.1 Introduction The role of coalescing solvents in coating formulations, the factors which affect the choice of coalescing solvents, and the recent developments in esters of low volatility, low odor and rapid biodegradability for use in coating formulations are discussed. The term “paint” is widely used to describe a coating applied to a variety of substrates for protective or decorative purposes. “Paint” also implies a pigmented species, whereas “surface coatings” is a broader term for coating systems with or without pigments used for any coating purpose. All paint systems may be considered as a combination of a small number of constituents. These are: continuous phase or vehicle (polymer and diluent), discontinuous phase or pigment and extender, and additives. This paper is not concerned with the detail of paint formulation, but the polymer types most widely encountered are alkyds, polyurethanes and nitrocellulose in solvent based systems, and acrylics, styrene-acrylics, and copolymers of vinyl acetate in water-based coatings.
970
David Randall
14.19.2.2 Water based coatings Until recent times, most paints were solvent based. Since many polymers were produced in solution, it was natural that the coating system, derived from these polymers, would use the solvent of reaction as the diluent for the polymer. These were almost invariably organic species. The objections to the use of organic solvents in paint formulations were at first confined to those with strong views regarding the loss of significant quantities of organic species to the environment. Latterly, these opinions have been reinforced by the role of some organic species in damage to the ozone layer, or to the production of smog in the lower atmosphere. This is as a result of the reaction between organic species present catalyzed by sunlight and exacerbated by the presence of other pollutants associated with motor vehicles, etc. Flammability of paints also poses a significant problem in storage and use. All this tended to reinforce the need to develop aqueous based systems, for which these problems would be eliminated. 14.19.2.3 Emulsion polymers Of course, the manufacture of polymers need not be carried out in solution; emulsion polymerization has a long and honorable tradition in the field of macromolecular chemistry. This polymerization technique is performed using water-insoluble monomers, which are caused to form an emulsion with the aqueous phase by the addition of a surfactant. Polymerization may be effected by the use of a variety of radical generating initiating species, and the molecular weight of the polymer is controlled by the use of chain transfer agents and the concentration of initiator employed. The final polymer dispersion is often described as a latex, named after natural rubber, which is also an emulsion polymer! This polymerization technique allows for the formation of copolymers in which the addition of relatively small quantities of comonomer may have a significant effect on the final properties of the polymer. This is particularly the case with the glass transition temperature, Tg, of the polymer. This is a physical transition, which occurs in the polymer when the amorphous structure of that polymer begins to change from a glassy to a rubbery state. At temperatures below a polymer’s Tg, it will be relatively brittle, and will be unlikely to form a coherent film. This effect has a major impact on the use of polymers in aqueous systems. In the case of a polymer in solution, the presence of the solvent plasticizes the polymer during film formation. A polymer with a high Tg, i.e., one greater than ambient temperature, can, when in solution be applied at temperatures below its Tg. In the case of an emulsion polymer, water is a non-solvent in the system and film formation below the polymer Tg is unlikely. The temperature at which a coherent film may be formed from a solution or emulsion based system is known as the Minimum Film Forming Temperature (MFFT or MFT). Figures 14.19.2.1-14.19.2.4 describe the formation of a discrete film from an emulsion system. When the film is applied onto the substrate, discrete polymer particles are dispersed in the aqueous phase. The system is stabilized by the presence of the surfactant at the water-particle interface. The particles are spherical, with an average diameter of 0.1-0.2 µm. As water is lost from the system, the mutual repulsive forces associated with the surfactants present inhibit the close packing of the particles and a cubic arrangement of the particles is formed.
14.19.2 Recent advances in coalescing solvents
Figure 14.19.2.1. Latex in contact with substrate.
Figure 14.19.2.2. Initial water evaporation and film formation.
Figure 14.19.2.3. Close packing of latex particles.
Figure 14.19.2.4. Coalescence to form a polymeric film.
971
As the water continues to evaporate, the particles become close packed with a solids volume of around 70%. Capillary forces continue to force the particles together. The final stage is achieved when most of the water is lost from the system. Here, the interparticular repulsive forces are overcome by increasing surface tension and the particles coalesce into a discrete film. This will only occur at temperatures in excess of the MFFT. 14.19.2.4 Role of a coalescing solvent Most coatings based on emulsion polymers are used in environments where they will be expected to form a coherent film at temperatures as low as 0°C. However, other physical properties besides film forming capability are required from the polymer. These include; abrasion resistance, hardness, chemical resistance, impact performance, etc. These can often be impossible to achieve with a polymer of low Tg. The polymers, which most clearly meet these criteria are acrylics or copolymers of vinyl acetate or styrene. These would, without additions of coalescing species, be brittle, forming incoherent films with little adhesion to the substrate at normal application temperatures. Coalescing solvents allow these polymeric systems to form films at ambient or sub-ambient temperatures. The presence of the coalescing solvent has the following effects. • It reduces the total surface energy of the system by reducing polymer surface area. • It increases the capillary forces by the controlled evaporation of the water. • It reduces the repulsive forces between the polymeric particles. • It allows deformation of the particles in contact with each other by effectively lowering the Tg of the polymer. An emulsion polymer consists of a dispersion of polymeric particles varying in size from 0.05-1 µm, dispersed in an aqueous environment. The coalescing solvent may be found in several different locations depending on their nature. They are classified according to their preferred location in the aqueous system.
972
David Randall
Hydrophobic substances such as hydrocarbons will prefer to be within the polymer particle; these are described as A Group coalescents. These tend to be inefficient coalescing agents. Molecules, which are more hydrophilic than hydrocarbons tend to be preferentially sited at the particle - aqueous interface, along with the surfactant system. These are described as AB Group coalescents. These exhibit the best efficiency as coalescing solvents. In the third group are the hydrophobic glycol ethers and similar species. These exhibit good coalescing power, but are partitioned between the polymer particle, the boundary layer, and the aqueous phase. More of these are required than AB Group coalescents, they are called ABC Group coalescents. Finally, hydrophilic species such as glycols and the more polar glycol ethers are inefficient coalescing agents, and are more commonly used as freeze thaw stabilizers. These are described as C Group coalescents. 14.19.2.5 Properties of coalescing agents 14.19.2.5.1 Hydrolytic stability A coalescing agent should have good hydrolytic stability (a high degree of resistance to hydrolysis) so that it can be used successfully in both low and high pH latex systems. 14.19.2.5.2 Water solubility It is desirable for a coalescing aid to have low water miscibility for the following reasons: • When added to a coating formulation, a coalescing aid with low water miscibility partitions into the polymer phase and softens the polymer; this improves pigment binding and polymer fusion. • There is less tendency for the evaporation of the coalescing aid to be accelerated by evaporation of the water during the early stages of drying. • The early water resistance of the coating is not adversely affected. • When a latex coating is applied to a porous substrate, water immiscible coalescing aids are not lost into the porous substrate along with the water; i.e., more coalescing aid will be available to coalesce the coating. 14.19.2.5.3 Freezing point The freezing point of a coalescing agent should be low (below -20°C) as materials with a high freezing point may require specialized (therefore more expensive) handling techniques in transport and storage. 14.19.2.5.4 Evaporation rate The evaporation rate of a coalescing aid should be slow enough to ensure good film formation of the emulsion coating under a wide range of humidity and temperature conditions; however, it should be fast enough to leave the coating film in a reasonable length of time and not cause excessive film softness. The evaporation rate of the coalescing aid should be less than that of water but not so slow that it remains in the film for an extended period of time causing dirt pickup. 14.19.2.5.5 Odor The odor of a coalescing agent should be minimal. This is especially important for interior coatings apFigure 14.19.2.5. Location of species in aqueplications. ous dispersions.
14.19.2 Recent advances in coalescing solvents
973
14.19.2.5.6 Color A coalescing agent should be colorless, to prevent discoloration of the product. 14.19.2.5.7 Coalescing efficiency The effectiveness of a coalescing agent is based upon the MFFT test. 14.19.2.5.8 Incorporation A good coalescing agent should be capable of addition during any stage of paint manufacture. Any particular coalescing agent may need premixing prior to addition with varying amounts of the water and surfactant used in the paint stage. 14.19.2.5.9 Improvement of physical properties A good coalescing agent can provide a significant improvement in scrub resistance, stain resistance, flexibility and weatherability. 14.19.2.5.10 Biodegradability A good coalescing agent should exhibit acceptably rapid biodegradation quickly. 14.19.2.5.11 Safety The use of the coating system requires that the nature of the coalescing agent should permit use in the domestic environment. The suitability of such a material is therefore obviously dependent on its toxicity, flash point, vapor pressure and VOC classification. 14.19.2.6 Comparison of coalescing solvents From the above, it may readily be seen that AB type coalescing solvents give the best performance in use. These are the high molecular weight esters and ester alcohols. Other species have some of the benefits of these materials but have side effects, which render them less suitable. Minimization of the quantity used tends to be a key financial issue, as well as offering the best environmental option. In virtually all cases, the efficiency of the AB group in application is so marked, that they have become the industry standards. Several other properties, perhaps less important, than efficiency must then be considered when selecting between these. These products are usually used in enclosed spaces, where product odor will be vital for acceptance to the consumer. The dibasic esters are virtually odor free. For legislative reasons, a solvent not classified as a VOC will be of definite benefit, given the current attitude to organics in consumer products. Again, the dibasic esters have initial boiling points considerably above the VOC threshold of 250°C. The stability of the product towards hydrolysis is also a significant factor in the selection of the optimum coalescing agent. Effectively, this parameter determines the storage stability of the product. Again, dibasic esters, particularly those produced from higher molecular weight alcohols appear to have advantage. The efficiency of the AB group of coalescents towards almost all polymer types has already been mentioned. For styrene-acrylics and vinyl acetate - Veova copolymers, the diesters have a significant edge in performance compared with other members of that group. This advantage is shown in the reduction in amount required to attain a particular MFFT or the actual MFFT for a given addition level. Coasol manufactured by Chemoxy is the blend of di-isobutyl esters of the dibasic acids; adipic, glutaric and succinic. These acids are produced as a by-product in the production of adipic acid in the manufacture of Nylon 6,6. The ratio of the acids is 15-25% adipic,
974
David Randall
50-60% glutaric, 20-30% succinic. Esterification using isobutanol is then performed and the product is isolated following distillation. Coasol was introduced in the early part of last decade as a coalescing agent to the aqueous based coatings industry. Coasol has excellent coalescing properties and is compatible with virtually all paint systems. It has very low vapor pressure at ambient temperature which ensures excellent plasticization throughout the film forming and drying process. Its extremely low odor also significantly improves the general view that aqueous systems have unpleasant odors. This is almost always as a result of the coalescing solvent. Its resistance to hydrolysis allows its use in a wide variety of formulations, including those, which require adjustment of pH to basic conditions. It rapidly biodegrades, both aerobically and anaerobically, since, as an ester, it is readily attacked by ubiquitous bio-species. Finally, its toxicological profile is virtually benign, so in use it can be handled with confidence in most working environments. Its properties are summarized in Table 14.19.2.1. Table 14.19.2.1. Properties of Coasol Physical Characteristic
Coasol
Comments
Boiling point
> 275°C
Not a VOC
Evaporation rate
< 0.01 (Butyl acetate = 1.0)
Slow, allowing excellent coalescence
Freezing point
< -55°C
No freeze-thaw issue
Water solubility
600 ppm
Hydrolytic stability
Very Good
No hydrolysis in normal use
Color
5 Hazen units
Imparts no color to coatings
Coalescing efficiency
Excellent for most polymer systems
Biodegradability
80% in 28 days
Odor
None discernible
Toxicology
Oral LD50 (Rat) >16,000 mg/kg
Biodegradable
Essentially non toxic
14.19.2.7 RECENT ADVANCES IN DIESTER COALESCING SOLVENTS We have also prepared the di-isopropyl esters of the higher adipic content stream. This has a vapor pressure similar to that of Coasol, but is slightly more water soluble. Finally, we have manufactured di-isopropyl adipate, which has the highest boiling point, the lowest vapor pressure and the lowest water solubility of all of this range of products. These preparations were undertaken to add to repertoire of products to suit the diverse requirements of the formulators of aqueous based systems. In virtually all cases, the dibasic esters gave a significant improvement in efficiency in reducing the MFFT for a given quantity of additive. The attempts are made to offer a tailor made solution to each individual polymer system employed in the development of aqueous based systems. The dibasic esters of the AGS acids group offer the opportunity for fine tuning, with the added advantage of low odor, low toxicity and “excellent” VOC status.
14.20 Petroleum refining industry
975
14.19.2.8 Appendix - Classification of coalescing solvents Coalescent Type Type A
Type of Species
Examples
Comments
Hydrocarbons
White Spirit
Type AB
Diesters
DBE Dimethyl esters DBE Diisobutyl esters Di-isobutyl adipate Di-isopropyl adipate Dibutyl phthalate
Estasol, Du Pont DBE’s Coasol, Lusolvan Chemoxy new products
Type AB
Ester alcohols
Diol Monoesters
Texanol
Type ABC
Glycol esters & Glycol ester ethers
PGDA Butyldiglycol acetate
Type ABC
Ether alcohols & diethers
PnBS 2-Butoxyethanol MPG Diethers
Type C
Glycols
DEG DPG TEG
Dow Products BASF and others Proglides and glymes
14.20 PETROLEUM REFINING INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The US petroleum refining industry generates sales of over $140 billion with only about 200 plants. It employs 75,000 people. About 90% of the products used in US are fuels of which 43% is gasoline. Figure 14.20.1 illustrates how the products breakdown. The process is described in detail in Chapter 3. Emissions of hydrocarbons to the atmosphere occur at almost every stage of the production process. Solvents are produced in various processes and they are also used to extract aromatics from lube oil feedstock, deasphalting of lubricating base stocks, sulfur recovery from gas stream, production of solvent additives for motor fuels such as methyl tert-butyl ether and tert-amyl methyl ether, and various cleaning operations. Emissions to atmosphere include fugitive emissions of the volatile components of crude oil and its fractions, emissions from incomplete combustion of fuel in heating system, and various refinery processes. Fugitive emissions arise from thousands of valves, pumps, tanks, pressure relief valves, flanges, etc. Individual leaks may be small but their combined quantity results in the petrochemical industry contributing the largest quantity of emissions and transfers. Tables 14.20.1 and 14.20.2 give solvent releases and transfers data for the petroleum refining industry. Transfers are small fraction of releases which means that most wastes are processed on-site. In addition to emissions to atmosphere, some plants have caused contamination of ground water by releasing cooling and process water.
976
George Wypych
Toluene, xylenes, and benzene constitute the majority of solvent emissions since they are native components of crude oil. Methyl ethyl ketone is also emitted in large quantities because of its use in lube oil dewaxing. Pollution prevention will become increasingly important to the petroleum industry as federal, state and municipal regulations become more stringent and waste disposal cost rises. The industry estimates that to comply with 1990 Clean Air Act Amendments it will require investment of $35-40 billion. Actions required to decrease pollution include process equipment modification, waste segregation and separation, recycling, and better training and supervision.
Figure 14.20.1. Diagram of production outputs from refineries. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995.
Table 14.20.1. Reported solvent releases from the petroleum refining industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
1,750,000
methyl isobutyl ketone
110,000
n-butyl alcohol
23,000
N-methyl-2-pyrrolidone
280,000
tert-butyl alcohol
28,000
methyl tert-butyl ketone
1,380,000
carbon tetrachloride
17,000
tetrachloroethylene
21,000
cresol
75,000
1,1,1-trichloroethane
50,000
cyclohexane
960,000
trichloroethylene
730
dichloromethane
8,000
1,2,4-trimethylbenzene
420,000
ethylbenzene
600,000
toluene
4,360,000
ethylene glycol
46,000
xylene
2,330,000
hexane
3,000,000
m-xylene
170,000
methanol
540,000
o-xylene
150,000
methyl ethyl ketone
2,100,000
p-xylene
1,000,000
14.21.1 Use of solvents in manufacture of drugs
977
Table 14.20.2. Reported solvent transfers from the petroleum refining industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
160,000
N-methyl-2-pyrrolidone
4,000
tert-butyl alcohol
900
methyl tert-butyl ketone
34,000
carbon tetrachloride
1,000
tetrachloroethylene
900
cresol
130,000
1,1,1-trichloroethane
6,500
cyclohexane
10,000
1,2,4-trimethylbenzene
31,000
ethylbenzene
61,000
toluene
270,000
ethylene glycol
58,000
xylene
340,000
hexane
13,000
m-xylene
11,000
methanol
180,000
o-xylene
30,000
methyl ethyl ketone
30,000
p-xylene
7,000
methyl isobutyl ketone
3,500
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.21 PHARMACEUTICAL INDUSTRY 14.21.1 USE OF SOLVENTS IN THE MANUFACTURE OF DRUG SUBSTANCES (DS) AND DRUG PRODUCTS (DP)
Michel Bauer International Analytical Department, Sanofi-Synthélabo, Toulouse, France
Christine Barthélémy Laboratoire de Pharmacie Galénique et Biopharmacie, Faculté des Sciences Pharmaceutiques et Biologiques, Université de Lille 2, Lille, France
14.21.1.1 Introduction Today the manufacturing of a pharmaceutical drug is almost totally the responsibility of: • the chemical industry for the preparation of the drug substance (active principle) and the excipients used for preparing the DP (finished product) • the pharmaceutical industry for the preparation of the DP itself. For the reader who is not familiar with the pharmaceutical industry, a reminder of the key points is given below.
978
Michel Bauer, Christine Barthélémy
One or several active compounds (DS) are prepared by organic synthesis, extracted from vegetable, animals, micro-organisms or obtained by biotechnology. The DS is generally associated in the product with several excipients of chemical, mineral or biological in nature either as monomers or as polymers. The goal is to formulate a processable DP stable over the time and allowing the active substance to be released in vitro and in vivo. Obviously the formulation is designed in relation to the route of administration: • oral route solid dosage forms (e.g., tablets, capsules, etc.) drinkable solutions etc. • ORL route (nasal solutions, spray) • local route (suppositories, transdermal systems, eye-drop formulation, spray) • intravenous and intramuscular route (injectable solution, lyophilizate, etc.) At practically every step of the manufacture of the drug substance and the excipients, solvents including water are utilized. This is equally true for the preparation of the pharmaceutical formulations. Ideally we would like to have available a universal stable solvent, ultrapure, non-toxic and which does not affect the solutes. This is an old dream of the alchemists who searched for a long time for the “Alkahest” or “Menstruum universal” as it was named by Paracelsus.1 The fact that a solvent is not a totally inert species allows it to play an important role in chemical equilibria, rates of chemical reactions, appearance of new crystalline forms, etc. and consequently contributes to the great wealth of compounds which the chemists are able to produce. But of course, there are drawbacks in using solvents. Because there are not totally inert they may favor the formation of undesirable impurities in the intermediates of synthesis and in the DS. Regarding the manufacture of the DP, the solvents, including water, may induce either polymorphic transformations or formation of solvates (hydrates) which, after drying, could lead to a desolvated solvate with quite different physical properties impacting potentially either positively or negatively on the DP performance.2 Another crucial aspect which deserves to be discussed, is the notion of purity. Impurities present in solvents could have an impact on the stability of drugs, for example, or on the crystallization process. Last, but not least, the toxicological aspects should be taken into account. Numerous solvents show different kinds of toxicity and this should be a matter of concern in relation to the health of workers exposed to them.3 But ultimately residual solvents still present in the DS and DP have to be assessed and systematically limited. We are now going to consider several aspects of the use of solvents in the manufacture of drug substances (DS) and drug products (DP) including their quality (purity) and influence on the quality, stability and physico-chemical characteristics of pharmaceutical products. The issue of residual solvents in pharmaceutical products will be considered in Chapter 16.2 and will focus amongst other things on the corresponding ICH Guideline.4
14.21.1 Use of solvents in manufacture of drugs
979
14.21.1.2 Where are solvents used in the manufacture of pharmaceutical drugs? 14.21.1.2.1 Intermediates of synthesis, DS and excipients 14.21.1.2.1.1 General points Raw materials are now produced by the chemical industry and involve the use of solvents at different steps in their production. These materials are usually produced by: • chemical synthesis • an extraction process, a fermentation process • or a biotechnology process The goal of this chapter is, of course, not to deal with the criteria for selection in relationship to their use in particular chemical reactions or extraction processes but rather to stress that impurities present in solvents could have an impact on the purity of the substances obtained, on their stability and potentially on their safety. These three concepts are of paramount importance in the pharmaceutical field. A list of solvents which are commonly used in the chemical industry5 is presented in Table 14.21.1.1. Table 14.21.1.1. Solvents commonly used in the chemical industry
Alcohols
Ketones
Ethanol
Acetone
Butanol
Methyl ethyl ketone
2-Ethylhexanol
Methyl isobutyl ketone
Isobutanol
Methyl isopropyl ketone
Isopropanol
Mesityl oxide
Ethylene bromide Chloroform Ethylene chloride Dichloromethane Tetrachloroethylene Carbon tetrachloride
Methanol Propanol
Ethers
Propyleneglycol
1.4-Dioxane Butyl ether
Amide
Ethyl ether
Dimethylformamide
Diisopropyl ether Tetrahydrofuran
Amine
Halogenated solvents
Tert-butyl methyl ether (MTBE)
Trichloroethylene
Sulphur-containing Dimethylsulfoxide
Aromatics hydrocarbons Toluene Xylene
Pyridine Nitriles Aliphatic hydrocarbons
Acetonitrile
Ethyl acetate
Cyclohexane Hexane
Esters
Water
It is generally relatively easy to know for pharmaceutical industry the nature of solvents to be looked for in a DS because it produces itself the active component or because it
980
Michel Bauer, Christine Barthélémy
can have by contract an access to the DS Mater File or because there is a compendial monograph giving occasionally some indications (e.g., search for benzene in carbomers). As a consequence of the ICH Guideline Q3C4 dealing with the residual solvents in pharmaceutical products (see Chapter 16.2), the use of solvents of class I (solvents to be avoided) like benzene is no more possible. It is known that carbopol resins (carbomer), used to modify the rheology of polar systems and as a binder in sustained release tablets, were up to now polymerized in benzenic medium. In the current quality of poloxamers it was possible to retrieve up to 1000 ppm of benzene. The ICH limit being 2 ppm, it was impossible to achieve this goal. The manufacturers have consequently developed new polymerization media6 containing either ethyl acetate alone or a mixture of ethyl acetate and cyclohexane, the first one belonging to class III (no safety concern), the second one belonging to class II (ICH limit 3880 ppm). 14.21.1.2.1.2 Criteria of purity This is a difficult matter. Purity in chemistry is an ideal concept referring to a situation where a product consists of one type of molecules only. This is a theoretical situation which can only be EXPERIMENTALLY approached more or less closely.7 The purity of a product is a relative notion and is dependant on the analytical methods used and their performances. More practically the quality finally chosen for a solvent will depend on the specific use for which this solvent is intended to be utilized.8 A solvent is considered sufficiently pure if it does not contain impurities able in nature and in quantity to interfere on the admissible quality of the product in the manufacture of which it participates.7 14.21.1.2.1.3 Solvents as reaction medium In this case the solvents should have a range between the melting point and the boiling point as extended as possible and a good thermal and chemical stability. The purity should be of good degree but could depend on the step considered of the global synthesis. As an example let us consider the case of the dimethylformamide (DMF).5 If it is used in reactions evolving in anhydrous media, it will be mandatory to control the level of water at the ppm level. The specification regarding the water content will be of course loosened if the DMF in the chemical step considered is used in conjunction with water as reaction media. We will see further that solvents contain actually a lot of chemical impurities which could be reactive vis a vis the main molecule undergoing the chemical reaction and leading to additional impurities other than those coming from the mechanism of reaction itself. 14.21.1.2.1.4 Solvents for crystallization They should be carefully chosen in such a way that they show a high solubility at high temperature and a low solubility at low temperature of the substance to be crystallized or recrystallized. Of course the solvent should be absolutely inert and of the highest achievable purity for at least two reasons: • The first one being identical to the one mentioned for solvents as reaction media: possibility to produce other impurities. • The second one being linked to the crystallization process itself. It is well known that the presence of impurities whatever the origin could have serious effects on the nucleation and growth process. We will tell a little bit more about that further in the text.
14.21.1 Use of solvents in manufacture of drugs
981
14.21.1.2.1.5 Solvents used for extraction and preparative chromatography As in the precedent cases they have to be absolutely inert (as far as it is possible) and with a high degree of purity for the reasons already evoked. In case of preparative chromatography a special care will be taken concerning the chemical inertia to adsorbate9 of the solvents constituting the mobile phase and the fact that impurities or additives contained in the solvents in a way not under control could impair significantly the reproducibility of the retention times. 14.21.1.2.1.6 Nature and origin of impurities contained in solvents10 It should be reminded here that a solvent used at the industrial level is rarely pure (we mean here no impurity analytically detectable). Industrial solvents may contain: • impurities coming from their origin or their manufacturing process • impurities originating from the container during transportation • stabilizers • denaturing agents • impurities resulting from a transformation of the solvent during the chemical reaction These impurities or side products should be look for as far as it is possible when assessing the purity of the solvent. In fact they could be less volatile than the main solvent and could finally concentrate in the pharmaceutical product. We will now review shortly the nature of all these kinds of impurities of the most often used solvents. 14.21.1.2.1.6.1 Impurities coming from the origin or the manufacturing process of the solvent1,10 Table 14.21.1.2. Solvent impurities Class of solvents
Possible impurities (according to the manufacturing process) Hydrocarbons
Toluene
Methylthiophene, benzene, paraffinic hydrocarbons
Xylene
Mixture of ortho, meta and para isomers, paraffinic hydrocarbons, ethyl benzene, sulfur compounds
Cyclohexane
Benzene, paraffinic hydrocarbons, carbonyl compounds Halogenated compounds
Dichloromethane
Chloroform, carbon tetrachloride, chloromethane
Chloroform
Chlorine, carbonyl chloride (phosgene), dichloromethane, carbon tetrachloride, hydrogen chloride
Carbon tetrachloride
Chlorides, chlorine, carbon disulfide Alcohols
Methanol
Water, acetone, formaldehyde, ethanol, dimethylether, carbon dioxide, ammonia
methyl
formate,
982
Michel Bauer, Christine Barthélémy
Ethanol
Aldehydes, ketones, esters, water, ethyl ether, benzene (if anhydrous ethanol)
2-propanol
Water, peroxides
N.B.: Some alcohols obtained by fermentation could contain pesticides. It is necessary to obtain from the purchaser some guaranty in requiring limit contents (expressed in Parathion e.g.). Aliphatic ethers/cyclic ethers Ethylether/isopropyl ether/monoalkylated ethers/ethylene glycol/diethylene glycol/etc.
Alcohols (from which they are prepared), water, corresponding aldehydes, peroxides
Tetrahydrofuran
Water, peroxides
Dioxane
Acetaldehyde, water, acetic acid, glycol acetal paraldehyde, crotonaldehyde/peroxides Ketones
Acetone
Methanol, acetic acid, water Esters
Methyl acetate
Acetic acid, water, methanol
Ethyl acetate
Acetic acid, ethanol, water Amides
Formamide
Formic acid, ammonium formate, water
N,N-Dimethylformamide
N-Methylformamide, formic acid, water Nitriles
Acetonitrile
Acetamide, ammonium acetate, ammoniac, water, toluene Nitro compounds
Nitrobenzene
Nitrotoluene, dinitrothiophene, dinitrobenzene, aniline
14.21.1.2.1.6.2 Impurities originating from the container during transportation It relates to contamination coming from tankers or drums not correctly cleaned. These concerns of course solvents of low quality conveyed in industrial quantity. In case of utilization of such solvents, the user has to bear in mind that some incidents or uncommon behavior may find an explanation based on this considerations. 14.21.1.2.1.6.3 Stabilizers It is of course very difficult to know every stabilizer used. There is here an important problem of confidentiality. We quote thereafter some of them which are well known.
14.21.1 Use of solvents in manufacture of drugs
983
Table 14.21.1.3 Stabilizers used in selected solvents Solvents
Stabilizers
Dichloromethane
Ethanol, 2-methyl-but-2-ene
Chloroform
Ethanol (1% V/V) for avoiding the phosgene formation, 2-methyl-but-2-ene
Diethylether
2,6-di-tert-butyl-4 methylphenol (BHT)
Tetrahydrofuran
BHT, p-cresol, hydroquinone, calcium hydride
14.21.1.2.1.6.4 Denaturing agents This process is relevant primarily to ethanol. Common denaturing agents are: methanol, isopropanol, ethyl acetate, toluene. 14.21.1.2.1.6.5 Transformation of the solvent during the chemical reaction Solvents are rarely chemically inert. During the reactions where solvents are involved, they can undergo chemical transformation generating impurities which can be found, for example in the DS. This is a huge field which cannot be exhaustively covered. We give below a few examples of well-known side reactions. • Acetone in acidic media is easily transformed into mesityl oxide:
So do not forget to test for it when performing residual solvents analysis on drugs. • In basic medium the diketone-alcohol is obtained:
• DMF can be hydrolyzed in presence of hydrochloric acid:
• Acids undergoing reaction in alcoholic media can be partially transformed into esters
• Transesterification reaction. Take care when, for example, recrystallization has to be performed for a molecule containing an ester group:
• Aldehydes (even ketones) can be transformed in alcoholic solutions into ketals:
984
Michel Bauer, Christine Barthélémy
• From time to time the solvent can react in lieu of the reagent. In a synthesis aimed to prepare 3-chloro-1-methoxy-2-propan-2-ol starting from chloromethyloxirane and methanol, traces of ethanol present in methanol gave the corresponding ethoxylated compound:
Another and final example concerns the preparation of a urea derivative using the reaction of an amine and isocyanate in presence of isopropanol:
Other examples could be found. These obvious examples stress the need for close collaboration between chemists and analysts when elaborating chemical syntheses and corresponding quality control monographs. 14.21.1.2.2 Drug products11,12 14.21.1.2.2.1 General points Because ultimately it is the DP which is administered to the patient, it is necessary to have the quality of the solvents potentially used in the design of pharmaceutical formulations under control. 14.21.1.2.2.2 Areas of utilization Solvents including water are used in different ways in pharmaceutical formulation: • either as a part of the final drug product: injectables, drinkable solutions, patches, sprays, microemulsions • or used as an intermediary vehicle which is removed at the end of the process: granulation coating sugar coating microencapsulation
14.21.1 Use of solvents in manufacture of drugs
985
We have listed in Table 14.21.1.4. the most commonly used solvents. Table 14.21.1.4 Solvents used in formulation Water
Dichloromethane
Ethyl acetate
Chloroform
Ethyl alcohol (denaturated with butanol and isopropanol)
Hexane
Isopropyl alcohol (denaturated with methyl ethyl ketone)
Cyclohexane
Methanol
Polyethylene glycol (low molecular weight)
Acetone
Manufacturers try progressively to replace the formulations using organic solvents such as chloroform, dichloromethane, cyclohexane belonging to the class 2 (ICH classification see Chapter 16.2), for example by developing aqueous coatings. 14.21.1.2.2.3 What should be the quality? Taking into account the fact that the solvents used in the DP manufacturing process, either as a component of the formulation or as a residual solvent, will be absorbed by the patient, their quality must be of the highest standard. From a regulatory point of view, in almost every country if not all, it is mandatory to use solvents covered by a pharmacopoeial monograph (e.g., European Pharmacopoeia, USP, JP, local Pharmacopoeias). Some examples are given below. 14.21.1.3 Impacts of the nature of solvents and their quality on the physicochemical characteristics of raw materials and DP. 14.21.1.3.1 Raw materials (intermediates, DS, excipients) The impurities contained in the solvents could have several effects on the raw materials: • When the solvents are removed, non-volatile or less volatile impurities will be concentrated in raw materials. • They can induce chemical reactions leading to side products. • They can affect the stability of the raw material considered. • They can modify substantially the crystallization process. 14.21.1.3.1.1 Concentration of less volatile impurities Due to the potential concentration of these impurities, they should be tested for in both DS and excipients and even in intermediates of synthesis if the latter constitute the penultimate step of the synthesis and if solvents belong to class 1 or class 2 solvent (see Chapter 16.2). 14.21.1.3.1.2 Side reactions This case has already been illustrated (see paragraph 14.21.1.2.1.6.5). The skills of the chemist together with those of the analyst are needed to ensure that the presence of unexpected impurities can be detected. By way of example, the reactions involving the keto-enol tautomerism deserve to be mentioned. The equilibrium is very sensitive to the solvent so that the presence of other solvents as impurities in the main solvent can modify the keto-enol ratio leading to irreproducibility in the chemical process.13,14 14.21.1.3.1.3 Consequences for stability Some solvents, as mentioned in the paragraph 14.21.1.2.1.5, can contain very active entities such as aldehydes and peroxides. For example, if the raw material contains primary or sec-
986
Michel Bauer, Christine Barthélémy
ondary amines and/or is susceptible to oxidation or hydrolysis, it is likely that degradants will be formed over the time, reducing potentially the retest date of the raw material. The same situation could affect the DP and therefore two examples are given in paragraph 14.21.1.3.2. 14.21.1.3.1.4 Solvent purity and crystallization This issue is may be less known except for chemists working in this specialized area. Because the consequences can be important for the processability, stability and occasionally bioavailability of the drug substance in its formulation, it is relevant to comment on this subject. 14.21.1.3.1.4.1 Role of the nature and the quality of solvent on crystallization Most of the drugs on the market are obtained as a defined crystalline structure and formulated as solid dosage forms. It is well known that a molecule can crystallize to give different crystalline structures displaying what is called polymorphism. The crystal structures may be anhydrous or may contain a stoichiometric number of solvent molecules leading to the formation of solvates (hydrates in case of water molecules). Pseudopolymorphism is the term used to describe this phenomenon. Another characteristic which plays a major role in the overall processability of the DS for DP manufacture is the “crystal habit”. This term is used15,16 to describe the overall shape of crystals, in other words, the differing external appearance of solid particles which have the same internal crystalline structure. Both structures (internal, external) are under the control of different parameters including the nature of the solvent used and its quality. The role of the solvent itself in the overall crystallization process, including the determination of the crystal structure and the crystal habit is well known.17 But it is equally worth noting that impurities coming from: • the product to be crystallized • the solvent used • the environment can selectively affect the nucleation process and the growth rates of different crystal faces.17-21 They can be selectively adsorbed to certain faces of the polymorphs thereby inhibiting their nucleation or retarding their growth to the advantage of others. Crystal shape (habit) can also be modified by a solvent without polymorphic change. Additives or impurities can block, for a defined polymorph, the growth rate of certain faces leading e.g. to needles or plates. It is possible to introduce deliberately additives to “steer” the crystallization process. An interesting example of this crystal engineering strategy have been published for e.g., adipic acid22 or acetaminophen.23 14.21.1.3.1.4.2 Solvent-solid association/overview After the crystallization of the product, solvents must be removed in order to obtain the minimum amount of residual solvents compatible with safety considerations and/or physicochemical considerations including stability, processability and occasionally microbiological quality (see Chapter 16.2). Different situations can be encountered. 14.21.1.3.1.4.2.1 Solvent outside the crystal The solvent remains outside the crystals at the time of crystal formation. It is adsorbed on the surface or in the crystal planes. In the first case, the solvent is easily removed. But in the second case, if a cleavage plane exists, the drying process can be very difficult. Two methods can be used to try to remove this type of residual solvent almost completely.
14.21.1 Use of solvents in manufacture of drugs
987
• Displacement by water vapor in an oven, keeping in mind that this method may introduce some degradation leading to processability problems.11,24
• Extraction by supercritical CO225,26 but the extraction power of CO2 is basically limited to slightly polar solvents. 14.21.1.3.1.4.2.2 Solvent inside the crystal The solvent remains inside the crystalline structure. Three situations can arise. 14.21.1.3.1.4.2.2.1 Occluded solvents During rapid crystallization, some degree of disorder (amorphous phases, crystalline defects) can arise, creating pockets where residual solvent can be occluded. Through a process of dissolution/recrystallization this “hole” moves towards the external faces of the crystal releasing the solvent at the end. This phenomenon is more frequent for large crystals (500 µm/600 µm) but rare for smaller crystals (1 - 100 µm). The solvent odor which is detected when opening a drum or a bag containing a substance which was dried in the normal way can be explained by this mechanism. 14.21.1.3.1.4.2.2.2 Solvates At the end of the crystallization process, the substance can be isolated as a solvate (hydrate), i.e., as a pseudopolymorph. The solvates generally have quite different physicochemical properties from the anhydrous form. Their stability can be questionable and in any case deserves to be investigated. In some cases it is possible to remove the solvent from the crystal without changing the structure of the lattice leading to an isomorphic desolvate which displays a similar X-ray diffraction pattern to that of the parent compound.2,27 The lattice of the desolvated solvate is in a high energy state relative to the original solvate structure. A better dissolution rate and compressability can be expected,28 but the drawbacks are hygroscopicity and physico-chemical instability. The lattice could undergo a relaxation process over time which increases the packing efficiency of the substance by reducing the unit cell volume. When developing a new chemical entity all these aspects have to be considered to avoid unpleasant surprises during development or once the drug is on the market. Due to the need for process scale-up and of making the manufacturing process more industrial, changes are introduced especially in the crystallization and the drying processes, (e.g., change from static drying to dynamic drying). Because the drying is a particularly disturbing process for the integrity of the lattice, defects and/or amorphous phases may be created favoring subsequent polymorphic or pseudopolymorphic transformations of the crystalline form developed so far, if it is not the most stable one. 14.21.1.3.1.4.2.2.3 Clathrates In contrast to solvates, clathrates do not show any stoichiometric relationship between the number of molecules of the substance and the number of molecules of solvent. Clathrates actually correspond to a physical capture of solvent molecules inside the crystal lattice without any strong bonds including hydrogen bonds. Molecules of one or several solvents can be trapped within the crystalline structure as long as the crystallization has been performed with a pure solvent or a mixture. The case of the sodium salt of warfarin giving “mixed” clathrates with water and isopropyl alcohol is well known and the existence of 8/4/0 or 8/2/2 proportions has been shown.11 It is fairly obvious that some powder properties like wettability can be modified by the formation of clathrates. Because their formation is not easy to control, some batch to batch inconsistency may be expected in this situation.
988
Michel Bauer, Christine Barthélémy
Figure 14.21.1.1. DSC and TG patterns.
In order to demonstrate the internal character of the clathrate relative to the crystal lattice, the example of a molecule developed in the laboratory of one of the authors is given below. Figure 14.21.1.1 shows the DSC pattern of a molecule with the melting event at 254°C and the corresponding TG pattern obtained at the same temperature scanning rate. At the time the melting occurs, a loss of weight is observed corresponding to the loss of 0.2 % of isopropanol. The nature and the amount of the solvent have been confirmed by GC after dissolving the substance. 14.21.1.3.2 Drug product As for the DS, the solvents used for DP manufacture can produce some negative effects by themselves or through their own impurities. For liquid or semi liquid formulations, the formulator has to ensure that the solvents themselves do not display chemical interactions with the DS or the excipients. Everything which has been said in paragraphs 14.21.1.2.1.5 and 14.21.1.3.1 remains true here. 14.21.1.3.2.1 Interaction of impurities contained in the solvent As said in paragraph 14.21.1.3.1 with the DS, impurities contained in the solvent especially if they are strongly reactive, like aldehydes or peroxides, can promote formation of degradants. Regarding aldehydes, the publication of Bindra and all29 should be mentioned. It relates to the degradation of the o-benzylguanine in an aqueous solution containing polyethylene glycol 400 (PEG 400). This type of solvent very often contains formaldehyde, which can lead to the formation of a precipitate over time:
14.21.1 Use of solvents in manufacture of drugs
989
PEG can also contain peroxides which can initiate over time, the formation of degradants via an oxidation process. Several publications have dealt with this phenomenon.30,31 14.21.1.3.2.2 Interaction with the container When the formulation is a solution which is prepared from water and different organic solvents, it is mandatory to investigate possible interactions between the medium and the container especially if the latter is of polymeric nature (PVC-PVDC, polyethylene, etc.) with or without elastomeric stoppers. A thorough investigation is necessary including: • an examination of the solution for plasticizers, antioxidants, monomers and oligomers, mineral impurities, potentially extracted from the container, • the evaluation of the absorption by the container of components (DS, excipients) contained in the solution. In the first case, the migration of impurities into the solution could initiate physicochemical instability and possibly some potential toxicity. In the second case, a decrease in the content of the DS and/or some excipient (e.g., organic solvents added to promote the solubility) could lead to some loss of therapeutic efficacy and in some case to physical instability (precipitation). 14.21.1.3.2.3 Solvates formation during the solid dosage form manufacture During the granulation process it is possible that the DS (occasionally the excipient) could transform into a solvated crystalline structure (solvate, hydrate). During the drying process, different situations can occur: • The solvate is poorly stable and the solvent is easily removed leading to either the original polymorphic form but creating a certain degree of disorder in the crystalline structure or to what is called a “desolvate solvate” form. In this last case, also named “isomorphic desolvate”, the desolvated solvate retains the structure of its parent solvated form. The X-ray diffraction patterns look similar between the parent and the daughter forms. In this situation we have the creation of a molecular vacuum which could substantially impact on the stability, hygroscopicity and mechanical characteristics of the DS and finally of the DP. • The solvate is stable within the formulation: we then have in a sense a new chemical entity. The properties of the solvate could be entirely different (solubility, kinetics of dissolution, stability, processability, etc.) and the consequence could be either positive or negative. The case where the kinetics of dissolution are affected by the formation of solvates should always be investigated. Papers on this subject have been published for molecules such as lorazepam,32 hydrocortisone,33 cephalexin,32 etc. • Obviously, as in the case of raw materials (14.21.1.3.1.4.2.2.3) clathrate formation should be considered in order to explain possible batch to batch inconsistency. 14.21.1.3.3 Conclusions We have seen that the solvent, far from being inert, plays a key role by itself and occasionally via its own impurities in different ways which are important for pharmaceutical development. We will now discuss how to set up sound specifications for solvents in relation to their field of use.
990
Michel Bauer, Christine Barthélémy
14.21.1.4 Setting specifications for solvents 14.21.1.4.1 Solvents used for the raw material manufacture For the raw materials we should distinguish between solvents: • used during the synthesis • and those used for the last step of the manufacture corresponding very often to the crystallization process. As a rule of thumb, the specifications set Table 14.21.1.5. Examples of for solvents used for the crystallization step will monographs be more stringent than those used during the synthesis. Solvents used during the synthesis (of well For the intermediates of synthesis, if the controlled origin) origin of the solvent is under control (e.g., exisCharacter/Appearance tence of contracts/Quality Assurance audits) a ) Identification (IR, GC or n 20 simplified monograph is completely adequate D (see Table 14.21.1.5) as long as the supplier proPurity GC (generally not less than 98%) vides a detailed certificate of analysis where impurities (including solvents) are properly Table 14.21.1.6. Examples of specified with acceptable limits. If the same solmonographs vent is used for the crystallization step additional purity tests are necessary (Table 14.21.1.6). Final crystallization solvents (of well conFor economic reasons, it may be necessary trolled origin) to recycle solvents. If so, the containers should Character/Appearance be fully identified in terms of storage: Identification (IR, GC or n 20 • If solvents can be efficiently purified (e.g., by D) redistillation) they must comply with the Tests same specifications as those of fresh solvents Water content (0.1 to 0.5 depending on and consequently can be used in any solvent type) synthesis. Residue on evaporation: not more than • If they still contain volatile impurities 0.01 per cent resulting from the reaction they come from, Purity GC not less than 99 % (examine for they can be recycled only for this reaction. In denaturing agents and other potential imthis case, the impurities should be identified purities) and their possible impact on the reaction evaluated. In Tables 14.21.1.7 and 14.21.1.8 we have summarized possible specifications for a fresh batch of ethyl acetate used for a defined chemical reaction and those for the recycled solvent. We recommend working with reliable solvent suppliers who can give every assurance on the quality of solvents provided to avoid any “unpleasant surprises”. Water should be mentioned separately. If it is used during the synthesis of intermediates the quality “drinking water” can be used without any problems. But if water is used during the last step of the process, its quality must be in compliance with the requirements of purified water as they are described in several pharmacopoeias. In Table 14.21.1.9 the requirements for the Ph. Eur and USP are given as examples. Purified water is generally obtained from the drinking water. It undergoes demineralization by either distillation or an ion-exchange process. Particular attention has to be paid to microbiological quality.
14.21.1 Use of solvents in manufacture of drugs
991
Table 14.21.1.7. Example of monograph Table 14.21.1.8. Example of monograph applied to recycled solvent: Ethyl acetate applied to new solvent: Ethyl acetate
Specifications
Specifications Controls Characters
Clear liquid, colorless
Complies
1.370 to 1.373
or C - Gas chromatography
Gas chromatography
Complies
Water content
Not more than 2.0%
Related substances Complies
Assay Ethyl acetate (purity)
Clear liquid, colorless
Tests
or B - Refractive index
Characters
Standards
Identification
Identification A - Infrared spectrum
Controls
Standards
Not less than 99.5%
Methanol
Not more than 1.0%
Ethanol
Not more than 2.0%
Ethyl chloride
Not more than 2.0%
Others impurities (sum)
Not more than 2.0%
Assay Ethyl acetate (purity)
Not less than 99.5%
14.21.1.4.2 Solvents used for the DP manufacture There is no other choice than to use the quality of solvents defined by a Pharmacopoeia. It is true that there are still discrepancies between the pharmacopoeias of different countries. It is hoped that the ICH process dealing with the harmonization of quality, safety and efficacy amongst three main zones of the world (EU, USA, Japan) will progressively reduce the remaining differences in dossiers submitted to Regulatory Authorities and the way the data are evaluated. As examples, Tables 14.21.1.10, 14.21.1.11, and 14.21.1.12 summarize specifications for acetone, ethanol and isopropanol given by the Ph. Eur. and USP. As can be seen only the Ph. Eur. monograph makes reference to volatile impurities to be tested for by GC. 14.21.1.5 Quality of solvents and analysis The solvents, including water, are used in almost every area of analytical sciences: spectroscopy, chromatography, potentiometry, electrochemistry. They should be characterized by a set of properties making them suitable for use for their intended purpose. 14.21.1.5.1 Quality of solvents used in spectroscopy As a general requirement, the solvents used in spectroscopy should be transparent and stable towards the relevant range of wavelengths. They should be able to dissolve the substance to be examined and not contain impurities affecting the stability of the substance or the validity of the method (selectivity, repeatability, limit of detection, analytical response). Theoretically the solvent chosen should have minimal interaction with the solute. But what could be seen as a disadvantage could also be an important source of structural information. What is called the solvent effect can help in UV, IR and NMR spectroscopies34 e.g. in struc-
992
Michel Bauer, Christine Barthélémy
Table 14.21.1.9. Water quality
ture elucidation. In the book by Reichardt,1 data regarding the cut-off points of solvents commonly used in UV/visible spectroscopy are provided. The cut-off point is defined as the wavelength in the ultraviolet region at which the absorbance approaches 1.0 using a 1-cm cell path with water as the reference. In the same way the range of transparency for IR-solvents are given. Complete IR-spectra of organic solvents can be found in the “Stadler IR spectra handbook of common organic solvents” . The solvents suppliers usually provide catalogues including a “spectroscopic grade” allowing the user to make a sound choice. In the case of the NMR spectroscopy problems arise with the residual protonated part of deuterated solvents (1H-NMR) and the 13C-NMR absorption bands of compounds used as solvents. References can be found1 where detailed data are given regarding these points.
14.21.1 Use of solvents in manufacture of drugs
993
Table 14.21.1.10. Specification for acetone
A common problem which can be met regularly in chemistry is the identification of signals from common contaminants in solvents of medium quality (see paragraph 14.21.1.2.1.6.1). Gottlieb and Col35 have published data on NMR chemical shifts of trace impurities contained in common laboratory solvents, making NMR spectroscopy the instrument of choice as a tool for routine quality control. 14.21.1.5.2 Quality of solvents used in chromatography The aim of this paragraph is not to focus on strategies for solvent selection in order to achieve extraction or liquid chromatography. A detailed literature review has been published by Barwick36 on this matter allowing the user to design relevant methodology in any
994
Michel Bauer, Christine Barthélémy
Table 14.21.1.11. Specification for ethanol
Impurities as per European Pharmacopoeia, Supplement 2000: A - 1,1-diethoxyrethane (acetal); B - Acetaldehyde; C - Acetone; D - Benzene; E - Cyclohexane; F - Methanol; G Butan-2-one (methyl ethyl ketone); H - 4-methylpentan-2-one (methyl isobutyl ketone); I - Propanol; J Propan-2-ol; K - Butanol; L - Butan-2-ol; M - 2-methylpropanol (isobutanol); N - Furan-2-carbaldehyde (furfural); O - 2-methylpropan-2-ol (1,1-dimethyl alcohol); P - 2-methylbutan-2-ol; Q - Pentan-2-ol; R - Pentanol; S - Hexanol; T - Heptan-2-ol; U - Hexan-2-ol; V - Hexan-3-ol
14.21.1 Use of solvents in manufacture of drugs
995
Table 14.21.1.12. Specification for isopropanol
particular case. It is preferable to focus on the impact of impurities present in the solvent on the chromatographic performances. Impurities in and additives to solvents can cause several problems and artifacts in liquid and gas chromatography.37-39 Primarily they can be the origin of irreproducible separations, enhanced UV-background and even of mechanical problems. De Schutter and Col37 have investigated this problem in purity of solvents used in high-performance thin-layer chromatography. A way to improve the stability of the chromatographic system by minimizing the role of solvent impurities is to add deliberately a controlled amount of organic modifier. For example, Lauren and Col40 have applied this technique, using decanol for improving the stability of the LC used for analyzing carotenoids. Middleditch and Zlatkis41 have listed an impressive range of stabilizers and additives which can be found in solvents which may help the chromatographer in explaining the occurrence of artifacts in chromatography. Zelvensky and Col38 have determined by gas chromatography the most common impurities contained in solvents for liquid chromatography. The series of solvents investigated include acetonitrile, methanol, ethanol, dichloromethane, formic acid, dimethylformamide,
996
Michel Bauer, Christine Barthélémy
pyridine, tetrahydrofuran and dimethyl sulphoxide. Parsons and Col39 have performed a search for trace impurities in solvents commonly used for gas chromatographic analysis of environmental samples. 14.21.1.5.3 Quality of solvents used in titrimetry For titrimetric determinations performed in an aqueous medium it is highly recommended to use distilled water and to perform a blank determination if necessary. For determinations performed in non aqueous media, the solvents should be as anhydrous as possible and, of course, inert to the titrant and the substance. Their purity should be such that they do not contain impurities which could react with the substance to be analyzed. A blank titration should be performed if necessary. 14.21.1.6 Conclusions Far from being inert and not affecting the molecules dissolved in it, the solvent can affect the behavior of the solute in different ways. This chapter has aimed to support the idea that it is important for the chemist and the pharmacist to control the quality of the solvents used in the different areas of pharmaceutical activity. As we have tried to show, many pitfalls can be avoided during the development of a drug if a thorough investigation of the quality of the solvents used is carried out. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
C. Reichardt, Solvents and solvents effect in organic chemistry, 2nd edition, 1998, VCH Verlagsgesellschaft mbH. G.A. Stephenson, E.G. Groleau, R.L. Kleemann, W. Xu and D.R. Rigsbee, J. Pharm. Sci., 87, 536 (1998). A. Picot, Le bon usage des solvants, Information Toxicologique n° 3, Unité de prévention du risque chimique - CNRS (France) (1995). ICH Harmonised Tripartite Guideline, Impurities: Guideline for Residual Solvents. M. Gachon, STP Pharma Pratiques 1, 531 (1991). Carbomer Monograph - Eur. Ph. Addendum 2000, p 494. J.A. Riddik and W.B. Bungler, Organic Solvents, Wiley Intersciences, New-York, 1970, pp 552-571. G.P.J. Ravissot, STP Pharma Pratiques, 9, 3 (1999). See Reference 1 p 427. M. Debaert, STP Pharma Pratiques, 1, 253 (1991). A.M. Guyot-Hermann, STP Pharma Pratiques, 1, 258 (1991). D. Chulia, M. Deleuil, Y. Pourcelot, Powder Technology and Pharmaceutical Processes, Elsevier, Amsterdam, 1994. J.L. Burdett and M.T. Rodgers, J. Am. Chem. Soc., 86, 2105 (1964). M.T. Rodgers and J.L. Burdett, Can. J. Chem., 43, 1516 (1965). J. Haleblian and W. McCrone, J. Pharm. Sci., 58, 911 (1969). P. York, Int. J. Pharm., 14, 1 (1983). S. Khoshkhoo and J. Anwar, J. Phys. D.; Appl. Phys., 26, B90 (1993). J. Shyh-Ming, Diss. Abst. Int., 57 (10), 6402-B (1997). W. Beckmann and W.H. OTTO, Chem. E. Res. Des., 74, 750 (1996). N. Rodriguez-Hornedo and D. Murphy, J. Pharm. Sci., 88, 651 (1999). R. David and D. Giron in Powder Technology and Pharmaceutical Processes, Handbook of Powder Technology, Elsevier Sciences, Ed., Amsterdam 1994, pp 193-241. A.S. Myerson, S.M. Jang, J. Cryst. Growth, 156, 459 (1995). A.H.L. Chow, D.J.W. Grant, Int. J. Pharm., 42, 123 (1988). C. Lefebvre-Ringard, A.M. Guyot-Hermann, R. Bouché et J. Ringard, STP Pharma Pratiques, 6, 228 (1990). D.C. Messer, L.T. Taylor, W.N. Moore and W.E. Weiser, Ther. Drug Monit., 15, 581 (1993). M. Perrut, Information Chimie, 321, 166 (1990). R.R. Pfeiffer, K.S. Yang and M.A. Tucker, J. Pharm. Sci., 59, 1809 (1970). R. Hüttenrauch, Pharm. Ind., 45, 435 (1983). D.S. Bindra, T.D. William and V.J. Stella, Pharm. Res., 11, 1060 (1994).
14.21.2 Predicting cosolvency
30 31 32 33 34 35 36 37 38 39 40 41
997
J.W. McGinity and J.A. Hill, J. Pharm. Sci., 64, 356 (1975). D.M. Johnson and W.F. Taylor, J. Pharm. Sci., 73, 1414 (1984). J. Joachim, D.D. Opota, G. Joachim, J.P. Reynier, P. Monges and L. Maury, STP Pharma Sci., 5, 486 (1995). S.R. Byrn, C.T. Lin, P. Perrier, G.C. Clay and P.A. Sutton, J. Org. Chem., 47, 2978 (1982). J. Wiemann, Y. Pascal, J. Chuche, Relations entre la structure et les propriétés physiques, Masson edit, Paris, 1965. H.E. Gottlieb, V. Kotlyar and A. Nudelman, J. Org. Chem., 62, 7512 (1997). V.J. Barwick, Trends in An. Chem., 16, 293 (1997). J.A. de Schutter, G. Van der Weken, W. Van den Bossche and P. de Moerloose, Chromatographia, 20, 739 (1985). V.Y. Zelvensky, A.S. Lavrenova, S.I. Samolyuk, L.V. Borodai and G.A. Egorenko, J. Chromatogr., 364, 305 (1986). W.D. Bowers, M.L. Parsons, R.E. Clement, G.A. Eiceman and F.W. Rarasek, J. Chromatogr., 206, 279 (1981). D.R. Lauren, M.P. Agnew, D.E. McNaughton, J. Liq. Chromatogr., 9, 1997 (1986). B.S. Middleditch, A. Zlatkis, J. Chromatogr. Sci., 25, 547 (1987).
14.21.2 PREDICTING COSOLVENCY FOR PHARMACEUTICAL AND ENVIRONMENTAL APPLICATIONS
An Li School of Public Health, University of Illinois at Chicago, Chicago, IL, USA
14.21.2.1 Introduction Cosolvency refers to the effects of adding one or more solvents (cosolvents), which are different from the existing solvent in a solution, on the properties of the solution or behavior of the solute. Cosolvency has found its applications in numerous engineering and scientific disciplines. The discussion in this section will be limited to aqueous phase cosolvency (the primary solvent is water), and cosolvents will include only pure organic solvents which are miscible with water either completely (in any proportion) or partially (in only certain proportions). The extent of cosolvency will be quantitatively described by the difference in solute solubilities in pure water and in a mixture of water and cosolvent(s). Cosolvency has been studied for decades. However, it remains a poorly understood phenomenon due in large measure to our limited awareness of the liquid structure and the intermolecular forces. At present, practical approaches to predicting cosolvency are to develop models based on established theories and to make use of correlation between experimental observations and properties of the substances involved. As with all modeling efforts, it is essential to make judicious simplifications at various levels. The efforts to date have given rise to several models, including the extended regular solution theory1-3 and its modification,4 excess free energy model,5-7 the phenomenological model,8,9 modified Wilson model,10,11 the combined nearly ideal binary solvent (NIBS) model,12 the mixture response surface model,13 and others. Many of these models, however, are considered to be more descriptive than predictive, because they inevitably involve one or more model parameters which are usually specific to a particular solute/solvent/cosolvent(s) system, and must be estimated from experimental data of solubility obtained for that system. On the other hand, purely empirical models, e.g., the double-log exponential equation,14 aim at satisfying mathematical descriptions of measured data, and often offer little insight to the process. Comparisons among cosolvency models have been made in several published papers.15-19
998
An Li
This discussion is intended to provide an easy approach for predicting the effect of cosolvents, which are frequently involved in pharmaceutical and environmental applications, on the solubility of organic chemicals. The starting point is the widely used log-linear model. 14.21.2.2 Applications of cosolvency in pharmaceutical sciences and industry Although many drugs are formulated and administrated in solid, vapor, powder, or other forms, using solutions as drug delivery vehicle has significant advantages. Most parenterally administered medicines are in the form of liquid solution, and for intravenous injection, liquid form dosage is the only possible form. However, drugs are usually designed with little concern about their level of solubility in solution. It is the task of pharmaceutical formulators to find appropriate forms for the drug to be effectively delivered into biological systems. The phenomena of cosolvency have been studied for more than a century by pharmaceutical scientists. Numerous experimental data are published in the literature, and most of the models mentioned in the introduction section have been developed from pharmaceutical research. In addition to the need of solubilizing drugs which are poorly water-soluble, controlling the dissolution of drugs administered as solids to optimize therapeutic activity also demands an improved understanding of drug solubilization. Approaches that may be pursued to enhance drug solubility in a liquid dosage formulation include adjusting pH, adding surfactants, cosolvents, or complexation agents. Choice of these techniques depends primarily on the drug’s chemical structure and physicochemical properties. For example, control of pH is applicable only when the drug is an electrolyte. To solubilize nonelectrolyte drugs, the use of cosolvents outweighs surfactants and complexing agents.20 Cosolvents that are routinely used in drug formulation include ethanol, propylene glycol, polyethylene glycol, and glycerin. Examples of pharmaceutical products containing these cosolvents are summarized in Table 14.21.2.1. Table 14.21.2.1. Selected pharmaceutical products containing cosolvents Trade Name
Cosolvent
Aclovate cream
Propylene glycol
Alurate elixir
Ethanol
Amidate
Propylene glycol
Amphojel
Glycerin
Apresoline
Propylene glycol
Aristocort cream
Propylene glycol
Ativan
Polyethylene glycol Propylene glycol
Bentyl syrup
Propylene glycol
Brevibioc
Ethanol
Cleocin T lotion
Glycerin
vol%
Manufacturer
Type
Schering
Topical
20
Roche
Oral liquid
35
Abbott
Parenteral
Wyeth-Ayerst
Oral
Ciba
Parenteral
Fujisawa
Topical
Wyeth-Ayerst
Parenteral
Lakeside
Oral
DuPont
Parenteral
Upjohn
Topical
10
20 80
25
14.21.2 Predicting cosolvency
Trade Name
999
Cosolvent
Comtrex cough
Ethanol
Cyclocort lotion
vol% 20
Manufacturer
Type
Bristol
Oral liquid
Polyethylene glycol
Lederle
Topical
Delsym
Propylene glycol
McNeil Consumer
Oral
Depo-Medrol
Polyethylene glycol
Upjohn
Parenteral
Dilantin
Ethanol Propylene glycol
10 40
Parke-Davis
Parenteral
Dramamine
Propylene glycol
50
Searle
Parenteral
Elocon lotion
Propylene glycol
Schering
Topical
Entex liquid
Glycerin
Norwich-Eaton
Oral
Fluonid solution
Propylene glycol
Herbert
Topical
Halog cream
Propylene glycol
Westwood-Squibb
Topical
Halog ointment
Polyethylene glycol
Westwood-Squibb
Topical
Kwell cream
Glycerin
Reed & Carnrick
Topical
Lanoxin
Ethanol Propylene glycol
10 40
Librium
Propylene glycol
20
Lidex
Propylene glycol
Luminal Sod
Propylene glycol
67.8
MVI-12
Propylene glycol
Nembutal
Ethanol Propylene glycol
Neoloid
Propylene glycol
Nitro-BID IV
Ethanol
Novahistine DH
Glycerin
Paradione
Ethanol
Pentuss
Burroughs Wellcome
Parenteral
Roche
Parenteral
Syntex
Topical
Winthrop
Parenteral
30
Armour
Parenteral
10 40
Abbott
Parenteral
Lederle
Oral
Marion
Parenteral
Lakeside
Oral
Abbott
Oral liquid
Propylene glycol
Fisons
Oral
Psorcon ointment
Propylene glycol
Dermik
Topical
Rondec DM
Glycerin
Ross
Oral
S-T Forte syrup
Ethanol
Scot-Tussin
Oral liquid
Sulfoxyl lotion
Propylene glycol
Stiefel
Topical
Tinactin
Polyethylene glycol
Schering
Topical
Trideslon cream
Glycerin
Miles
Topical
Tussar
Propylene glycol
Rorer
Oral
Tussionex
Propylene glycol
Fisons
Oral
70
65
5
1000
An Li
Trade Name
Cosolvent
Tylenol
Propylene glycol
Valium
Ethanol Propylene glycol
Vepesid
Ethanol
vol%
10 40 30.5
Manufacturer
Type
McNeil Consumer
Oral
Roche
Parenteral
Bristol-Myers
Parenteral
(Data are from reference 21)
14.21.2.3 Applications of cosolvency in environmental sciences and engineering The significance of cosolvency research in environmental sciences stems from the need for accurately modeling the distribution and movement of organic pollutants, and cleaning up polluted soils and sediments. Since the late 1970s, environmental research on the effect of cosolvents has grown steadily. Most published research papers have focused on the effects of adding cosolvents on the aqueous solubility18,22-31 and soil sorption28,32-44 of pollutants of interest. A few researchers have also examined cosolvent effects on liquid phase partitioning.45,46 In the cases of industrial waste discharges, liquid fuel and paint spills, storage tank leakage, landfill leaching, and illegal dumping, various organic solvents may find their way into the natural environment. These solvents may not only act as pollutants themselves, but also bring substantial changes on the distribution, movement, and fate of other environmental pollutants with high concern. In environmental cosolvency studies, the majority of the solutes are hydrophobic organic compounds (HOCs), including benzene and its derivatives, polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins and furans (PCDDs and PCDFs), and various pesticides. These chemicals are toxic, and many of them are mutagenic and carcinogenic. The 1990 Clean Air Act Amendment has stimulated research on gasoline additives such as methyl t-butyl ether (MTBE) and formulated fuels like gasohol. Their cosolvent effects on the solubility and sorption of a few pollutant groups have also been examined.25,47 Meanwhile, environmental engineers have put cosolvents to work in cleaning up contaminated sites. As a consequence of the failure of using traditional pump-and-treat remediation for soils contaminated with organic pollutants, a few new approaches have been experimented since the late 1980s. Among those involving cosolvents, Ex situ solvent extraction was developed to treat excavated soils, sediment, or sludge. A typical one is the basic extractive sludge treatment (B.E.S.T.) process certified by USEPA.48,49 Triethylamine was selected as the extracting solvent due mainly to its inverse miscibility property - it is completely miscible with water below 60oF but separates from water above 90oF. This property makes it easier to recycle the solvents after separating the treated solids from liquids containing the solvent, pollutants, and water. For PCBs in various soils, it is typical to achieve an extraction efficiency higher than 99% using the B.E.S.T technique. More attractive are in situ remediation approaches, which often cost less. Cosolvents promote the mobilization of organic chemicals in soils, thus accelerating the cleanup of contaminated site. Cosolvent flushing has been developed using the same principles as those used in solvent flooding, a technique to enhance petroleum recovery in oil fields. It involves injecting a solvent mixture, mostly water plus a miscible cosolvent, into the vadose
14.21.2 Predicting cosolvency
1001
or saturated zone upgradient of the contaminated area. The solvent with the removed contaminants is then extracted downgradient and treated above ground. Precise formulations for the water/cosolvent mixture need to be determined by laboratory and pilot studies in order to achieve the desired removal.50-53 A few field-scale evaluations of this technique were carried out at Hill Air Force Base, Utah, where the aquifer had been severely contaminated by jet fuel, chlorinated solvents, and pesticides during 1940s and 1950s. These contaminants had formed a complex non-aqueous phase liquid (NAPL) containing more than 200 constituents, which covered the surfaces of soil particles and was trapped in pores and capillaries over the years. One of the evaluations consisted of pumping ternary cosolvent mixture (70% ethanol, 12% n-pentanol, and 18% water) through a hydraulically isolated test cell over a period of 10 days, followed by flushing with water for another 20 days.54,55 The removal efficiency varied from 90-99% at the top zone to 70-80% at the bottom near a confining clay layer. Similar removal efficiencies were obtained from another test cell using a combination of cosolvent n-pentanol and a surfactant at a total of 5.5 wt % of the flushing solution.56 In order to remove gasoline residuals at a US Coast Guard base in Traverse City, Michigan, it was demonstrated that the contaminants were mobilized when cosolvent 2-propanol was used at 50% concentration, while methanol at either 20% or 50% showed little effect.57 Cosolvent flushing was also proven to be effective in treating NAPLs which were denser than water. Methanol, isopropanol, and t-butanol were used in treating soils contaminated with tri- and tetra-chlorinated ethylenes.58 The applicability of solvent flushing, however, is often limited by the characteristics of the soil, especially the particle size distribution. While sandy soils may result in uncontrolled fluid migration, clayey soils with particles size less than 60 µm are often considered unsuitable for in situ solvent flushing due to low soil permeability. In an attempt to remove PAHs from poorly permeable soils, Li, et al.59 investigated the possibility of combining cosolvent flushing with the electrokinetic technique. Electrokinetic remediation involves application of a low direct electrical current to electrodes that are inserted into the ground. As water is continuously replenished at anodes, dissolved contaminants are flushed toward the cathode due to electroosmosis, where they can be extracted and further treated by various conventional wastewater treatment methods. Their column experiment of removing phenanthrene from soil was moderately successful with the assistance of cosolvent n-butylamine at 20%(v). Retardation factor (ratio of the water linear velocity to that of the chemical) of phenanthrene was reduced from 753 in pure water to 11 by the presence of n-butylamine, and 43% of the phenanthrene was removed after 127 days or 9 pore volumes. However, significant removal of phenanthrene was not attained in their experiments with acetone and hydrofuran as cosolvents. 14.21.2.4 Experimental observations Numerous experimental data exist in the literature on the solubility of organic solutes, including both drugs and environmental pollutants, in various mixtures of water and cosolvents. Experimental observations are often illustrated by plotting the logarithm of solubility of the solute versus the volume fraction of cosolvent in the solvent mixture. A few examples of solubilization curves are shown in Figure 14.21.2.1, which shows three typical situations for solutes of different hydrophobicity in the mixture of water and ethanol. The classification of solute/cosolvent/water systems based on their relative polarity was suggested by Yalkowsky and Roseman.61 Solutes which are less polar than both water
1002
An Li
Figure 14.21.2.1. Effects of ethanol on the solubilities of selected organic compounds. (a): n benzene, o naphthalene, ¡ biphenyl, ¯ anthracene, p benzo(a)pyrene, r perylene, ¡ chrysene; (b) n hydantoic acid, o hydantoin, ¡ methyl hydantoic acic, ¯ 5-ethyl hydantoin, r 5-isobutyl hydantoin; (c) o triglycine, ¡ diglycine, r glycine [Adapted, by permission, from Li and Yalkovsky, J. Pharm. Sci., 83, 1735 (1994).]
and the cosolvent are considered as “nonpolar”, those which have a polarity between those of water and the cosolvent as “semipolar”, and those which are more polar than both water and cosolvent as “polar”. Figure 14.21.2.1-a illustrates the behavior of relatively hydrophobic compounds, which tend to have monatonically increasing solubilization curves. The solubility enhancement is greater for the more hydrophobic solutes. Curves with opposite trends were mostly observed for polar solutes. The monatonical desolubilization is greater for more hydrophilic solutes, as evidenced by the curves in Figure 14.21.2.1-c. In-between are semipolar solutes with slightly parabolic curves shown in 14.21.2.1-b. The impact of adding cosolvents is much less profound for the semipolars than for the other two groups. On a linear solubility scale, the parabola tends to be more obvious than on the log scale. The same general trends were seen for the cosolvents glycerine61 and propylene glycol61,62 and presumably many other water-miscible cosolvents. It is more difficult to evaluate the effects of cosolvents which have limited miscibility with water. In the literature, such organic solvents have been termed as both cosolvents and cosolutes, and there is no clear criteria for the distinction. Cosolvent is usually miscible with water, or to be used in an attempt to increase the aqueous solubility of the solute. Cosolute, on the other hand, may be organic chemicals which have a similar chemical structure or behave similarly with the solute when they exist in water alone. The effects of cosolutes have been examined in a limited number of published papers.63-73 Partially water-miscible organic solvents (PMOSs) may act as either cosolvents or cosolutes, and the research in the past has shown the complexity of their effects.23,27-30,73-75 It was demonstrated that in order to exert effects on solubility or sorption of HOCs, PMOSs must exist as a component of the solvent mixture in an appreciable amount: Munz and Roberts23 suggested a mole fraction of greater than 0.005 and Rao and coworkers27,28 proposed a volume percent of 1% or a concentration above 104 mg/L. Cosolvents with relatively high water solubility are likely to demonstrate observable effects on the solubilities of solutes, up to their solubility limits, in a similar manner to cosolvents of complete miscibility with water. A few experimental examples of the effects of PMOSs include 1-butanol and
14.21.2 Predicting cosolvency
1003
1-pentanol acting on PCB congeners30 and naphthalene.26 and butanone on anthracene and fluoranthene.75 Even more hydrophobic organic solvents produce little or even negative influence on the solubility of HOCs. For instance, the presence of benzene does not increase the aqueous solubility of PCBs up to their saturation concentration.29 Solubility of a few PCB congeners in water were found to be depressed by dissolved dichloromethane and chloroform.73 On the other hand, PCB solubility showed little change when cosolvent benzyl alcohol, 1-hexanol, 1-heptanol, or 1-octanol was present.29,30 Similar “no change” observations were made for naphthalene with cosolvents dichloromethane and chloroform,73 and for solutes benzene and hexane with cosolvent MTBE.25 Much of the complexity with hydrophobic cosolvents, or rather, cosolutes, can be explained by the fact that these cosolvents may partition into the solute phase, thus the physical state of the solute is no longer the same as is in pure water. Instead of a basically pure crystalline or liquid phase of solute, the solute and the cosolvent form an organic mixture, and the composition and ideality of this mixture will very much determine the concentrations of its components in the aqueous phase. Such a situation may be better investigated along the line of phase partitioning, where Raoult’s law defines an ideal system. 14.21.2.5 Predicting cosolvency in homogeneous liquid systems The log-linear model Yalkowsky and Roseman introduced the log-linear model in 1984 to describe the phenomenon of the exponential increase in aqueous solubility for nonpolar organic compounds as the cosolvent concentration is increased.61 They showed that log S mi = f log S c + (1 − f ) log Sw
[14.21.2.1]
Rearranging equation [14.21.2.1] results in
(
log S mi / Sw
) = f log(S
c
/ Sw ) = σf
[14.21.2.2]
The left side of equation [14.21.2.2] reflects the extent of solubilization; f defines how much cosolvent is required to reach the desired solubilization. The constant σ is the end-to-end slope of the solubilization curve and defined by: σ = log S c − log Sw = log(S c / Sw )
[14.21.2.3]
The model can be extended to systems containing a number of cosolvents:
(
log S mi / Sw
)=∑σ f
i i
[14.21.2.4]
where the subscribe i denotes the ith component of the solvent mixture. Two measured solubilities will define the value of σ that is specific to a solute/cosolvent pair. The value of σ is also dependent of the solubility unit selected and on whether 10-based or e-based logarithm is used. The magnitude of σ reflects the difference in molecular interactions between solute/cosolvent and solute/water. When applied to describe cosolvency, σ is like a microscopic partition coefficient if water and cosolvent are thought of as two independent entities. There had been other definitions of σ, such as the
1004
An Li
partial derivative ∂(log Sm)/∂f,76 or the regressional (not end-to-end) slope of the solubilization curve.77 The σ defined in these ways will depend on the range of f and on the accuracy of all data points over the entire range of f. These definitions are not desirable because they make σ difficult to predict and interpret in light of the concept of ideal solvent mixture on which the log-linear model is based. Note also that σ is not related to the crystalline structure of the solute, since the contributions from the free energy of melting to the two solubilities cancel out. However, it may change if the solute exists in pure cosolvent with a chemical identity different from that in water, as in the cases where solute degradation, solvation, or solvent-mediated polymorphic transitions occur in either solvent. Estimation of σ Laboratory measurements of Sw and Sc can be costly and difficult. Various methods, including group contribution technique and quantitative structure (or property) property relationships (QSPRs or QPPRs),78 are available to estimate Sw and Sc, from which σ values can be derived. A direct approach of predicting σ has also been established based on the dependence of cosolvency on solute hydrophobicity. Among a number of polarity indices, octanol/water partition coefficient, Kow, was initially chosen by Yalkowsky and Roseman61 for correlation with σ, due mainly to the abundance of available experimental Kow data and the wide acceptance of the Hansch-Leo fragment method79 for its estimation. Kow is a macroscopic property which does not necessarily correlate with micro-scale polarity indices such as dipole moment, and only in a rank order correlates with other macroscopic polarity indicators such as surface tension, dielectric constant, and solubility parameter. Correlation between σ and solute Kow takes the form: σ = a + b log K ow
[14.21.2.5]
where a and b are constants that are specific for the cosolvent but independent of solutes. Their values have been reported for various cosolvents and are summarized in Table 14.21.2.2. From Table 14.21.2.2, the slopes of equation [14.21.2.5], b, are generally close to unity, with few below 0.6 or above 1.2. Most of the intercepts a are less than one, with a few negative values. In searching for the physical implications of the regression constants a and b, Li and Yalkowsky80 derived equation [14.21.2.6]:
(
)
(
)
(
σ = log K ow + log γ ∞0 * / γ c + log γ w / γ w∞* + log V0* / Vc
)
[14.21.2.6]
According to this equation, σ~log Kow correlation will indeed have a slope of one and a ∞* predictable intercept of log (Vo*/Vc) if both γ ∞* 0 / γ c and γ w / γ w terms equal unity. Vo* = -1 0.119 L mol based on a solubility of water in octanol of 2.3 mol L-1, and Vc ranges from 0.04 to 0.10 L3 mol-1, thus log (Vo*/Vc) is in the range of 0.08 to 0.47, for the solvents included in Table 14.21.2.2 with the exclusion of PEG400. However, both γ ∞* 0 / γ c and γ w / γ w∞* are not likely to be unity, and their accurate values are difficult to estimate for many solutes. The ratio γ ∞* 0 / γ c compares the solute behavior in water-saturated octanol under dilute conditions and in pure cosolvent at saturation, while the γ w / γ w∞* term reflects both the effect of dissolved octanol on the aqueous activity coefficient and the variation of the activity coefficient with concentration. Furthermore, the magnitudes of both terms will vary from one solute to another, making it unlikely that a unique regression intercept will be ob-
14.21.2 Predicting cosolvency
1005
served over a wide range of solutes. Indeed, both a and b were found to be dependent on the range of the solute log Kow used in the regression. For instance, for solutes with log Kow ≤ 0, 0.01 to 2.99, and ≥3, the correlation of σ versus log Kow for cosolvent ethanol have slopes of 0.84, 0.79, and 0.69, respectively, and the corresponding intercepts increase accordingly; the slope of the overall correlation, however, is 0.95. Much of the scattering on the σ~log Kow regression resides on the region of relatively hydrophilic solutes. Most polar solutes dissociate to some extent in aqueous solutions, and their experimental log Kow values are less reliable. Even less certain is the extent of specific interactions between these polar solutes and the solvent components. According to equation [14.21.2.6], σmay not be a linear function of solute log Kow on a theoretical basis. However, despite the complexities caused by the activity coefficients, quality of the regression of σ against log Kow is generally high, as evidenced by the satisfactory R2 values in Table 14.21.2.2. This can be explained by the fact that changes in both γ ratios are much less significant compared with the variations of Kow for different solutes. In addition, the two γ terms may cancel each other to some degree for many solutes, further reducing their effects on the correlation between σ and log Kow. It is convenient and reliable to estimate σ from known log Kow of the solute of interest, especially when the log Kow of the solute of interest falls within the range used in obtaining the values of a and b. Dependence of σ on the properties of cosolvents has been less investigated than those of solutes. While hundreds of solutes are involved, only about a dozen organic solvents have been investigated for their cosolvency potentials. A few researchers examined the correlations between σ and physicochemical properties of cosolvents for specific solutes, for instance, Li et al. for naphthelene,31 and Rubino and Yalkowsky for drugs benzocaine, diazepam, and phenytoin.81 In both studies, hydrogen bond donor density (HBD), which is the volume normalized number of proton donor groups of a pure cosolvent, is best for comparing cosolvents and predicting σ. Second to HBD are the solubility parameter and interfacial tension (as well as log viscosity and ET-30 for naphthalene systems), while log Kow, dielectric constant, and surface tension, correlate poorly with σ. The HBD of a solvent can be readily calculated from the density and molecular mass with the knowledge of the chemical structure using equation [14.21.2.7]. The disadvantage of using HBD is that it cannot distinguish among aprotic solvents which have the same HBD value of zero. HBD = (number of proton donor groups)(density)/(molecular mass)
[14.21.2.7]
In an attempt to generalize over solutes, Li and Yalkowsky82 investigated the possible correlations between cosolvent properties and slope of the σ~log Kow regressions (b). Among the properties tested as a single regression variable, octanol-water partition coefficient, interfacial tension, and solubility parameter, are superior to others in correlating with b. Results of multiple linear regression show that the combination of log Kow and HDB of the cosolvent is best (equation [14.21.2.8]). Adding another variable such as solubility parameter does not improve the quality of regression. b = 0.2513 log Kow - 0.0054 HBD + 1.1645
[14.21.2.8]
(N = 13, R2 = 0.942, SE = 0.060, F = 81.65) where log Kow (range: -7.6 ~ 0.29) and HBD (range: 0 ~ 41) are those of the cosolvent.
1006
An Li
Equation [14.21.2.8] can be helpful to obtain the b values for cosolvents not listed in Table 14.21.2.2. In order to estimate cosolvency for such cosolvents, values of the intercept a are also needed. However, values of a can be found in Table 14.21.2.2 for only about a dozen cosolvents, and there is no reliable method for its estimation. To obtain a for a cosolvent, a reasonable starting value can be log (Vo*/Vc), or -(0.92 + log Vc). The average absolute difference between a values listed in Table 14.21.2.2 and log (Vo*/Vc) is 0.18 (N=7) for alcohols and glycols, 0.59 (N = 6) for aprotic cosolvents, and > 1 for n-butylamine and PEG400.82 Table 14.21.2.2. Summary of regression results for relationship between σ and solute log Kow Cosolvent
N
a
b
R2
log Kow range
Ref.
Methanol
79
0.36±0.07
0.89±0.02
0.96
-4.53 ~ 7.31
80
Methanol
16
1.09
0.57
0.83
2.0 ~ 7.2
29
Methanol
16
1.07
0.68
0.84
n.a.
24
Ethanol
197
0.30±0.04
0.95±0.02
0.95
-4.90 ~ 8.23
80
Ethanol
107
0.40±0.06
0.90±0.02
0.96
-4.9 ~ 6.1
60
Ethanol
11
0.81
0.85
0.94
n.a.
24
1-Propanol
17
0.01±0.13
1.09±0.05
0.97
-3.73 ~ 7.31
80
2-Propanol
20
-0.50±0.18
1.11±0.07
0.94
-3.73 ~ 4.49
80
2-Propanol
9
0.63
0.89
0.85
n.a.
24
Acetone
22
-0.10±0.24
1.14±0.07
0.92
-1.38 ~ 5.66
80
Acetone
14
0.48
1.00
0.93
0.6 ~ 5.6*
24
Acetonitrile
10
-0.49±0.42
1.16±0.16
0.86
-0.06 ~ 4.49
80
Acetonitrile
8
0.35
1.03
0.90
n.a.
24
Dioxane
23
0.40±0.16
1.08±0.07
0.91
-4.90 ~ 4.49
80
Dimethylacetamide
11
0.75±0.30
0.96±0.12
0.87
0.66 ~ 4.49
80
Dimethylacetamide
7
0.89
0.86
0.95
n.a.
24
Dimethylformamide
11
0.92±0.41
0.83±0.17
0.73
0.66 ~ 3.32
80
Dimethylformamide
7
0.87
0.87
0.94
n.a.
24
Dimethylsulfoxide
12
0.95±0.43
0.79±0.17
0.68
0.66 ~ 4.49
80
Dimethylsulfoxide
7
0.89
0.87
0.95
n.a.
24
Glycerol
21
0.28±0.15
0.35±0.05
0.72
-3.28 ~ 4.75
80
Ethylene glycol
13
0.37±0.13
0.68±0.05
0.95
-3.73 ~ 4.04
80
Ethylene glycol
7
1.04
0.36
0.75
n.a.
24
Propylene glycol
62
0.37±0.11
0.78±0.04
0.89
-7.91 ~ 7.21
80
14.21.2 Predicting cosolvency
Cosolvent
1007
N
a
b
R2
log Kow range
Ref.
Propylene glycol
47
0.03
0.89
0.99
-5 ~ 7
61
Propylene glycol
8
0.77
0.62
0.96
n.a.
24
PEG400
10
0.68±0.43
0.88±0.16
0.79
-0.10 ~ 4.18
80
Butylamine
4
1.86±0.30
0.64±0.10
0.96
-1.69 ~ 4.49
80
*estimated from Figure 3 in Reference 24. n.a. = not available.
This empirical approach using equations [14.21.2.8], [14.21.2.5], and [14.21.2.2] can produce acceptable estimates of log (Sm/Sw) only if the solubilization exhibits a roughly log-linear pattern, such as in some HOC/water/methanol systems. In addition, it is important to limit the use of equations [14.21.2.5] and [14.21.2.8] within the ranges of log Kow used in obtaining the corresponding parameters. 14.21.2.6 Predicting cosolvency in non-ideal liquid mixtures Deviations from the log-linear model Most solubilization curves, as shown in Figure 14.21.2.1, exhibit significant curvatures which are not accounted for by the log-linear model. A closer look at the solubilization curves in Figure 14.21.2.1 reveals that the deviation can be concave, sigmoidal, or convex. In many cases, especially with amphiprotic cosolvents, a negative deviation from the end-to-end log-linear line is often observed at low cosolvent concentrations, followed by a more significant positive deviation as cosolvent fraction increases. The extent of the deviation from the log-linear pattern, or the excess solubility, is measured by the difference between the measured and the log-linearly predicted log Sm values:
(
)
log S m / S mi = log S m − (log Sw + ∑ σ i f i )
[14.21.2.9]
The values of log (Sm/Sim) for naphthalene, benzocaine, and benzoic acid in selected binary solvent mixtures are presented in Figures 14.21.2.2-a, -b, and -c, respectively. The log-linear model is based on the presumed ideality of the mixtures of water and cosolvent. The log-linear relationship between log (Sm/Sw) and f is exact only if the cosolvent is identical to water, which cannot be the case in reality. Deviation is fortified as any degradation, solvation, dissociation, or solvent mediated polymorphic transitions of the solute occur. The problem is further compounded if the solute dissolves in an amount large enough to exert significant influence on the activity of solvent components. Due to the complexity of the problem, efforts to quantitatively describe the deviations have achieved only limited success. A generally accepted viewpoint is that the deviation from the log-linear solubilization is mainly caused by the non-ideality of the solvent mixture. This is supported by the similarities in the patterns of observed log Sm and activities of the cosolvent in solvent mixture, when they are graphically presented as functions of f. Based on the supposition that solvent non-ideality is the primary cause for the deviation, Rubino and Yalkowsky87 examined the correlations between the extent of deviation and various physical properties of solvent mix-
1008
An Li
Figure 14.21.2.2a. Deviations from log-linear model (equation [14.21.2.2], triangle) and the extended log-linear model (equation [14.21.2.10], circle) for solute naphthalene in various water cosolvent systems. Experimental data are from Ref. 83.
tures. However, none of the properties consistently predicted the extrema of the deviation, although density corresponded in several cases. Non-ideality of a mixture is quantitatively measured by the excess free energy of mixing. From this standpoint, Pinal et al.75 proposed that a term Σ(fi ln γi) be added to equation [14.21.2.4] to account for the effect of the non-ideality of solvent mixture:
(
log S mii / Sw
)=∑σ f
i i
+ 2.303∑ f i log γ i
[14.21.2.10]
where γi is the activity coefficient of solvent component i in solute-free solvent mixture. Values of γ‘s can be calculated by UNIFAC, a group contribution method for the prediction of activity coefficients in nonelectrolyte, nonpolymeric liquid mixtures.88 UNIFAC derived activity coefficients are listed in Table 14.21.2.3 for selected cosolvent-water mixtures. They are calculated with UNIFAC group interaction parameters derived from vapor-liquid equilibrium data.89,90 The difference between the experimental log Sm and that predicted by the extended log-linear model, i.e., equation [14.21.2.10], is
14.21.2 Predicting cosolvency
1009
Figure 14.21.2.2b. Deviations from log-linear model (equation [14.21.2.2], triangle) and the extended log-linear model (equation [14.21.2.10], circle) for solute benzocaine in various water cosolvent systems. Experimental data are from Refs. 84 and 85.
(
)
log S m / S mii = log S m − (log Sw + ∑ σ i f i + 2.303∑ f i log γ i ) [14.21.2.11] Results of equation [14.21.2.11] for naphthalene, benzocaine, and benzoic acid in selected binary solvent mixtures are also included in Figure 14.21.2.2. A few other examples can be found in Pinal et al.75 The extended log-linear model outperforms the log-linear model in more than half of the cases tested for the three solutes in Figure 14.21.2.2. The improvement occurs mostly in regions with relatively high f values. In the low f regions, negative deviations of solubilities from the log-linear pattern are often observed as discussed above, but are not accounted for by the extended log-linear model as presented by equation [14.21.2.10]. In some cases, such as naphthalene in methanol and propylene glycol, and benzoic acid in ethylene glycol, the negative deviations occur over the entire f range of 0~1. In these cases, the extended log-linear model does not offer better estimates than the original log-linear model. With the activity coefficients listed in Table 14.21.2.3, the extended log-linear model generates worse estimates of log (Sm/Sw) than the log-linear model for systems containing dimethylacetamide, dimethylsulfoxide, or dimethylformamide. There is a possibility that
1010
An Li
Figure 14.21.2.2c. Deviations from log-linear model (equation [14.21.2.2], triangle) and the extended log-linear model (equation [14.21.2.10], circle) for solute benzoic acid in various water cosolvent systems. Experimental data are from Refs. 61 and 86.
the UNIFAC group interaction parameters involved in these systems are incorrect. With all the systems tested in this study with solute naphthalene, benzocaine, or benzoic acid, it is also found that replacing fi in the last term of equation [14.21.2.11] with mole fraction xi offers slight improvement in only a few cases. Dropping the logarithm conversion constant 2.303 results in larger estimation errors for most systems. An apparent limitation of this modification is the exclusion of any active role the solute may play on the observed deviation. Little understanding of the influence of solute structure and properties on deviations from the log-linear equation has been obtained. Although the patterns of deviations tend to be similar among solutes, as mentioned above, the extent of deviation is solute-dependent. For instance, C1~C4 alkyl esters of p-hydroxybenzoates and p-aminobenzoates demonstrated similar characteristics of solubilization by propylene glycol, with a negative deviation from the log-linear pattern occurring when f is low, followed by a positive one when f increases.91 The magnitude of the negative deviation, however, was found to be related to the length of the solute alkyl chain in each group, while that of the positive deviation to the type of the polar groups attached.91
14.21.2 Predicting cosolvency
1011
Both the hydrophobicity and hydrogen bonding property of the solutes seem to be important in influencing the extent of the deviation from the ideal log-linear pattern. Additional deviations related to the solute’s behavior may occur. For organic electrolytes, the acid dissociation constant Ka may decrease as cosolvent fraction f increases.40,75,92 This, in turn, will affect the patterns of solubilization by cosolvents. Furthermore, a high concentration of solutes may invalidate the log-linear model, which presumes negligible volume fraction of solute and no solute-solute interactions. For solid solutes, solvent induced polymorphism may also bring additional changes in their solubilization profile. Another approach to quantitatively address the deviations of solubilization from the log-linear model makes use of an empirical parameter β:
(
β = log(S m / Sw ) / log S mi / Sw
)
[14.21.2.12]
The modified log-linear equation then takes the form: log(S m / Sw ) = β∑ σ i f i Table 14.21.2.3. UNIFAC water-cosolvent systems f
0.1
Methanol
[14.21.2.13]
derived
0.2
activity
0.4
MW = 32.04
coefficients
0.6
for
selected
0.8
binary
0.9
Density = 0.7914
mol/L
2.4700
4.9401
9.8801
14.8202
19.7603
22.2303
x
0.0471
0.1000
0.2286
0.4001
0.6401
0.8001
γ, cosolvent
1.972
1.748
1.413
1.189
1.052
1.014
γ, water
1.003
1.013
1.055
1.14
1.298
1.424
Ethanol
MW = 46.07
Density = 0.7893
mol/L
1.7133
3.4265
6.8530
10.2796
13.7061
15.4194
x
0.0331
0.0716
0.1705
0.3163
0.5523
0.7351
γ, cosolvent
5.550
4.119
2.416
1.564
1.152
1.050
γ, water
1.005
1.022
1.097
1.256
1.57
1.854
1-Propanol
MW = 60.1
Density = 0.8053
mol/L
1.3399
2.6799
5.3597
8.0396
10.7195
12.0594
x
0.0261
0.0569
0.1385
0.2657
0.4910
0.6846
γ, cosolvent
12.77
8.323
3.827
2.001
1.248
1.077
γ, water
1.006
1.024
1.111
1.301
1.706
2.093
2-Propanol mol/L
MW = 60.1 1.3058
2.6116
Density = 0.7848 5.2233
7.8349
10.4466
11.7524
1012
An Li
f
0.1
0.2
0.4
0.6
0.8
0.9
x
0.0255
0.0555
0.1355
0.2607
0.4846
0.6790
γ, cosolvent
12.93
8.488
3.921
2.040
1.258
1.080
γ, water
1.006
1.023
1.107
1.294
1.695
2.084
Acetone
MW = 58.08
Density = 0.7899
mol/L
1.3600
2.7200
5.4401
8.1601
10.8802
12.2402
x
0.0265
0.0577
0.1403
0.2686
0.4947
0.6878
γ, cosolvent
8.786
6.724
3.952
2.370
1.484
1.196
γ, water
1.004
1.015
1.075
1.222
1.616
2.211
Acetonitrile
MW = 41.05
Density = 0.7857
mol/L
1.9140
3.8280
7.6560
11.4840
15.3121
17.2261
x
0.0369
0.0793
0.1868
0.3407
0.5795
0.7561
γ, cosolvent
10.26
7.906
4.550
2.536
1.501
1.126
γ, water
1.005
1.021
1.11
1.366
2.076
3.571
Dioxane
MW = 88.11
Density = 1.0329
mol/L
1.1723
2.3446
4.6891
7.0337
9.3783
10.5506
x
0.0229
0.0501
0.1233
0.2404
0.4577
0.6551
γ, cosolvent
14.71
8.743
3.616
1.833
1.171
1.124
γ, water
1.006
1.026
1.112
1.284
1.604
1.666
DMA
MW = 87.12
Density = 0.9429
mol/L
1.0823
2.1646
4.3292
6.4938
8.6584
9.7407
x
0.0212
0.0464
0.1149
0.2261
0.4380
0.6368
γ, cosolvent
0.121
0.141
0.204
0.330
0.602
0.826
γ, water
0.999
0.994
0.962
0.872
0.651
0.453
DMF
MW = 73.1
Density = 0.9445
mol/L
1.2921
2.5841
5.1683
7.7524
10.3365
11.6286
x
0.0252
0.0549
0.1342
0.2586
0.4819
0.6767
γ, cosolvent
0.833
0.873
0.930
0.962
0.983
0.985
γ, water
0.999
0.997
0.991
0.984
0.972
0.969
DMSO
MW = 78.13
Density = 1.10
mol/L
1.4079
2.8158
5.6316
8.4475
11.2633
12.6712
x
0.0274
0.0596
0.1445
0.2754
0.5034
0.6952
γ, cosolvent
0.07956
0.110
0.211
0.399
0.715
0.899
0.996
0.981
0.913
0.774
0.540
0.386
γ, water
14.21.2 Predicting cosolvency
f
0.1
Glycerol
1013
0.2
0.4
MW = 92.1
0.6
0.8
0.9
Density = 1.2611
mol/L
1.3693
2.7385
5.4771
8.2156
10.9542
12.3235
x
0.0267
0.0580
0.1411
0.2699
0.4964
0.6893
γ, cosolvent
1.257
1.066
0.903
0.899
0.969
0.996
γ, water
1.003
1.010
1.027
1.025
0.979
0.942
Ethylene glycol
MW = 62.07
Density = 1.1088
mol/L
1.7864
3.5727
7.1455
10.7182
14.2910
16.0773
x
0.0345
0.0744
0.1765
0.3254
0.5626
0.7432
γ, cosolvent
2.208
1.923
1.494
1.214
1.053
1.013
γ, water
1.002
1.01
1.047
1.12
1.247
1.338
Propylene glycol
MW = 76.09
Density = 1.0361
mol/L
1.3617
2.7234
5.4467
8.1701
10.8934
12.2551
x
0.0265
0.0577
0.1405
0.2688
0.4951
0.6881
γ, cosolvent
3.392
2.498
1.567
1.177
1.044
1.019
γ, water
1.005
1.018
1.069
1.145
1.224
1.267
Butylamine
MW = 73.14
Density = 0.7414
mol/L
1.0137
2.0273
4.0547
6.0820
8.1094
9.1231
x
0.0199
0.0436
0.1084
0.2149
0.4219
0.6215
γ, cosolvent
6.532
4.498
2.318
1.391
1.042
0.998
γ, water
1.004
1.016
1.071
1.175
1.326
1.384
Under the assumptions that the solute is chemically stable and has little influence on the activity of solvent component, β reflects the extent of deviation caused by the nonideality of the solvent mixture, as suggested by Rao et al.28 However, since β itself is a complicated function of f, equation [14.21.2.13] does not provide additional aid for predicting cosolvency. 14.21.2.7 Summary Applications of cosolvency in pharmaceutical and environmental research and industries are briefly summarized. Using ethanol as an example, the effects of adding a cosolvent on the solubilities of various organic solutes are presented in Figure 14.21.2.1. The log-linear solubilization model, equation [14.21.2.2] or [14.21.2.4], is the simplest theory of cosolvency developed so far. It discovers general trends and major determinant factors of cosolvency, thus providing guidelines for predicting solubility of organic chemicals in mixed solvents. The cosolvency power of a specific cosolvent towards a solute of interest, σ, can be estimated with equation [14.21.2.5] with the knowledge of the solute octanol-water partition coefficient Kow. Sources of error associated with this estimation method are discussed based on equation [14.21.2.6]. The slope of the σ~log Kow regression, b, can be
1014
An Li
estimated from the log Kow and hydrogen bond donor density of the cosolvent, as presented by equation [14.21.2.8]. One of the previously published modifications to the log-linear model, equation [14.21.2.10], is evaluated. The difference between the measured log Sm and those predicted by the log-linear and the extended log-linear model are presented in Figure 14.21.2.2 for solutes naphthalene, benzocaine, and benzoic acids in selected water and cosolvent mixtures. Notations a b f Kow Sc Sm Sim Siim Sw Vo* Vc Vw* Vw β σ γ ∞* 0 γc γw γ ∞w *
intercept of σ ~ log Kow regression slope of σ ~ log Kow regression volume fraction of cosolvent in mixed solvent with water. n-octanol water partition coefficient solubility in pure cosolvent solubility in the mixture of water and cosolvent solubility in the mixture of water and cosolvent, predicted by the log-linear model (Eq. [14.21.2.2]) solubility in the mixture of water and cosolvent, predicted by the extended log-linear model (Eq. [14.21.2.10]) solubility in pure water molar volume of 1-octanol saturated with water, 0.119 L mol-1 (based on a solubility of water in octanol of 2.3 mol L-1) molar volume of cosolvent molar volume of water saturated with 1-octanol, ≈ 0.018 L mol-1 molar volume of water, 0.018 L mol-1 empirically obtained water-cosolvent interaction parameter cosolvency power, σ = log (Sc/Sw) infinite dilution activity coefficient of solute in 1-octanol saturated with water activity coefficient of solute in cosolvent activity coefficient of solute in water infinite dilution activity coefficient of solute in water saturated with 1-octanol
References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
A. Martin, J. Newburger, and A. Adjel, J. Pharm. Sci., 68, 4 (1979). A. Martin, J. Newburger, and A. Adjel, J. Pharm. Sci., 69, 487 (1980). A. Martin, A. N. Paruta, and A. Adjel, J. Pharm. Sci., 70, 1115 (1981). P. Bustamante, B. Escalera, A. Martin, and E. Selles, Pharm. Pharmacol., 45, 253 (1993). N. A. Williams, and G. L. Amidon, J. Pharm. Sci., 73, 9 (1984). N. A. Williams, and G. L. Amidon, J. Pharm. Sci., 73, 14 (1984). N. A. Williams, and G. L. Amidon, J. Pharm. Sci., 73, 18 (1984). D. Khossrani, and K. A. Connors, J. Pharm. Sci., 82, 817 (1993). D. Khossravi, and K. A. Connors, J. Pharm. Sci., 81, 371 (1992). A. Jouyban-Gharamaleki, Chem. Pharm. Bull., 46, 1058 (1998). W. E. Acree, Jr., J. W. McCargar, A. I. Zvaigzne, and I. L. Teng, Phys. Chem. Liq., 23, 27 (1991) W. E. Acree, Jr. and A. I. Zvaigzne, Thermochimica. Acta, 178, 151 (1991). A. B. Ochsner, R. J. Belloto Jr., and T. D. Sololoski, J. Pharm. Sci., 74, 132 (1985). M. Barzegar-Jalali, and J. Hanaee, Int. J. Pharm., 109, 291 (1994). A. Li, and A. W. Andren, Environ. Sci. Technol., 29, 3001 (1995). J. K. Fu, and R. G. Luthy, J. Environ. Eng., 112, 328 (1986). M. Barzegar-Jalali, and A. Jouyban-Gharamaleki, Int. J. Pharm., 140, 237 (1996). R. M. Dickhut, D. E. Armstrong, and A. W. Andren, Environ. Toxicol. Chem., 10, 881 (1991). A. Jouyban-Gharamaleki, L. Valaee, M. Barzegar-Jalali, B. J. Clark, and W. E. Acree, Jr., Intern. J. Pharm., 177, 93 (1999). S. H. Yalkowsky, in Techniques of solubilization of drugs; S. H. Yalkowsky, Ed.; Dekker, New York, 1984, Chapter 1.
14.21.2 Predicting cosolvency
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
1015
S. C. Smolinske, Handbook of Food, Drug, and Cosmetic Excipients, CRC Press, Ann Arbor, MI, 1992. F. Herzel, and A. S. Murty, Bull. Environ. Toxicol., 32, 53 (1984). C. Munz, and P. Roberts, Environ. Sci. Technol., 20, 830 (1986). K. R. Morris, R. Abramowitz, R. Pinal, P. Davis, and S. H. Yalkowsky, Chemosphere, 17, 285 (1988). F. R. Groves, Jr, Environ. Sci. Technol., 22, 282 (1988). R. M. Dickhut, A. W. Andren, and D. E. Armstrong, J. Chem. Eng. Data, 34, 438 (1989). R. Pinal, P. S. C. Rao, L. S. Lee, P. V. Cline, and S. H. Yalkowsky, Environ. Sci. Technol., 24, 639 (1990). P. S. C. Rao, L. S. Lee, and A. L. Wood, EPA/600/M-91/009 (1991). A. Li, W. J. Doucette, and A. W. Andren, Chemosphere, 24, 1347 (1992). A. Li, and A. W. Andren, Environ. Sci. Technol., 28, 47 (1994). A. Li, A. W. Andren, and S. H. Yalkowsky, Environ. Toxicol. Chem., 15, 2233 (1996). P. Nkedi-Kizza, P. S. C. Rao, and A. G. Hornsby, Environ. Sci. Technol., 19, 975 (1985). J. K. Fu, and R. G. Luthy, J. Environ. Eng., 112, 346 (1986). R. W. Walters, and A. Guiseppl-Elie, Environ. Sci. Technol., 22, 819 (1988). P. Nkedi-Kizza, M. L. Brusseau, P. S. C. Rao, and A. G. Hornsby, Environ. Sci. Technol., 23, 814 (1989). A. L. Wood, D. C. Bouchard, M. L. Brusseau, and P. S. C. Rao, Chemosphere, 21, 575 (1990). P. S. C. Rao, L. S. Lee, and R. Pinal, Environ. Sci. Technol., 24, 647 (1990). F. C. Spurlock, and J. W. Biggar, Environ. Sci. Technol., 28, 1003 (1994). W. J. M. Hegeman, C. H. Van der Weijden, and J. P. G. Loch, Environ. Sci. Technol., 29, 363 (1995). L. S. Lee, and P. S. C. Rao, Environ. Sci. Technol., 30, 1533 (1996). V. A. Nzengung, E. A. Voudrias, P. Nkedi-Kizza, J. M. Wampler, and C. E. Weaver, Environ. Sci. Technol., 30, 89 (1996). R. P. Singh, Colloids and Surfaces A., 122, 63 (1997). V. A. Nzengung, P. Nkedi-Kizza, R. E. Jessup, and E. A. Voudrias, Environ. Sci. Technol., 31, 1470 (1997). T. C. Harmon, T. J. Kim, B. K. D. Barre, and C. V. Chrysikopoulos, J. Environ. Eng., January, 87 (1999). M. A. Ei-Zoobi, G. E. Ruch, and F. R. Groves Jr., Environ. Sci. Technol., 24, 1332 (1990). W. F. Lane, and R. C. Loehr, Environ. Sci. Technol., 26, 983 (1992). C. S. Chen, and J. J. Delfino, J. Environ. Eng., April 354 (1997). G. R. Jones, Environ. Prog., 11, 223 (1992). USEPA, Assessment and Remediation of contaminated Sediments (ARCS) Program: Remediation Guidance Document. EPA 905-R94-003, (1994) p.180. USEPA, EPA 542-K-94-006 (1995). C. T. Jafvert, Ground-Water Remediation Technologies Analysis Center, Echnology Evaluation Report, TE-96-02 (1996). R. W. Falta, GWMR, Summer, 94 (1998). D. C. M. Augustijin, R. E. Jessup, P. S. Rao, and A. L. Wood, J. Environ. Eng., 120, 42 (1994). R. K. Sillan, M. D. Annable, P. S. C. Rao, D. Dai, K. Hatfield, W. D. Graham, A. L. Wood, and C. G. Enfield, Water Resources Res., 34, 2191 (1998). P. S. C. Rao, M. D. Annable, R. K. Sillan, D. Dai, K. Hatfield, and W. D. Graham, Water Resources Res., 33, 2673 (1997). J. W. Jawitz, M. D. Annable, P. S. C. Rao, and R. D. Rhue, Environ. Sci. Technol., 32, 523 (1998). A. T. Kan, M. B. Tomson, and T. A. McRae, Proceedings of the 203rd American Chemical Society National Meeting, San Francisco, CA (1992). D. Brandes, and K. J. Farley, J. Water Environ. Res., 65, 869 (1993). A. Li, K. A. Cheung, and K. Reddy, J. Environ. Eng., 126, 527 (2000). A. Li, and S. H. Yalksowsky, J. Pharm. Sci., 83, 1735 (1994). S. H. Yalkowsky, and T. J. Roseman, in Techniques of solubilization of drugs; S. H. Yalkowsky, Ed.; Dekker, New York, 1984, Chapter 3. J. T. Rubino, and S. H. Yalkowsky, J. Pharm. Sci., 74, 416 (1985). P. J. Leinonen, and D. Mackay, Can. J. Chem. Eng., 51, 230 (1973). R. P. Eganhouse, and J. A. Calder, Geochimca et Cosmochimica Acta, 40, 555 (1976). Y. B. Tewari, D. E. Martire, S. P.; Wasik, and M. M. Miller, J. Solution Chem., 11, 435 (1982). S. Banerjee, Environ. Sci. Technol., 18, 587 (1984). D. R. Burris, and W. G. MacIntyre, Environ. Toxicol. Chem., 4, 371 (1985). H. H. Hooper, S. Michel, and J. M. Prausnitz, J. Chem. Eng. Data, 33, 502 (1988). D. Mackay, J. Contam. Hydrol. 8, 23 (1991). A. Li, and W. J. Doucette, Environ. Toxicol. Chem., 12, 2031 (1993). S. Lesage, and S. Brown, J. Contam. Hydrol., 15, 57 (1994).
1016
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92
George Wypych
K. Brololm, and S. Feenstra, Environ. Toxicol. Chem., 14, 9 (1995). G. T. Coyle, T. C. Harmon, and I. H. Suffet, Environ. Sci. Technol., 31, 384 (1997). P. S. C. Rao, L. S. Lee, P. Nkedi-Kizza, and S. H. Yalkowsky, in Toxic Organic Chemicals in Porous Media, Z. Gerstl, Eds. Springer-Verlag, New York, 1989, Chapter 8. R. Pinal, L. S. Lee, and P. S. C. Rao, Chemosphere, 22, 939 (1991). S. H. Yalkowsky, and J. T. Rubino, J. Pharm. Sci., 74, 416 (1985). J. T. Rubino, and S. H. Yalkowsky, J. Parent. Sci. Technol., 41, 172 (1984). W. J. Lyman, W. f. Reehl, and D. H. Rosenblatt, Handbook of Chemical Property Estimation Methods: Environmental Behavior of Organic Compounds. ACS Publications, Washington, DC, 1990. C. Hansch, and A. J. Leo, Substituent constants for Correlation Analysis in Chemistry and Biology. John Wiley, New York, 1979. A. Li, and S. H. Yalkowsky, Ind. Eng. Chem. Res., 37, 4470 (1998). J. T. Rubino, and S. H. Yalkowsky, Pharm. Res., 4, 220 (1987). A. Li, and S. H. Yalkowsky, Ind. Eng. Chem. Res., 37, 4476 (1998). R. Abramowitz, Ph.D. Dissertation, University of Arizona, 1986. A. Li, and S. H. Yalkowsky, unpublished data. J. T. Rubino, Ph.D. Dissertation, University of Arizona, 1984. S. H. Yalkowsky, unpublished data. J. T. Rubino, and S. H. Yalkowsky, Pharm. Res., 4, 231 (1987). A. Fredenslund, R. L. Jones, and J. M. Prausnitz, A.I.Ch.E. J., 21, 1086 (1975). J. Gmehling, P. Rasmussen, and A. Fredenslund, Ind. Eng. Chem. Process Des. Dev., 21, 118 (1982). E. A. Macedo, U. Weidlich, J. Gmehling, and P. Rasmussen, Ind. Eng. Chem. Process Des. Dev., 22, 678 (1983). J. T. Rubino, and E. K. Obeng, J. Pharm. Sci., 80, 479 (1991). L. S. Lee, C. A. Bellin, R. Pinal, and P. S. C. Rao, Environ. Sci. Technol., 27, 165 (1997).
14.22 POLYMERS AND MAN-MADE FIBERS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The resin production industry has over 450 plants in the USA with total sales of $33 billion/year and about 60,000 employees. The man-made fiber industry has over 90 plants. It employs about 45,000 people and it has sales of $13 billion/year. In the polymer manufacture industry, production processes are diverse both in technology and equipment design. They have common steps which include preparation of reactants, polymerization, polymer recovery, polymer extrusion (if in pelletized form), and supporting operations. In some preparation operations, solvents are used to dissolve or dilute monomer and reactants. Solvent are also used to facilitate the transportation of the reaction mixture throughout the plant, to improve heat dissipation during the reaction, and to promote uniform mixing. Solvent selection is optimized to increase monomer ratio and to reduce polymerization costs and emissions. The final polymer may or may not be soluble in the solvent. These combinations of polymers and solvents are commonly used: HDPE - isobutane and hexane, LDPE - hydrocarbons, LLDPE - octene, butene, or hexene, polypropylene - hexane, heptane or liquid propylene, polystyrene - styrene or ethylbenzene, acrylic dimethylacetamide or aqueous inorganic salt solutions. These examples show that there are options available. Excess monomer may replace solvent or water can be used as the solvent. During polymer recovery unreacted monomer and solvents are separated from polymer (monomers and solvents are flashed off by lowering the pressure and sometimes degassing under vacuum), liquids and solids are separated (the polymer may be washed to remove sol-
14.22 Polymers and man-made fibers
1017
Figure 14.22.1. Schematic diagram of emissions from the polymer manufacturing industry. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995.]
vent), and residual water and solvent are purged during polymer drying. Residual solvents are removed by further drying and extrusion. Solvents are also used in equipment cleaning. Solvents are often stored under a nitrogen blanket to minimize oxidation and contamination. When these systems are vented solvent losses occur. Figure 14.22.1 shows a schematic diagram of potential emissions during polymer manufacture. Manufacture of man-made fibers involves polymerization (usually the core part of the process), preparation of the solution, spinning, washing and coagulation, drying and other operations. Fibers are formed by forcing the viscous liquid through small-bore orifices. A suitable viscosity can be achieved either by heating or dissolution. The rheological properties of the solution are governed to a large degree by the solvents selected. Wastes generated during the spinning operation include evaporated solvent and wastewater contaminated by solvent. The typical solvents used in the production of fibers are dimethylacetamide (acrylic), acetone or chlorinated hydrocarbon (cellulose acetate), and carbon disulfide (rayon). In the dry spinning process a solution of polymer is first prepared. The solution is then heated above the boiling temperature of the solvent and the solution is extruded through spinneret. The solvent evaporates into the gas stream. With wet spinning the fiber is directly extruded into a coagulation bath where solvent diffuses into the bath liquid and the coagulant diffuses into the fiber. The fiber is washed free of solvent by passing it through an additional bath. Each process step generates emissions or wastewater. Solvents used in production are normally recovered by distillation. Figure 14.22.2 is a schematic diagram of fiber production showing that almost all stages of production generate emissions.
1018
George Wypych
Figure 14.22.2. Schematic diagram of emissions from the man-made fiber manufacturing industry. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995.]
Tables 14.22.1 and 14.22.2 provide data on releases and transfers from both polymer manufacturing and man-made fiber production in the USA. Carbon disulfide, methanol, xylene, and ethylene glycol are used in the largest quantities. Carbon disulfide is used in manufacture of regenerated cellulose and rayon. Ethylene glycol is used in the manufacture of polyethylene terephthalate, the manufacture of alkyd resins, and as cosolvent for cellulose ethers and esters. Methanol is used in several processes, the largest being in the production of polyester. This industry is the 10th largest contributor of VOC and 7th largest in releases and transfers. There have been many initiatives to reduce emissions and usage of solvents. Man-made fiber manufacturing no longer uses benzene. DuPont eliminated o-xylene and reduced methanol and ethylene glycol use in its Wilmington operation. This change resulted in annual savings of $1 million. Process modification in a polymer processing plant resulted in a decrease in total emissions of 74% and a reduction in the release of cyclohexane by 96%. Monitoring of thousands of valves in Eastman Texas plant resulted in a program of valve replacement which eliminated 99% of the emissions. Plant in Florida eliminated solvents from cleaning and degreasing. These examples show that in many cases pollution can be reduced by better equipment, organization, and care.
14.22 Polymers and man-made fibers
1019
Table 14.22.1 Reported solvent releases from the polymer and man-made fiber industry in 1995 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
allyl alcohol
29,000
1,4-dioxane
10,000
benzene
60,000
ethylbenzene
130,000
n-butyl alcohol
480,000
ethylene glycol
1,400,000
sec-butyl alcohol
25,000
hexane
880,000
tert-butyl alcohol
16,000
methanol
3,600,000
carbon disulfide
27,500,000
methyl ethyl ketone
260,000
carbon tetrachloride
100
methyl isobutyl ketone
98,000
chlorobenzene
19,000
pyridine
67,000
chloroform
14,000
tetrachloroethylene
4,000
cresol
4,000
1,1,1-trichloroethane
120,000
cyclohexane
98,000
trichloroethylene
39,000
1,2-dichloroethane
98,000
1,2,4-trimethylbenzene
12,000
dichloromethane
1,300,000
toluene
900,000
N,N-dimethylformamide
19,000
xylene
460,000
Table 14.22.2. Reported solvent transfers from the polymer and man-made fiber industry in 1995 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
allyl alcohol
120,000
ethylbenzene
880,000
benzene
160,000
ethylene glycol
49,000,000
n-butyl alcohol
330,000
hexane
8,000,000
sec-butyl alcohol
12,000
methanol
5,600,000
tert-butyl alcohol
160,000
methyl ethyl ketone
460,000
carbon disulfide
14,000
methyl isobutyl ketone
43,000
carbon tetrachloride
200,000
N-methyl-2-pyrrolidone
780,000
chlorobenzene
570,000
pyridine
70,000
chloroform
59,000
tetrachloroethylene
330,000
cresol
20,000
1,1,1-trichloroethane
21,000
cyclohexane
420,000
trichloroethylene
76,000
dichloromethane
250,000
1,2,4-trimethylbenzene
98,000
N,N-dimethylformamide
300,000
toluene
2,800,000
1,4-dioxane
11,000
xylene
7,800,000
1020
George Wypych
New technology is emerging to reduce solvent use. Recent inventions disclose that, in addition to reducing solvents, the stability of ethylene polymers can be improved with the new developed process.3 A proper selection of solvent improved a stripping operation and contributed to the better quality of cyclic esters used as monomers.4 Solvent was used for the recovery of fine particles of polymer which were contaminating water.5 A new process for producing fiber for cigarette filters uses reduced amounts of solvent.6 Optical fibers are manufactured by radiation curing which eliminates solvents.7 A new electrospinning process has been developed which produces unique fibers by the dry spinning method, providing a simpler separation and regeneration of the solvent.8
REFERENCES 1 2 3 4 5 6 7 8
EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. M M Hughes, M E Rowland. C A Strait, US Patent 5,756,659, The Dow Chemical Company, 1998. D W Verser, A Cheung, T J Eggeman, W A Evanko, K H Schilling, M Meiser, A E Allen, M E Hillman, G E Cremeans, E S Lipinsky, US Patent 5,750,732, Chronopol, Inc., 1998. H Dallmeyer, US Patent 5,407,974, Polysar Rubber Corporation, 1995. J N Cannon, US Patent 5,512,230, Eastman Chemical Company, 1996. P J Shustack, US Patent 5,527,835, Borden, Inc., 1996. A E Zachariades, R S Porter, J Doshi, G Srinivasan, D H Reneker, Polym. News, 20, No.7, 206-7 (1995).
14.23 PRINTING INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The number of printing and publishing operations in the US is estimated at over 100,000. 1.5 million people are employed. The value of shipments is over $135 billion. 97% of printing is done by lithography, gravure, flexography, letterpress, and screen printing on substrates such as paper, plastic metal, and ceramic. Although, these processes differ, the common feature is the use of cleaning solvents in imaging, platemaking, printing, and finishing operation. Most inks contain solvents and many of the adhesives used in finishing operations also contain solvents. Many processes use the so-called fountain solutions which are applied to enable the non-image area of the printing plate to repel ink. These solutions contain primarily isopropyl alcohol. But the printing operation is, by itself, the largest contributor of VOCs. Each printing process requires inks which differ drastically in rheology. For example, gravure printing requires low viscosity inks which contain a higher solvent concentrations. Tables 14.23.1 and 14.23.2 provide data on the reported releases and transfers of solvents by the US printing industry. These data show that there are fewer solvents and relatively low releases and transfers compared with other industries. In terms of VOC contribution, the printing industry is 5th and 10th in the total emissions and transfers. Current literature shows that there is extensive activity within and outside industry to limit VOCs and reduce emissions. Cleaning operations are the major influence on emissions. Shell has developed a new cleaning formulations containing no aromatic or chlori-
14.23 Printing industry
1021
nated hydrocarbons.3 An additional requirement was to optimize the solvent mixture to prevent swelling of the rubber in blanket cylinders and rollers. It is predicted that the European industry will increase rate of the introduction of radiation-cured inks and eliminate isopropanol from fountain solutions.4 It is expected that radiation-cured flexographic inks will grow by 30%/year in the next five years.5 In Germany, 70-80% of emissions or 47,000 ton/year will have to be eliminated by the year 2007.6 Beginning in 1999, the UK industry must keep VOC concentration below 5 tonnes/year per plant.7 VOC concentration in outside atmosphere must not exceed 150 mg/m3 (50 mg/m3 if there is more than 5% aromatic solvents). Reactive hot melts are being used in book binding.8 This will eliminate emissions from currently used solvent adhesives. Solvent replacements are not the only solution at hand. Solvent-containing systems often give better quality than replacement systems, therefore methods have been developed to make the solvent based materials more acceptable. A soil bed biofiltration system was tested in California with excellent results.9 This biofilter is a bed of soil impregnated with microorganisms which use VOC as their food. Present California regulations require that such a treatment system has a 67% capture and VOC destruction efficiency. The new method was proven to have 95.8% efficiency. In addition to environmental issues with solvents, the printing industry has addressed the source of their raw materials. Present systems are based on petroleum products which are not considered renewable resources. Terpenes are natural products which are now finding applications in the print industry.10 In Denmark, of 70% cleaning solvents are vegetable oil based. These and other such innovations will continue to be applied to reduce solvent use and emissions. It is also reported10 that water-based system replaces fountain solutions. Other factors are driving changes. Odors in packaging materials and the migration of solvent to foods are unacceptable. Most odors in packaging materials are associated with process and coalescing solvents.11 Foods which do not contain fat are more susceptible to the retaining the taste of solvents. Printing inks which may be acceptable for foods containing fat may not be suitable for fat-free applications (see more on this subject in Chapter 16.1).12 Many recent inventions have also been directed at solving the current environmental problems of printing industry.13-21 The solvent in gravure printing inks not only contribute to pollution but also to the cost of solvent recovery and/or degradation. A new technology is proposed in which a solvent free ink with a low melting point can be processed in liquid state and then be solidified on cooling.13 A non-volatile solvent for printing inks was developed based on a cyclic keto-enol tautomer and a drying oil.14 An alcohol soluble polyamide for rotary letterpress printing inks was developed15 and subsequently adapted to flexographic/gravure inks.18 A polyamide was also used in a rotary letterpress ink which enabled low alcohols to be used as the solvent with some addition of an ester.16 This new ink is compatible with water-based primers and adhesives which could not be used with solvent-based inks. Inks for jet printers are water sensitive. One solvent-based technology was developed using esters and glycols17 and the other using low alcohols.20 Another recent invention describes aqueous ink containing some low alcohols.19 UV and electron beam cured ink concentrates were also developed.21 This information from open and patent literature clearly indicates that industry is actively working on the development of new technological processes to reduce emissions of solvents.
1022
George Wypych
Table 14.23.1. Reported solvent releases from the printing and publishing industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
n-butyl alcohol
43,000
methyl isobutyl ketone
170,000
dichloromethane
59,000
N-methyl-2-pyrrolidone
31,000
1,4-dioxane
8,000
tetrachloroethylene
34,000
ethylene glycol
46,000
1,1,1-trichloroethane
180,000
ethylbenzene
23,000
trichloroethylene
13,000
hexane
50,000
1,2,4-trimethylbenzene
36,000
isopropyl alcohol
27,000
toluene
12,200,000
methanol
170,000
xylene
700,000
methyl ethyl ketone
960,000
Table 14.23.2. Reported solvent transfers from the printing and publishing industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
n-butyl alcohol
7,000
methyl isobutyl ketone
63,000
dichloromethane
43,000
N-methyl-2-pyrrolidone
28,000
1,4-dioxane
340
tetrachloroethylene
27,000
ethylene glycol
16,000
1,1,1-trichloroethane
39,000
ethylbenzene
9,200
trichloroethylene
4,000
hexane
12,000
1,2,4-trimethylbenzene
33,000
isopropyl alcohol
12,000
toluene
2,800,000
methanol
17,000
xylene
240,000
methyl ethyl ketone
700,000
REFERENCES 1 2 3 4 5 6 7
EPA Office of Compliance Sector Notebook Project. Profile of the Printing and Publishing Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. N C M Beers, M J C M Koppes, L A M Rupert, Pigment & Resin Technol., 27, No.5, 289-97 (1998). D Blanchard, Surface Coatings International, 80, No.10, 476-8 (1997). B Gain, Chem. Week, 160, No.14, 28-30 (1998). W Fleck, Coating, 31(1), 23-25 (1998). C H Williams, Converter, 34, No.9, 11-2 (1997).
14.24 Pulp and paper
8 9 10 11 12 13 14 15 16 17 18 19 20 21
1023
Hughes F, TAPPI 1997 Hot Melt Symposium. Conference Proceedings. TAPPI. Hilton Head, SC, 15th-18th June 1997, p.15-21. A Mykytiuk, Paper, Film & Foil Converter, 72, No.8, 120-3 (1998). A Harris, Paper, Film & Foil Converter, 72, No.5, 198-9 (1998). R M Podhajny, Paper, Film & Foil Converter, 72, No.12, 24 (1998). T Clark, Paper, Film & Foil Converter, 70, No.11, 48-50 (1996). R Griebel, K A Kocherscheid, K Stammen, US Patent 5,496,879, Siegwerk Druckfarben GmbH, 1996. D Westerhoff, US Patent 5,506,294, 1996. P D Whyzmuzis, K Breindel, R A Lovald, US Patent 5,523,335, Henkel Corporation, 1996. R J Catena, M C Mathew, S E Barreto, N Marinelli, US Patent 5,658,968, Sun Chemical Corporation, 1997. J M Kruse, US Patent 5,663,217, XAAR Ltd., 1997 P D Whymusis, US Patent 5,714,526, Henkel Corporation, 1998. H Yanagi, S Wakabayashi, K Kaida, US Patent 5,736,606, Kao Corporation, 1998. M Shinozuka, Y Miyazawa, M Fujino, T Ito, O Ishibashi, US Patent 5,750,592, Seiko Epson Corporation, 1998. W R Likavec, C R Bradley, US Patent 5,866,628, Day-Glo Color Corporation, 1999.
14.24 PULP AND PAPER George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The US pulp and paper industry operates over 550 facilities which employ over 200,000 people. Total shipments are $60 billion with an additional $80 billion in converted products. Several processes contribute to the emission of solvents. These include chemical pulping kraft process (terpenes, alcohols, methanol, acetone, chloroform), bleaching (acetone, dichloromethane, chloroform, methyl ethyl ketone, carbon disulfide, chloromethane, and trichloroethane), wastewater treatment (terpenes, alcohols, methanol, acetone, chloroform and methyl ethyl ketone), and evaporators in chemical recovery systems (alcohols and terpenes).
Table 14.24.1. Reported solvent releases from the pulp and paper industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
320,000
methanol
63,000,000
n-butyl alcohol
46,000
methyl ethyl ketone
700,000
chloroform
4,500,000
methyl isobutyl ketone
10,000
chloromethane
260,000
1,2,4-trimethylbenzene
17,000
cresol
410,000
toluene
580,000
ethylbenzene
22,000
xylene
49,000
ethylene glycol
37,000
o-xylene
260
hexane
150,000
1024
George Wypych
Table 14.24.2. Reported solvent transfers from the pulp and paper industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
24,000
hexane
8,600
n-butyl alcohol
16,000
methanol
23,000,000
chloroform
150,000
methyl ethyl ketone
36,000
chloromethane
120
1,2,4-trimethylbenzene
1,400
cresol
3,600
toluene
23,000
ethylene glycol
190,000
xylene
4,000
Table 14.25.1. Reported solvent releases from the rubber and plastics industry in 1995 [Data from Ref. 2] Solvent benzene n-butyl alcohol sec-butyl alcohol tert-butyl alcohol carbon disulfide chlorobenzene chloroform chloromethane cresol cyclohexane dichloromethane N,N-dimethylformamide 1,4-dioxane ethylbenzene
Amount, kg/year 5,800 380,000 17,000 240 5,500,000 5,000 46,000 47,000 9,000 480,000 11,700,000 350,000 2,600 210,000
Solvent ethylene glycol hexane isopropyl alcohol methanol methyl ethyl ketone methyl isobutyl ketone N-methyl-2-pyrrolidone tetrachloroethylene 1,1,1-trichloroethane trichloroethylene 1,2,4-trimethylbenzene toluene xylene m-xylene
Amount, kg/year 120,000 1,700,000 28,000 4,000,000 5,500,000 1,100,000 32,000 160,000 3,000,000 660,000 260,000 7,600,000 2,200,000 6,000
Tables 14.24.1 and 14.24.2 give the reported releases and transfers of solvent data for the US pulp and paper industry. If not for the emissions of methanol and chloroform the industry would be a much less serious polluter. It is 7th in VOC contributions and 8th in total releases and transfers.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Pulp and Paper Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.25 Rubber and plastics
1025
Table 14.25.2. Reported solvent transfers from the rubber and plastics industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
benzene
15,000
hexane
50,000
n-butyl alcohol
370,000
isopropyl alcohol
14,000
sec-butyl alcohol
1,100
methanol
1,400,000
tert-butyl alcohol
85,000
methyl ethyl ketone
3,500,000
carbon disulfide
150,000
methyl isobutyl ketone
450,000
chloroform
1,200
N-methyl-2-pyrrolidone
120,000
chloromethane
330
tetrachloroethylene
47,000
cresol
3,000
1,1,1-trichloroethane
160,000
cyclohexane
350,000
trichloroethylene
170,000
dichloromethane
900,000
1,2,4-trimethylbenzene
14,000
N,N-dimethylformamide
570,000
toluene
2,200,000
1,4-dioxane
49,000
xylene
940,000
ethylbenzene
350,000
m-xylene
5,700
ethylene glycol
15,300,000
14.25 RUBBER AND PLASTICS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The US rubber and plastics industry employs over 800,000 people and operates over 12,000 plants. Its total production output is estimated at over $90 billion. The industry produces a wide diversity of products some of which do not contain solvents but many of which require the use of process solvents. Solvents are contained in adhesives used in finishing operations. Large quantities of solvents are used for surface cleaning and cleaning of equipment. Tables 14.25.1 and 14.25.2 provide data on the reported releases and transfers of solvents by the US rubber and plastics industry. These industries contribute small amounts of VOC which are in the range of 0.00001-0.00005 kg VOC/kg of processed rubber. It was the ninth largest contributor to releases and transfers of all US industries. Dichloromethane, toluene, carbon disulfide, methyl ethyl ketone, methanol, 1,1,1-trichloroethane, hexane, methyl isobutyl ketone, and xylene are emitted in very large quantities.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Rubber and Plastics Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
1026
Mohamed Serageldin, Dave Reeves
14.26 USE OF SOLVENTS IN THE SHIPBUILDING AND SHIP REPAIR INDUSTRY Mohamed Serageldin U.S. Environmental Protection Agency, Research Triangle Park, NC, USA
Dave Reeves Midwest Research Institute, Cary, NC, USA
14.26.1 INTRODUCTION The focus of this chapter will be on the use of solvents in the shipbuilding and ship repair industry. This industrial sector is involved in building, repairing, repainting, converting, or alteration of marine and fresh water vessels. These vessels include self-propelled vessels, those propelled by other vessels (barges), military and Coast Guard vessels, commercial cargo and passenger vessels, patrol and pilot boats, and dredges. The industry sector is also involved in repairing and coating navigational aids such as buoys. This chapter begins with an overview of operations in a typical shipbuilding and/or ship repair facility (shipyard), to identify those operations that generate significant volatile organic compound (VOC) emissions and/or hazardous air pollutant (HAP) emissions from the use of organic solvents. Organic solvents that are VOCs contribute to formation of ozone in the troposphere. Other organic solvents such as chlorinated fluorocarbons (CFCs) cause depletion of the ozone layer in the stratosphere. Therefore, VOCs and other air toxics, such as those compounds listed as HAPs, are both indirectly and directly detrimental to the general public’s health. Because many solvents are VOCs and often contain large amounts of HAPs, many state agencies1,2 and the United States Environmental Protection Agency (U.S. EPA) have issued regulations to limit their content in materials used for surface coating and cleaning operations at shipyards.3-7 14.26.2 SHIPBUILDING AND SHIP REPAIR OPERATIONS Most facilities engaged in shipbuilding or ship repair activities (shipyards) have several manufacturing areas in common, each including one or more “unit operations”. These areas include: (a) surface preparation of primarily steel surfaces, which may include cleaning with multiple organic solvents; (b) assembly operations, which involve assembly of blocks that were constructed from sub-assembled parts (this step involves steel cutting and material movement using heavy equipment such as cranes); (c) cleaning operations (other than surface preparation) such as equipment and parts cleaning; and (d) coating operations.8,9 There are secondary operations such as chrome plating, asbestos removal, fuel combustion, carpentry, and, to various degrees, polyester lay-up operations (composite materials construction activities). We will next discuss those operations that involve the use of organic cleaning solvents. 14.26.3 COATING OPERATIONS Marine coatings can be applied by the use of spraying equipment, brushes, or rollers. Coating operations at shipyards are typically conducted at two primary locations: (1) outdoor work areas or (2) indoor spray booths. The outdoor work areas can include ship exteriors and interiors. Most shipyards report that typically only a small percentage (10%) of the coating operations are done indoors. However, in large construction yards a larger propor-
14.26 Use of solvents in the shipbuilding
1027
tion (up to 30 %) of the coatings are applied indoors.10 Coating and cleaning operations constitute the major source of VOC and HAP emissions from shipyards. If the metal surface is not well prepared before a coating is applied or if the coating is applied at the wrong ambient conditions, the coating system may fail and the work may have to be redone. The amount of cleaning necessary will depend on the type and extent of the problem and the coating system that is being used. 14.26.4 CLEANING OPERATIONS USING ORGANIC SOLVENTS In most industrial applications involving metal substrates, organic cleaning solvents are used to remove contaminants or undesirable materials from surfaces before a coating is applied to clean equipment and parts utilized to apply the coating or soiled during that operation. Solvents are used for general maintenance of equipment parts. These surfaces are typically made of steel. However, vessels may also be made from natural materials such as wood and synthetic materials such as fiberglass. Therefore, a solvent must be selected that will not attack the substrate being cleaned. For material accounting purposes, we can classify cleaning (unit) operations as follows:11,12 1. Surface preparation of large manufactured components (stage before a coating is applied). 2. Surface preparation of small manufactured components (stage before a coating is applied). 3. Line cleaning (includes piping network and any associated tanks). 4. Gun cleaning (manually or in a machine). 5. Spray booth cleaning (walls and floor). 6. Tank cleaning (mostly inner tank surfaces and any associated pipes). 7. Parts (machine) cleaning (simple dip tanks and large machines). 8. Cleaning of equipment and other items (e.g., bearings, buckets, brushes, contact switches). 9. Floor cleaning (organic solvents are no longer used). These categories are similar to those found in other industries involved in the application of surface coating. However, the number of cleaning categories varies from one industry to another. For example, the automotive manufacturing industry (SIC code 3711) and the furniture industry are involved to various degrees in all nine types of cleaning operations. On the other hand, the photographic supplies (chemicals) industry will not include the first three listed cleaning operations.11 14.26.4.1 Surface preparation and initial corrosion protection Large manufactured ship components are often cleaned with an organic solvent as the first of a number of cleaning steps that are required before a coating is applied. The method of surface preparation is selected to work with a chosen coating system. Surface preparation may include application of chemicals such as etching agents, organic solvents cleaners, and alkaline cleaners. Organic solvents such as mineral spirits, chlorinated solvents, and coal tar solvents are used to remove unwanted materials such as oil and grease.13 If a ship is being repaired, existing coatings usually need to be removed. Solvents such as dichloromethane are commonly used for removing (stripping off) old or damaged coatings. However, aqueous systems involving caustic compounds are now being used more frequently for such purposes.14 Pressure washing and hydro blasting are other cleaning techniques used. But, the
1028
Mohamed Serageldin, Dave Reeves
predominant method is still particulate blasting (using abrasive media), which is used to remove mil scale, extra weld material, rust, and old coatings. The angle at which the surface is blasted is chosen to generate the desired peaks and valleys on the substrate, that will accommodate the viscosity, chemistry (polar groups) of the primer coating. The surface profiling will also help the primer coating adhere mechanically to the substrate, contributing to the longevity of the coating system.15 Pre-construction primers are sometimes used immediately following surface preparation (blasting) to prevent steel from oxidizing (rusting). This primer is removed by particulate blasting, before the protective coating system (one or more coatings) is applied to the assembled parts or blocks. Removal of such primers (when they cannot be welded-through) can result in emissions of VOCs and HAPs. 14.26.4.2 Cleaning operations after coatings are applied Surface coating operations at shipyards use predominantly solvent-based coatings. Hence, relatively large amounts of organic solvents are used for cleaning and thinning activities. Table 14.26.1 shows the most common organic solvents used for thinning and cleaning, based on 1992 data.16 Table 14.26.2 gives examples of solvent products that can be used for both thinning coatings and for cleaning surfaces after coatings are applied and for maintenance cleaning. The solvent products are listed in decreasing order of evaporative rate. Acetone, a ketone solvent is commonly used for cleaning and thinning polyester resins and gel coats. However, it is also used in formulating low-VOC and low-HAP products. Methyl ethyl ketone (MEK) and methyl isobutyl ketone (MIBK) are fast evaporative solvents that are used for thinning and cleaning vinyl coatings, epoxy coatings, and many other high performance coatings. Fast evaporative coatings that can improve application properties for a good finish may also be formulated by blending different solvents. Examples are shown in Table 14.26.2. The fast evaporative mix includes solvents varying in polarity and solubility parameters. They include an oxygenated solvent (MIBK), aromatic hydrocarbon solvents that contain less than 10 percent (by mass) HAPs, and aromatic hydrocarbons like xylene that are 100 percent HAPs as will be shown later. Together they produce the correct solvency for the polymer (resin). Table 14.26.1. Predominant solvents used in marine coatings [from ref. 16 ] and EPA regulatory classifications Organic solvent
VOC
HAP, Sec. 112 (d)
Toxic chemicals, Sec. 313
Butyl alcohol
Y
Y
Y
Ethyl alcohol
Y
N
N
Isopropyl alcohol
Y
N
Ya
Xylene
Y
Y
Y
Toluene
Y
Y
Y
Ethyl benzene
Y
Y
Y
ALCOHOLS
AROMATICS
14.26 Use of solvents in the shipbuilding
Organic solvent
1029
VOC
HAP, Sec. 112 (d)
Toxic chemicals, Sec. 313
Ethylene glycol ethers
Y
Y
Y
Propylene glycol ethers
Y
N
N
Acetone
N
N
Y
Methyl ethyl ketone
Y
Y
Y
Methyl isobutyl ketone
Y
Y
Y
Methyl amyl ketone
Y
N
N
Y
Yb
N
High-flash naphtha
Y
Y
b
Y
n-Hexane
Y
Y
N
ETHERS
KETONES
PARAFFINIC Mineral spirits
VOC = volatile organic compound; HAP = Hazardous air pollutant; Sec 313 of the Emergency Right-to-know Act (EPCRA), also known as Title III of the Superfund Amendments and Reauthorization Act of 1986 (40 CFR Part 372). aUse of strong acid process, no supplier notification. bLigroine (light naphtha), VM&P naphtha, Stoddard solvent, and certain paint thinners are also commonly referred to as mineral spirits. These distillation fractions contain less than 10 % by mass HAPs (see Table 14.26.4).
Vapor press., mmHg @ 20oC
Surface tension, dynes/cm 20oC
Av. sp. gr, @ 25oC
Viscosity, cP @ 25oC
6.1
186
56
27.1
0.787
0.31
MEK*
Vinyl
9.3
4.0
70
80
24.2
0.806
0.43
MIBK**, 24 %
8.58
1.7
28
116
23.3
0.796
0.54
N-butyl alc., 24%
11.6
0.44
5.5
118
23.4
0.806
2.62
Toluene, 52%
8.93
2.0
22
111
28.2
0.863
0.57
MIBK, 23 %
8.58
1.7
28
116
23.3
0.796
0.54
EGBE***, 26%
10.2
0.072
0.6
>169
26.9
0.899
3.0
AHC****, 30%
7.7
0.16
2.0
>160
23.4
0.775
0.88
1,2,4-Trimethylbenzene, 16%
8.9
19
2.1
168
30.2
0.871
0.94
Xylene (mixed), 2%
9.9
0.77
6.0
>135
27.6
0.856
0.63
Spraying thinner & solvent Epoxy b cleaner (fast)
Brushing thinner & solvent cleaner b (medium)
Epoxy
MEK, 100 % (approx.)
B.P., oC @ 760 mmHg
Relative rate, nBUOAc =1.0
9.8
Compound, wt%
Polyester Acetone, 100% (approx.)
Typical coating
Acetone
Thinner & cleaning solvents
Solubility parameter, (cal/cm3)1/2
Table 14.26.2. Selected products that are used as both solvent thinners and solvent cleaners
B.P., oC @ 760 mmHg
Surface tension, dynes/cm 20oC
Av. sp. gr, @ 25oC
0.072
0.6
>169
26.9
0.899
3.0
AHC, 30%
7.7
0.16
2.0
>160
23.4
0.775
0.88
1,2,4-Trimethylbenzene, 16%
8.9
19
2.1
168
30.2
0.871
0.94
Xylene (mixed), 1%
9.9
0.77
6.0
>135
27.6
0.856
0.63
Relative rate, nBUOAc =1.0
10.2
Solubility parameter, (cal/cm3)1/2
EGBE, 51%
Compound, wt%
Vapor press., mmHg @ 20oC
Lacquer retarder (thinLacquer ner) & cleaner b (slow)
Viscosity, cP @ 25oC
Mohamed Serageldin, Dave Reeves
Typical coating
Thinner & cleaning solvents
1030
*MEK=methyl ethyl ketone, **MIBK=methyl isobutyl ketone, ***EGBE=ethylene glycol monobutyl ether, ****AHC=aromatic hydrocarbon solvent; aPhysical properties mainly from Industrial Solvents Handbook, 110 -114. bMaterial Safety Data Sheet (Mobile Paint Co. Alabama)
The lower specific gravity of ketones (see Table 14.26.2) than other materials such as glycol ethers helps reduce total mass of VOCs (or HAPs) per volume of nonvolatiles (solids) in a container of coating. Glycol ethers are good solvents for epoxies and acrylics. They also have good coupling abilities in blends of poorly miscible solvents17 and have low evaporative rates. The properties of a solvent product are dependent on the chemical structure and distillation range of the solvent mix in the product. The latter will affect the evaporative rate from a coating or cleaner, affecting the solubility of the resin in the coating and viscosity of the coating and solvent cleaner. Therefore, the viscosity of the solvent product must be close to that of the resin in a coating.18 The surface tension of a solvent provides a measure of the penetrability of a cleaning solvent. A low surface tension also means the solvent spreads more readily, which is an important property for a cleaning product. However, several properties in Table 14.26.2 come into play in determining the effectiveness of a cleaning solvent. Most coating operations, due to the size and accessibility of ships, occur in open air in drydocks, graving docks, railway, or other locations throughout a facility. Because of the size of ships, the predominant application method is airless spray guns. The thickness of the coating will determine if the application equipment needs to be cleaned during application of the coating or after the job is completed. The lines from the supply tanks to the spray gun may in some instances exceed 46 m (150 ft) in length. The ensemble of equipment and items that have to do with the application of the coating or “unit operation system (UOS)” is shown schematically in Figure 14.26.1. The representation depicts a layout for outdoor application of coatings. It includes the container used to hold the coating, attached feed pump, line transferring the coating to the spray gun, the spray gun itself, and any other item soiled with a coating that will need to be cleaned with organic solvent before it can be reused. The need and frequency for cleaning will depend on the individual facility or company cleanliness standard (i.e., requirements) and the number of coating formulation or color changes. Cleaning of spray guns, internal transfer lines, and associated tanks account for a large part of organic solvent usage. At most shipyards, a small percentage of the coatings are applied indoors, in spray-booths. The walls and floors of these booths are cleaned by wiping with a solvent laden cloth. The coating application equipment UOS for most facilities will look very similar to that shown in Figure 14.26.1, except that the coating transfer
14.26 Use of solvents in the shipbuilding
1031
lines will be shorter if the coating storage tanks are positioned close to the spray booths. The transfer lines, that will need to be cleaned with solvent, will be longer if the coating tanks are located away from the application area. When this is the case, the transfer lines typically run underground at the facility and another representation than the one shown in Figure 14.26.1 will need to be used, to clearly identify the emission points and waste streams for properly quantifying solvent losses. The latter may include a unit for recycling or reclaiming solvents. Spray gun cleaning procedures may be a once-through type with collection of spent solvent in a container for disposal or reuse. Some facilities use commercial gun washers. Because gun washers are enclosed and recirculate solvent, they can reduce the amount of solvent lost by evaporation. In either case, the emissions are calculated as the difference between the amount used and the amount recovered. To calculate the emissions associated with cleaning a spray gun it is recommended that a material balance around a “unit operation system” be considered. Several examples are provided in the Alternative Control Techniques (ACT) document on industrial cleaning solvents.19 Several types of part cleaners are used at shipbuilding and ship repair facilities. The types used in such facilities vary from the more simple sink and spray systems20 to more elaborate parts (machine) cleaners of the cold or vapor types.4 Most of the parts cleaners in shipyards are small — around 1.5 m x 1 m and 1 m deep - usually located in the machine shops, not the paint rooms. Most of the parts are small components being cleaned prior to being joined to other small parts into assemblies and sub-assemblies or being cleaned as part of some type of repair operation. Most of parts cleaners used were basket-type design with the parts loaded into a basket and dropped through the vapor zone several times to clean off the oils and dirt. Some shipyards use contractors to come in and change out the solvent on a routine schedule. 14.26.4.3 Maintenance cleaning of equipment items and components Shipyards also undertake scheduled maintenance cleaning of many ship components such as contacts and switches and equipment items such as bearings and packaging machines. This is mostly done by hand-wiping the parts with organic solvents. These operations will generally consume a relatively small amount of the overall volume of organic solvents used for cleaning in shipyards. Solvents are also used in machine shop areas and thus contribute to the waste stream. 14.26.5 MARINE COATINGS There are several categories of marine coatings that are used to protect the surface of a ship from the aggressive marine environment Figure 14.26.1 Schematic diagram of marine coating application equipment.
1032
Mohamed Serageldin, Dave Reeves
and for other performance requirements such as preventing corrosion and fouling; protecting cargoes from contamination; providing safety warnings and informational markings; providing cosmetic and camouflage colors; preventing slipping and sliding on walking surfaces; reducing fire hazards; and providing cathodic protection.21 The coating systems of marine coatings are selected to meet: • the type of marine environment to which a vessel will be exposed • the time a vessel is to remain operational before it needs to be reworked. General areas of a ship include: (1) underwater hull, (2)superstructures and freeboard, (3) interior habitability areas, (4) exterior deck areas, and (5) fuel, water ballast and cargo tank.22 The freeboard is the area above water hull. These areas have different characteristics and operational requirements. Table 14.26.323 shows the predominant resin and solvent types used on ships based on a 1991/1992 survey of the industry obtained as part of the shipbuilding and ship repair regulation was being developed. The summary table also gives average VOC and HAP content for the various coating category types. Epoxy coatings constitute a large percentage of the coatings used. The epoxy films are strongly resistant to most chemicals and are very good anti-corrosion coatings, and require little surface preparation. Table 14.26.3. Summary of marine coating usage (by coating type)23 Coating types
Average usage in U.S. shipyards, %
Average VOC content, g/L (lb/gal)
Average HAP content, g/L (lb/gal)
General use types alkyd based
10
474
(3.95)
355
(2.98)
epoxy based
59
350
(2.92)
56
(0.47)
antifouling (multiple resins)1
11
388
(3.23)
268
(2.25)
inorganic zinc based
10
545
(4.54)
274
(2.30)
other speciality categories
10
400
(3.33)
144
(1.20)
TOTAL
100
Speciality types
1
resins: epoxy, polyurethane, vinyl, and chlorinated rubber
The coating system used will depend on service requirements. Maximum protection at an economical price can be achieved when the user understands the protection needed and the functions performed by the coatings. Coatings are designed for spray viscosity, drying time, pot-life, and cure profile; all of these parameters affect shelf stability.24 The physical parameters and properties of a coating are affected by the volatile constituents (mainly organic solvents) in a coating, some of which are VOCs, HAPs, ozone depleters, and SARA 313 toxic chemicals that need to be reported.25 14.26.6 THINNING OF MARINE COATINGS The sprayability of a coating is determined by its viscosity at application. The viscosity is a measure of the ability of a material to resist flow and is an important formulation design parameter. Application viscosity is affected by the ambient conditions and by the degree of
14.26 Use of solvents in the shipbuilding
1033
mixing that occurs within the applicator. The thinner is often the same material as the cleaner (as indicated in Table 14.26.2). The solvent material is often a blend of miscible materials. Together they will dissolve a dry resin that needs to be removed or give the coating certain needed properties such as reduced/increased viscosity or shorter drying times. Standard spraying equipment will apply coatings up to some maximum viscosity. Above that maximum value, thinning solvents are required. Thinning solvent is sometimes added to enhance brushability or sprayability of a coating. The appropriate viscosity is provided by the coating manufacturer or supplier; it will depend on the solvent content of the coating and temperature at the point of application.26 Since most coatings are applied outdoors, extreme weather conditions may require adding thinning solvents to the coating. Organic thinning solvents are added to coatings to alter their flowing properties. However, the flow properties of a coating may be altered by using special heaters or a combination of solvent and heat. The effect of a heater on the viscosity of a coating depends on the physical properties of the coating and on the flow rate in the in-line heater. Under cold weather conditions, in-line heaters may provide good viscosity control, but may not be able to solve all application problems that are encountered in the field. Under extremely low temperatures, the substrate surface can act as a heat sink, which may inhibit the setting or curing of the coating. In-line heaters which are used for low volume coatings are not suitable for large volume coatings. As a result, thinning solvents are still needed to transfer the fluid from storage to pumps and hoses. Under hot and humid weather conditions, certain coatings (e.g., lacquers) can rapidly lose organic solvent prior to and during application. Often under these situations a facility will add solvent blends to make up for the reduction in viscosity and to overcome condensation on the surface (blushing).27 Evaporative losses can be minimized by adopting good work practices and by using formulations that contain organic solvents with low vapor pressures. 14.26.7 SOLVENT EMISSIONS Several states with their own rules regulating marine coatings have separate rules addressing solvent cleaning operations. While marine coating rules typically address VOC contents and types of application equipment, the cleaning solvent rules are more generic and address cleaning solvents used at any and all metal-related manufacturing operations. Many types of solvents are used in marine coatings and in their associated cleaning materials as shown in Table 14.26.1. Almost all solvents used at shipyards are VOC and approximately one in three solvents contain HAPs. Of the HAPs reported, several are included on the list of 17 high priority chemicals targeted by U.S. EPA for the 33/50 program.28 These included xylene (commercial), toluene, and the ketones. Commercial grade xylene represents the major portion of the volatile HAPs reported. Many of the commonly known solvents are actually petroleum distillation fractions and are composed of a number of compounds (e.g., mineral spirits and naphthas). There are two general types of solvents derived from petroleum, aliphatics or aromatics. Aromatics are stronger solvents than aliphatics since they dissolve a wider variety of resins. Most major solvent suppliers (chemical manufacturers) produce several types and variations of these solvents and the associated HAP contents can vary significantly from manufacturer to manufacturer and from batch to batch. These types of solvents are used extensively and are present in the majority of marine coatings. Table 14.26.4 provides a summary of common petroleum distillate solvents and solvent blends and their associated HAP content. For any
1034
Mohamed Serageldin, Dave Reeves
solvent or solvent blend that is not listed as specified in Table 14.26.4, another table (Table 14.26.5) was developed to provide solvent groupings and associated HAP component/content values. The HAP values for Tables 14.26.4 and 14.26.5 were adapted from estimates provided in 1998 by the Chemical Manufacturer Association’s Solvent’s Council. Table 14.26.4. HAP content of single solvents and solvent blends. [Adapted, by permission from Chemical Manufacturer Association’s Solvent’s Council] Solvent/solvent blend
CAS No.
HAP content range, wt%
Average HAP content, wt%
Typical HAP, wt%
Toluene
108-88-3
100
100
toluene
Xylene(s)
1330-20-7
100
100
xylenes, ethylbenzenea
Hexane
110-54-3
49-55
50
n-hexaneb
Ethylbenzene
100-41-4
100
100
ethylbenzene
Aliphatic 140
0
0
none
Aromatic 100
<5
2
1% xylene, 1% cumene
Aromatic 150
<10
9
naphthalene
Aromatic naphtha
64742-95-6
<5
2
1% xylene, 1% cumene
Aromatic solvent
64742-94-5
<10
10
naphthalene
Exempt mineral spirits
8032-32-4
0
0
none
Ligroines (VM & P)
8032-32-4
0
0
none
VM & P
8032-32-4
0
0
none
Lactol spirits
64742-89-6
12-20
15
toluene
Low aromatic white spirit
64742-82-1
0
0
none
Mineral spirits
64742-88-7
<2
1
xylenes
Hydrotreated naphtha
64742-48-9
<0.01
0
none
Hydrotreated light distillate
64742-47-8
<0.1
0.1
toluene
Stoddard solvent
8052-41-3
1-2
1
xylenes
Super high-flash naphtha
64742-95-6
0-6
5
xylenes
Varsol® solvent
8052-49-3
<1.1
1
0.5% xylenes, 0.5% ethylbenzene
VM & P naphtha
64742-89-8
5-10
6
3% toluene, 3% xylene
Petroleum distillate mixture
68477-31-6
4-9
8
4% naphthalene, 4% biphenyl
14.26 Use of solvents in the shipbuilding
1035
Table 14.26.5. Petroleum solvent groups and associated HAP contents. [Adapted, by permission from Chemical Manufacturer Association’s Solvent’s Council] Solvent Type
Solvent Name
HAP Component HAP Concentration, wt%
Mineral spirits 135
Xylene
1
Mineral spirits 150 EC
Toluene
1
Naphtha
Ethylbenzene
1
Mixed hydrocarbon Aliphatic hydrocarbon Aliphatic
Naphthol spirits Petroleum spirits Petroleum oil Petroleum naphtha Solvent naphtha Solvent blend Estimated total HAP for each solvent type
Aromatic
3
Medium-flash naphtha
Xylene
4
High-flash naphtha
Toluene
1
Aromatic naphtha
Ethylbenzene
1
Light aromatic naphtha Light aromatic hydrocarbons Aromatic hydrocarbons Light aromatic solvent Estimated total HAP for each solvent type
6
The percentage of solvent used that evaporates during cleaning operations will depend on the volatility of the solvent and on the type cleaning operation. Any materials such as rags or sponges that come into contact with the solvent must also be considered part of the solvent cleaning system and disposed of properly. Since marine coatings are mostly air dried, it is generally impractical to capture air emissions from coating operations and until recently, to duct air emissions into abatement devices such as incinerators or adsorbers. The use of low solvent (VOC or HAP) containing coatings and high performance coatings provide a suitable means of achieving lower volatile emissions from coating of ships. Such coatings will lead to lower life cycle costs and emissions, since the ship will require less rework. EPA has recently approved the use of technology for capturing and destroying air pollutant emissions from the application of coatings to the hull of a ship.29 14.26.8 SOLVENT WASTE The waste cleaning solvent is handled differently by different shipyards. Facilities that use different type solvents for cleaning and thinning, or utilize multiple cleaners may select disposal rather than recycling for re-use. This decision is typically based on cost. Other facili-
1036
Mohamed Serageldin, Dave Reeves
ties will first mix their waste solvents with unused coating. Incineration is generally selected as the disposal method when it is determined to be less expensive than recycling. The type of solvent waste generated from shipbuilding and ship repair facilities does not vary much. Surface preparation and cleaning operations following application of coatings does not represent the most significant part of the organic solvent waste. Certain coatings, such as antifouling coatings, contain toxic pigments (metals) such as chromium, lead, and tributyltin. The waste will likely be hazardous waste if the waste material contained coating residues or other materials that are toxic. Solvents used in cleaning are often mixed with other waste fuels such as hydraulic fluids, waste fuels, and other maintenance by-products. Some of the larger shipyards segregate the solvent waste or recycle it for re-use. 14.26.9 REDUCING SOLVENT USAGE, EMISSIONS, AND WASTE The amount of organic solvent usage for cleaning operations makes up about 20 percent of the total solvent usage in cleaning operations and coating operations at shipyards. The waste from solvent cleaning represents a major portion of the waste in a plant that has surface coating operations. Line cleaning, gun cleaning, and tank cleaning are three operations that utilize most of the cleaning solvents. There are many steps that a facility can undertake to reduce the use of solvent for cleaning. It can reuse some of the solvent, scrape off the paint left on the inside of a mix tank, or modify existing cleaning practices and procedures.5,30 Some shipyards have modified their coating operations to use less solvent. For planning purposes, a facility needs to know where the major solvent usage, emissions, and waste occur in the facility. This may be achieved by: • Developing a solvent accounting system for tracking the usage, fate, and costs of organic solvents in coating operations and cleaning operations; • Identifying actions management will take to reduce solvent content in coatings and reduce usage of cleaning materials containing organic solvents.5,31 The steps are not mutually independent and require full management support and financial backing. Obtaining reliable data is one of the major problems that a manager or supervisor involved in pollution reduction faces. Any program adopted should be one that can identify accurate and specific material usage, emission data, and waste data that will be useful for evaluating pollution cost and risk.32 The “unit operation system” (UOS) approach was used to categorize the universe of cleaning into nine main categories equivalent to those identified here in the section discussing cleaning operations. The concept of a UOS provides the foundation for a standardized material accounting method that provides transportable data, that a secondary user can understand, for setting up a material balance, for comparing alternatives, or evaluating risk. The visual representation of a set up, as was done in Figure 14.26.1, is an essential component of this approach. When facility managers and operators are able to identify with some confidence, at the source, final solvent disposition and cost of cleaning, they are in a better position to make informed decisions and take the necessary actions for reducing solvent usage. Facilities have the added incentive to reduce worker exposure to toxics contained in solvents because of OSHA requirements.33 The facility owners and operators also have to abide by state, local, and federal regulations that limit the amount of VOCs and HAPs emitted to the air and water streams. Facilities that generate large amounts of solvent waste streams will have to deal with RCRA hazardous waste regulations.34
14.26 Use of solvents in the shipbuilding
1037
14.26.10 REGULATIONS AND GUIDELINES FOR CLEANING SOLVENTS The U.S. EPA developed a guideline document to help state and local agencies reduce VOC emissions for the use of cleaning solvents. The document does not provide emission limits, but instead recommends the use of a management system for tracking usage, emissions, and waste from the use of organic solvents and provides a list of definitions that should be used for that purpose.35 Less than 20 states had regulated VOC emissions from the use of organic cleaning solvents by 1997.36 Many states only had minimal record-keeping requirements. Some regulations are driving facilities to use solvents with lower vapor pressures. These low vapor pressure solvents evaporate more slowly at ambient temperatures. However, the emission benefits cannot be guaranteed to last forever. An organic solvent should eventually all evaporate unless it is reused, recycled, or sealed in a container. Many of the cleaning solvents used, such as 1,1,1-trichloroethane, are not considered VOC in the United States. Those solvents are regulated as HAPs and are being phased out because of the Montreal Protocol which aims to reduce ozone depleters. Alternative cleaners are now being used and further developed. To meet regulatory requirements, a facility will often replace or substitute a cleaning product containing HAP material(s) with one or more products containing no HAP solvents (but may contain VOC solvents or vice versa). Acetone, once the preferred solvent for fiberglass boat manufacturers, was replaced starting in 1992 with other solvents such as diacetone alcohol.37 The latter solvent is as effective as acetone, but has a much lower vapor pressure (0.80 mm Hg at 20oC) and flash point (52oC) to reduce the amount of VOC emitted and hazardous reportable releases. Acetone was recently determined not to have significant photochemical reactivity and is no longer considered a VOC in the United States. As a result, acetone is now making a comeback in cleaners and in some coatings. HCFCs are now being used instead of CFCs, but will soon need to be replaced because of the Montreal Protocol. There are many articles in the literature that discuss alternatives that will reduce VOC and HAP emissions.38-42 The use of waterborne materials or other material(s) containing no HAPs or VOCs is the ideal solution, when such coatings or cleaners will do the job. Software and internet sites also exist to help in the selection of the appropriate solvent cleaner or method of cleaning. Disclaimer The views expressed in this chapter are those of the authors and do not necessarily reflect those of their affiliation. Mentioning the names of organizations does not constitute an endorsement. REFERENCES 1 2 3 4 5
6 7 8 9
South Coast Air Quality Management District (SCAQMD), Rule 1106: Marine Coating Operations, 1995. SCAQMD, Rule 1171: Solvent Cleaning Operations, (1996) and Rule 1122: Solvent Degreasers, (1996). 40 CFR Part 63, Subpart II, (1995), U.S. Government Printing Office, Washington, DC. 40 CFR, Part 63, Subpart T, (1996), U.S. Government Printing Office, Washington, DC. U.S. Environmental Protection Agency (EPA), Control Techniques Guideline Shipbuilding and Ship Repair Operations (Surface Coating), Office of Air Quality Planning and Standards, North Carolina, EPA-453/R-94-015, (NTIS no. PB94-156791). U.S. EPA, Federal Register, August 27, 1996 (vol. 61, no. 167), pp. 44050-44057. M. A. Serageldin in Electrochemical Society Proceedings, Volume 95-16, Electrochemical Society Inc., Chicago, Illinois 1997, pp. 1-16. T. J. Snider, An Analysis of Air Pollution Control Technologies for Shipyard Emittted Volatile Organic Compounds (VOCs), National Shipbuilding Research Program, Report No. 0376. March 1993, pp. 3-5. U.S. EPA, EPA Office of Compliance Sector Notebook: Profile of The Shipbuilding and Repairing Industry, EPA Office of Compliance, Washington, D.C, EPA/310-R-97-008, 1997, pp. 75-78.
1038
10
11 12
13 14 15 16
17 18 19 20 21 22
23
24 25 26 27 28 29
30 31 32 33
34 35 36
Mohamed Serageldin, Dave Reeves
U.S. EPA, Surface Coating Operations at Shipbuilding and Ship Repair Facilities—Background Information for Proposed Standards, OAQPS, North Carolina, EPA-453/D-94-011a, (NTIS no. PB95-197471), 1994, pp. 6-7 to 6-9. U.S. EPA, Alternative Control Techniques Document—Industrial Cleaning Solvents, Office of Air Quality Planning and Standards, North Carolina, EPA-453/R-94-015, 1994, pp. 3-17 to 3-28. M. A. Serageldin in Emission.Inventory Living in a Global Environment, Proc..Special. Conf., December 8-10, 1998, New Orleans, LA, Air and Waste Management Association, Pittsburgh, PA, VIP-88, vol.1, 1999, pp. 341-48. Metals Handbook, 9th edition, v.5 Surface Cleaning, Finishing, and Coating, Table 21, American Society for Metals, Metal Park, Ohio, W.G. Wood, Ed., 1990, p. 506. U.S. EPA, Guide to Pollution Prevention: The Marine Maintenance and Repair Industry, Office of Research and Development, Washington D.C., EPA/625/7-91/015, 1991, pp. 5-10. Metals Handbook, 9th edition, v.5 Surface Cleaning, Finishing, and Coating, American Society for Metals, Metal Park, Ohio, W.G. Wood, Ed., 1990, pp. 504 - 505. U.S. EPA, Surface Coating Operations at Shipbuilding and Ship Repair Facilities—Background Information for Proposed Standards, Office of Air Quality Planning and Standards, North Carolina, EPA-453/D-94-011a, (NTIS no. PB95-197471), 1994, p. 318. C. H. Hare, Protective Coatings Fundamentals of Chemistry and Composition, 26. Solvent Families, Technology Publishing Company, Pittsburgh, PA, 1994, pp. 371-383. H. Burrell, 1995 Official Digest of the Federation of Society for Paint Technology, vol. 27, no. 369, 1955, p. 726. U.S. EPA, Alternative Control Techniques Document—Industrial Cleaning Solvents, Office of Air Quality Planning and Standards, North Carolina, EPA-453/R-94-015, 1994, pp. H-1 to H-6. U.S. EPA, Alternative Control Techniques Document—Industrial Cleaning Solvents, Office of Air Quality Planning and Standards, North Carolina, EPA-453/R-94-015, 1994, pp. C-1 to C-18. H. Bleile and S. Rodgers, Marine Coatings. Federation Series on Coatings Technology. 1989. pp. 8. U.S. EPA, Surface Coating Operations at Shipbuilding and Ship Repair Facilities—Background Information for Proposed Standards, OAQPS, North Carolina, EPA-453/D-94-011a, (NTIS no. PB95-197471), 1994, pp. 3-6 to 3-11. U.S. EPA, Surface Coating Operations at Shipbuilding and Ship Repair Facilities—Background Information for Proposed Standards, Office of Air Quality Planning and Standards, North Carolina, EPA-453/D-94-011a, (NTIS no. PB95-197471), 1994, pp 3-18. C. H. Hare in Protective Coatings Fundamentals of Chemistry and Composition, 1. The Basic Composition of Paint, Technology Publishing Company, Pittsburgh, PA, 1994, pp.1-13. 40 CFR 372.65 (1986), U.S. Government Printing Office, Washington, D.C. A. T. Chen and R. T. Wojcik in Metal Finishing , 97’ Organic Finishing Guidebook and Directory Issue, Metal Finishing, Pittsfield, MA, 1997, pp. 167-179. G. L. Muir in Metal Finishing , 97’ Organic Finishing Guidebook and Directory Issue, Metal Finishing, Pittsfield, MA, 1997, pp.391-398. U.S. EPA 33/50 Program: The Final Record, Office of Pollution Prevention and Toxics, EPA-745-R-99-004, Washington, D.C., March 1999. U. S. EPA, Evaluation of Application for Approval of an Alternative Methodology for Compliance with the NESHAP for Shipbuilding and Recommended Requirements for Compliance, Research Triangle Park, N.C., EPA-453/R-99-005, 1999. P.M. Randall in Proc: Poll. Prevent. Conf. Low- and No VOC Coating Technol., Air and Engineering Research Lab, RTP, NC, EPA-600/R-94-022, 1994, pp.489-499. M. A. Serageldin, in Waste Minimization Through Process Design, A. P. Rossiter, Ed., McGraw-Hill, New York, April 1995, pp. 289-303 . M.A. Serageldin, The Unit Operation System for Measuring Plant Environmental Performance (with selected examples), NTIS PB98-124399INZ, November 1996. Occupational Safety and Health Administration, Shipyard Industry, U.S. Department of labor. Document No. 2268, [Online]. URL: Available from http://www.osha-slc.gov/OshDoc/Additional.html#OSHG-HAZ, (revised 1998). 42 U.S.C. s/s 321 et seq. (1976) U.S. EPA, Alternative Control Techniques Document—Industrial Cleaning Solvents, OAQPS, North Carolina, EPA-453/R-94-015, 1994, pp. A1- A7. Emission Inventory Improvement Program: Volume III - Area Sources: Chapter 6— Solvent Cleaning, 1997 [Online]. Available from URL: http://www.epa.gov/ttn/chief/eiip/iii06fin.pdf.
14.26 Use of solvents in the shipbuilding
37 38 39 40 41 42
1039
M. Melody, Boatmakers find solvent substitute cuts emission, costs, Hazmat World, February 1992, pp36-39. 1991 UNEP Solvents, Coatings, and Adhesives Technical Options Report, UNEP, Paris, December1991, pp.58-69 and pp. 153-178. P. Milner in The Coating Agenda America 1955/1996, Campden Publishing Limited (in association with National Paint and Coating Association, U.S.A), London, U.K. pp. 125 and 127. F. Hussey in Advanced Coating Technology, Proc. 3rd, Ann. ESD Adv. Coating. Conf., Dearborn, MI, 9-11 November 1993, ESD, The Engineering Society, Ann Arbor, Michigan, pp. 250 -261. L. Clark, S. Kaldon, E. Reid, and D. Ellicks, Putting Solvent Back in Paint Legally, Metal Finishing, 1998, pp 24-28. C. Mouche, Army Tests Enviro-Friendly Solvents Alternative, Pollution Engineering, September, 1999, p. 35.
14.27 STONE, CLAY, GLASS, AND CONCRETE George Wypych ChemTec Laboratories, Inc., Toronto, Canada
This is a very diverse industry segment which manufactures glass, glassware, cement, concrete, clay products pottery, gypsum, and plaster products. In the USA the industry employs about 500,000 people and operates over 16,000 facilities. There is a broad range of
Table 14.27.1. Reported solvent releases from the stone, clay, glass, and concrete industry in 1995 [Data from Ref. 2] Solvent
Amount. kg/year
Solvent
Amount. kg/year
benzene
4,000
ethylene glycol
42,000
n-butyl alcohol
74,000
hexane
12,000
sec-butyl alcohol
1,700
isopropyl alcohol
300
tert-butyl alcohol
4,000
methanol
560,000
carbon disulfide
5
methyl ethyl ketone
700,000
carbon tetrachloride
14
methyl isobutyl ketone
25,000
chlorobenzene
14
tetrachloroethylene
44,000
chloroform
14,000
1,2,4-trichlorobenzene
46,000
cresol
610
1,1,1-trichloroethane
330,000
cyclohexane
77,000
trichloroethylene
110,000
1,4-dichlorobenzene
14,000
toluene
890,000
1,2-dichloroethane
17
xylene
500,000
dichloromethane
120,000
m-xylene
450
1,4-dioxane
230
o-xylene
540
ethylbenzene
60,000
p-xylene
110
1040
George Wypych
Table 14.27.2. Reported solvent transfers from the stone, clay, glass, and concrete industry in 1995 [Data from Ref. 2] Solvent
Amount. kg/year
Solvent
Amount. kg/year
benzene
6,000
ethylene glycol
29,000
n-butyl alcohol
44,000
hexane
15,000
sec-butyl alcohol
600
isopropyl alcohol
2,500
tert-butyl alcohol
2,000
methanol
300,000
carbon disulfide
5
methyl ethyl ketone
760,000
carbon tetrachloride
5,000
methyl isobutyl ketone
92,000
chlorobenzene
4,000
tetrachloroethylene
26,000
chloroform
20,000
1,2,4-trichlorobenzene
10,000
cresol
7,000
1,1,1-trichloroethane
60,000
cyclohexane
12,000
trichloroethylene
54,000
1,4-dichlorobenzene
3,000
toluene
800,000
1,2-dichloroethane
3,000
xylene
470,000
dichloromethane
20,000
m-xylene
600
N,N-dimethylformamide
5,000
o-xylene
1,200
ethylbenzene
67,000
p-xylene
340
processes involved which use a wide variety of chemicals. Solvents are used for cleaning and are components of product decoration materials. Tables 14.27.1 and 14.27.2 give data on the reported releases and transfers of solvents by the US stone, clay, glass, and concrete industry. Numerous solvents are used but most are used in small quantities the only high volume ones being toluene, xylene, methanol, 1,1,1-trichloroethane, and methyl ethyl ketone. Although the industry does not release large quantities of solvents, their total VOC contribution puts them second among the US industries. Total releases and transfers are one of the smallest of all industries.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Stone, Clay, Glass, and Concrete Products Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
14.28 Textile industry
1041
14.28 TEXTILE INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The US textile industry operates over 5,500 plants, employs over 600,000 people and has shipments exceeding $70 billion. In addition to conventional methods of processing and finishing, this industry also produces coated fabrics, tire cord and fabrics, and cordage and twine. Numerous operations are employed by the textile manufacturing and finishing but only a few of these operations involve the use of solvents. These include printing (volatile solvents and xylenes), finishing (e.g., methanol), scouring (glycol ethers and mineral spirits), cleaning knit goods, desizing (glycol ethers and methanol in PVA desizing), and coating (methyl ethyl ketone and toluene). Tables 14.28.1 and 14.28.2 provide data on the reported releases and transfers of solvents by the US textile industry. Methyl ethyl ketone, toluene, and methanol are emitted in the greatest quantities. The total number of solvents used is low. The textile industry is among the lowest contributors to VOC and is the second smallest (after shipbuilding industry) in the amount of emitted and transferred solvents.
REFERENCES 1 2
EPA Office of Compliance Sector Notebook Project. Profile of the Textile Industry. US Environmental Protection Agency, 1997. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.
Table 14.28.1. Reported solvent releases from the textile industry in 1995 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
n-butyl alcohol
900
methyl isobutyl ketone
190,000
1,4-dichlorobenzene
700
N-methyl-2-pyrrolidone
180,000
1,2-dichloroethane
4,000
tetrachloroethylene
29,000
dichloromethane
230,000
1,2,4-trichlorobenzene
21,000
N,N-dimethylformamide
53,000
1,1,1-trichloroethane
150,000
ethylene glycol
66,000
trichloroethylene
130,000
hexane
59,000
1,2,4-trimethylbenzene
24,000
isopropyl alcohol
11,000
toluene
1,600,000
methanol
1,300,000
xylene
380,000
methyl ethyl ketone
2,200,000
1042
George Wypych
Table 14.28.2. Reported solvent transfers from the textile industry in 1995 [Data from Ref. 1] Solvent
Amount, kg/year
Solvent
Amount, kg/year
1,2-dichloroethane
3,500
tetrachloroethylene
26,000
dichloromethane
9,000
1,2,4-trichlorobenzene
51,000
N,N-dimethylformamide
52,000
1,1,1-trichloroethane
2,000
ethylene glycol
210,00
trichloroethylene
170,000
methanol
120,000
1,2,4-trimethylbenzene
20,000
methyl ethyl ketone
640,000
toluene
330,000
methyl isobutyl ketone
61,000
xylene
67,000
N-methyl-2-pyrrolidone
83,000
14.29 TRANSPORTATION EQUIPMENT CLEANING George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Transportation equipment cleaning includes cleaning of interior of trucks, rail cars, barges, intermodal tank containers, intermediate tank containers, tank interiors, and the exterior of aircraft. These operations rarely involve solvents. Transportation equipment is also cleaned before refurbishing which includes old paint removal, surface cleaning, and painting. These processes usually do require solvents. Aircraft deicing is done using a mixture containing ethylene glycol. There is no data available on the amount of solvents used and emitted from these processes.
REFERENCES 1
EPA Office of Compliance Sector Notebook Project. Profile of the Transportation Cleaning Industry. US Environmental Protection Agency, 1995.
14.30 WATER TRANSPORTATION George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Water transportation equipment is divided into self-propelled vessels and barges. The US industry is composed of over 13,000 establishments employing over 150,000 people operating about 40,000 vessels. Solvents are used in maintenance for paint removal, painting, degreasing engines, and cleaning carburetors. There is no information available regarding solvent use by the industry.
14.31 Wood furniture
1043
REFERENCES 1
EPA Office of Compliance Sector Notebook Project. Profile of the Water Transportation Industry. US Environmental Protection Agency, 1997.
14.31 WOOD FURNITURE George Wypych ChemTec Laboratories, Inc., Toronto, Canada
The US wood furniture industry operates over 11,000 manufacturing facilities which employ over 250,000 people. Total sales are over $22 billion. Several operations use solvents, including veneer application (adhesives), derosination (certain types of wood contain rosin which may interfere with the effectiveness of certain finishes; rosin is removed by application of a mixture of acetone and ammonia), pretreatment (coating contains coalescing solvent (slow evaporating) and diluent (fast evaporating), a new UNICARB® technology uses carbon dioxide in the mixture with a coalescing solvent which eliminates the need for a diluent), several finishing operations (stains, paints, fillers, inks used to print simulated wood grain on plywood, topcoat, and cleaning liquids). Nitrocellulose lacquers are commonly used in the USA; polyurethane and unsaturated polyester finishes are still rarity. Solvents and thinners used in paints include toluene and xylene. Adhesives usually contain solvents such as methyl isobutyl ketone, methyl ethyl ketone, xylene, toluene, and 1,1,1-trichloroethane. Table 14.31.1 shows the relative amounts of VOC emissions from different finishing components depending on furniture type and type of process used. The long process consists of applications of 3 or more coats of stain, a washcoat, a filler, a sealer, a highlight, and two to three topcoat applications. In the short-sequence process two stain coats are applied, followed by a washcoat, a sealer, and two top coat applications. Office furniture uses only one application of a stain, a sealer, and a topcoat. Tables 14.31.2 and 14.31.3 provide data on the reported releases and transfers of solvents by the US furniture and fixtures industry. Toluene, xylene, methanol, and methyl ethyl ketone are released in the largest quantities. The furniture and fixtures industry is one of lesser contributors of VOCs releases and transfers among the US manufacturing industries. Several alternative technologies are under development and some are at the field trial stage.3-12 Waterborne clear varnishes are being used increasingly but the market is still dominated by solvent-based varnishes. One reason may be that waterborne varnishes are 2-3 times more expensive than solvent-based products.3 It is also not clear if waterborne varnishes have lower odor and can be applied at the same rate as solvent-based products. Waterborne varnishes have one important advantage in clear coatings for they have no color therefore the natural wood color is more vivid. Two-component water-based polyurethane coatings have been introduced recently.4 They have a very low VOC content and have very good drying speed, gloss, and potlife. The drying rate and potlife are controlled by the ratio of isocyanate and hydroxyl components. Increasing this ratio increases drying speed but decreases potlife. However, at 4 hours, potlife is acceptable. Polyurethane dispersions are used in furniture adhesive applications, and are especially useful in the lamination of kitchen tables.5
1044
George Wypych
Table 14.31.1. Relative VOC emissions. [Reproduced from EPA draft Guidelines for the Control of Volatile Compound Emissions from Wood Furniture Coating Operations] Percentage of total emissions
Operation
Long process
Short process
Office furniture
Stain
26
28
32
Washcoat
4
4
-
Filler
3
-
-
Wiping stain/glaze
8
-
-
Sealer
18
32
32
Highlight
1
-
-
Topcoat
40
36
36
Table 14.31.2. Reported solvent releases from the furniture and fixtures industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
n-butyl alcohol
940,000
methyl isobutyl ketone
590,000
cyclohexane
10
tetrachloroethylene
7,300
dichloromethane
180,000
1,2,4-trichlorobenzene
90
ethylbenzene
360,000
1,1,1-trichloroethane
160,000
ethylene glycol
30
trichloroethylene
29,000
hexane
25,000
1,2,4-trimethylbenzene
140,000
isopropanol
105,000
toluene
5,300,000
methanol
3,600,000
xylene
3,300,000
methyl ethyl ketone
1,900,000
One problem related to all water-base coatings is the degree to which they penetrate into wood grain. Waterborne coatings are composed of relatively large emulsified particles which are frequently the size of the pit openings. After the emulsion droplets dry out they become very viscous and penetration becomes even more difficult. The penetration of alkyd waterborne coatings was tested by microautoradiography.6 The results indicate that the penetration rate is controlled by the viscosity and form of the material (solution vs. emulsion) is less important. UV curable coatings have also problems in application to wood finishing. These finishes need to have low viscosity but it is difficult to formulate 100% solids curable coatings which have a sufficiently low application viscosity.7 It possible to formulate waterborne UV curable coatings but their performance is still an issue.
14.31 Wood furniture
1045
Table 14.31.3. Reported solvent transfers from the furniture and fixtures industry in 1995 [Data from Ref. 2] Solvent
Amount, kg/year
Solvent
Amount, kg/year
n-butyl alcohol
53,000
methyl isobutyl ketone
145,000
cyclohexane
110
1,2,4-trichlorobenzene
110
dichloromethane
28,000
1,1,1-trichloroethane
5,700
ethylbenzene
340,000
trichloroethylene
300
hexane
16,000
1,2,4-trimethylbenzene
160,000
isopropanol
11,000
toluene
1,030,000
methanol
570,000
xylene
1,700,000
methyl ethyl ketone
380,000
Dunlop developed a two-part water-based adhesive which performed well in lamination of wood and foam.8 Also, a new reactive hot-melt system for attaching moldings in furniture assembly is in use.9 A water reducible coating for wood finishing has been developed based on copolymer which can be cured by crosslinking.11 Another solvent-free material was developed to cover scratches and other damage of finished wood articles.12 Replacement of solvent based adhesives is essential if this industry, which uses 12% of all adhesives, is to significantly reduce its solvent usage.10
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12
EPA Office of Compliance Sector Notebook Project. Profile of the Wood Furniture and Fixtures Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. P B Bell, J J Bilancieri, Paint & Ink International, 9, No.5, , 6-10 (1996). M J Dvorchak, J. Coat. Technol., 69, No.866, 47-52 (1997). M Moss, Pigment Resin Technol., 26, No.5, 296-9 (1997). R M Nussbaum, E J Sutcliffe, A C Hellgrn, J. Coat. Technol., 70, No.5, 49-57 (1998) S Peeters, J P Bleus, Z J Wang, J A Arceneaux, J Hall, Paint & Ink International, 11, No.1, 6-9 (1998). Plast. Rubber Weekly, No.1691, 1997. Hughes F, TAPPI 1997 Hot Melt Symposium. Conference Proceedings. TAPPI. Hilton Head, SC, 15th-18th June 1997, p.15-21. L White, Eur. Rubber J., 179, No.4, 24-5 (1997). S-H Guo, US Patent 5.646,225, ARCO Chemical Technology, 1997. I J Barlow, US Patent 5,849,838, SC Johnson & Sons, Inc., 1998.
14.32 SUMMARY The data from the US industry given in the various sections of this chapter allow analysis of releases and transfers of the various solvents and helps us to understand the patterns of solvent use. Also industries can be ranked based on the amounts of solvents released. This information is provided in Tables 14.32.1, 14.32.2, 14.32.3, 14.32.4, 14.32.5, and 14.32.6 which contain total release of solvents from all reporting industries, the total transfers of solvents from all reporting industries, the sum of releases and transfers from these industries and the total release of solvents by each industry, the total transfers of solvents by each in-
1046
George Wypych
Table 14.32.1. Reported solvent releases from all industries in 1995 [Data from EPA Office of Compliance Sector Notebooks] Solvent
Amount, ton/year
Solvent
Amount, ton/year
acetone
870
2-ethoxyethanol
19
allyl alcohol
60
ethylbenzene
3,810
benzene
3,378
ethylene glycol
8,122
n-butyl alcohol
5,640
hexachloroethane
16
sec-butyl alcohol
216
hexane
6,571
tert-butyl alcohol
482
isopropyl alcohol
205
carbon disulfide
33,085
methanol
93,965
carbon tetrachloride
30
2-methoxyethanol
23
chlorobenzene
181
methyl ethyl ketone
23,315
chloroform
458
methyl isobutyl ketone
6,876
chloromethane
310
N-methyl-2-pyrrolidone
1,111
cresol
800
methyl tert-butyl ether
1,477
m-cresol
320
pyridine
187
o-cresol
270
tetrachloroethylene
1,692
p-cresol
162
toluene
41,015
cyclohexane
3,001
1,2,4-trichlorobenzene
1,245
cyclohexanone
303
1,1,1-trichloroethane
8,631
cyclohexanol
1,100
trichloroethylene
6,611
1,2-dichlorobenzene
15
trichlorofluoromethane
84
1,4-dichlorobenzene
15
1,2,4-trimethylbenzene
2,308
1,2-dichloroethane
222
xylene
26,544
dichloromethane
16,332
m-xylene
269
N,N-dimethylformamide
447
o-xylene
82
1,4-dioxane
33
p-xylene
1,660
dustry, and the total release and transfer of solvents by each industry, respectively. The three highest entries in the tables (either amount, release, or transfer) are marked by bold (first), italic (second), or underlining (third). Methanol, ethylene glycol, toluene, xylene, and carbon disulfide occupy the highest places in Tables 14.32.1-14.32.3, followed by methyl ethyl ketone, dichloromethane, 1,1,1-trichloroethane, tert-butyl alcohol, hexane, and methyl isobutyl ketone. These eleven solvents constitute 86% of all 550,738 tons of solvents released and transferred per year. Methanol alone constitutes over a quarter (27.1%) of the released and transferred solvents.
14.32 Summary
1047
Table 14.32.2. Reported solvent transfers from all industries in 1995 [Data from EPA Office of Compliance Sector Notebooks] Solvent
Amount, ton/year
Solvent
Amount, ton/year
acetone
767
2-ethoxyethanol
2
allyl alcohol
330
ethylbenzene
4,691
benzene
792
ethylene glycol
72,549
n-butyl alcohol
3,636
hexane
8,910
sec-butyl alcohol
1,723
isopropyl alcohol
168
tert-butyl alcohol
12,749
methanol
55,287
carbon disulfide
260
2-methoxyethanol
8
carbon tetrachloride
224
methyl ethyl ketone
15,119
chlorobenzene
704
methyl isobutyl ketone
6,811
chloroform
322
N-methyl-2-pyrrolidone
1,497
chloromethane
0.5
methyl tert-butyl ether
246
cresol
594
pyridine
103
m-cresol
720
tetrachloroethylene
1,189
o-cresol
57
toluene
18,717
p-cresol
870
1,2,4-trichlorobenzene
69
cyclohexane
1,711
1,1,1-trichloroethane
1,778
cyclohexanone
1
trichloroethylene
1,967
cyclohexanol
4
trichlorofluoromethane
4
1,2-dichlorobenzene
4
1,2,4-trimethylbenzene
729
1,4-dichlorobenzene
3
xylene
27,000
1,2-dichloroethane
3
m-xylene
98
dichloromethane
2,895
o-xylene
491
N,N-dimethylformamide
1,298
p-xylene
9
1,4-dioxane
60
Methanol + xylene + toluene constitute over half (52.6%) of all released and transferred solvents by the analyzed industries. This is in fact good news because neither of the five most frequently released and transferred solvents is considered to have carcinogenic effect. But four (methanol, xylene, ethylene glycol, and toluene) contribute to the pollution of lower atmosphere and xylene and toluene cause formation of ozone in lower atmosphere by which it may affect the respiratory system. Carbon disulfide causes formation of hydroxyl radicals which have relatively long half-life (a few days) and thus participate in a variety of photochemical processes. All five solvents are relatively easy to biodegrade.
1048
George Wypych
Table 14.32.3. Reported solvent releases and transfers from all industries in 1995 [Data from EPA Office of Compliance Sector Notebooks] Solvent
Amount, ton/year
Solvent
Amount, ton/year
acetone
1,637
2-ethoxyethanol
21
allyl alcohol
390
ethylbenzene
8,501
benzene
4,170
ethylene glycol
80,671
n-butyl alcohol
9,276
hexachloroethane
16
sec-butyl alcohol
1,939
hexane
15,481
tert-butyl alcohol
13,231
isopropyl alcohol
373
carbon disulfide
33,345
methanol
149,252
carbon tetrachloride
254
2-methoxyethanol
31
chlorobenzene
885
methyl ethyl ketone
38,434
chloroform
780
methyl isobutyl ketone
13,687
chloromethane
311
N-methyl-2-pyrrolidone
2,608
cresol
1,394
methyl tert-butyl ether
1,723
m-cresol
1,040
pyridine
290
o-cresol
327
tetrachloroethylene
2,881
p-cresol
1,032
toluene
59,732
cyclohexane
4,712
1,2,4-trichlorobenzene
1,314
cyclohexanone
304
1,1,1-trichloroethane
10,409
cyclohexanol
1,104
trichloroethylene
8,578
1,2-dichlorobenzene
19
trichlorofluoromethane
88
1,4-dichlorobenzene
18
1,2,4-trimethylbenzene
3,037
1,2-dichloroethane
225
xylene
53,544
dichloromethane
19,227
m-xylene
367
N,N-dimethylformamide
1,745
o-xylene
573
1,4-dioxane
93
p-xylene
1,669
The largest releases and transfers of methanol are from the pulp and paper (57.7%), organic chemical (21.3%), and the polymer and man-made fibers industry (6.2%) which contribute 85.2% of the total methanol releases and transfers. Ethylene glycol releases and transfers are mostly from the polymer and fiber industry (62.5%), the rubber and plastics (19.1%), and the organic chemical industry (15.9%). These three industries contribute 97.5% of all releases and transfers of ethylene glycol from the reporting industries. Toluene and xylene are typical paint, ink, cleaning and process solvents. The largest source of these
14.32 Summary
1049
Table 14.32.4. Total releases from different industries in the USA in 1995. [Data from EPA Office of Compliance Sector Notebooks] Industry
Release ton/year
Percent total %
Release/employee kg/year
Release/$1000 of shipment, g
Aerospace
3,483
1.1
4.3
48
Metal fabrication
24,310
7.9
221
2455
Inorganic chemicals
643
0.2
5.8
21
Iron and steel
2,434
0.8
6.1
24
Metal casting
7,051
2.3
28
237
Motor vehicle
30,675
10.0
77
153
Organic chemicals
24,678
8.0
197
380
Petroleum
19,419
6.3
259
139
Polymer & fiber
37,646
12.3
358
818
Printing
14,750
4.8
9.8
109
Pulp & paper
70,101
22.8
351
876
Rubber & plastics
45,119
14.7
56
501
Stone, clay, ...
3,630
1.2
7.3
-
Textile
6,629
2.2
11
95
Wood furniture
16,636
5.4
67
756
Total
307,204
100
Average = 111
Average = 472
is from the motor vehicle assembly industry (20%), the printing industry (13.7%), the rubber and plastics industry (11.4%), and the metal fabrication industry (10.7%). In total, these industries contribute over half of all releases and transfers of toluene and xylene (55.8%). The majority of carbon disulfide (82.5%) is contributed by the man-made fiber industry from rayon production. The solvents ranked below the top five in the releases and transfers are mostly used in the paint and surface cleaning. The majority of emitted methyl ethyl ketone comes from the metal fabricating industry (25.2%), the rubber and plastics industry (23.4%), and the vehicle assembly industry (11.5%) which jointly contribute 60.1% of the total releases and transfers. 1,1,1-trichloroethane is released and transferred from the metal fabricating industry (36.6%), the rubber and plastics industry (30.4%), and the motor vehicle assembly industry (8.4%), totals 75.4% of all releases and transfers of this solvent. Tert-butyl alcohol is almost solely contributed by the organic chemical industry (94.6%). Hexane comes from polymer and fiber (57.4%), petroleum (19.5%), and rubber and plastics industry (11.3%) which together give 88.2% of all releases and transfers. Methyl isobutyl ketone is mostly contributed by the motor vehicle industry (56.7%) but metal fabricating (13%) and rubber and plastics industry (11.3%) also release and transfer large amounts of this solvent. Together all three industries account for 81% of its total release and transfer. Dichloromethane
1050
George Wypych
Table 14.32.5. Total transfers from different industries in the USA in 1995. [Data from EPA Office of Compliance Sector Notebooks] Industry
Transfer ton/year
Percent total %
Transfer/employee kg/year
Transfer/$1000 of shipment, g
Aerospace
1,080
0.4
1.3
15
Metal fabrication
13,164
5.3
120
1330
Inorganic chemicals
324
0.1
2.9
11
Iron and steel
4,328
1.7
10.8
43
Metal casting
178
0.1
0.7
6
Motor vehicle
23,980
9.6
60
60
Organic chemicals
63,680
25.3
509
980
Petroleum
1,382
0.6
18
10
Polymer & fiber
78,584
31.4
748
1708
Printing
4,051
1.6
2.7
30
Pulp & paper
23,457
9.4
117
293
Rubber & plastics
27,215
10.9
34
302
Stone, clay, ...
2,736
1.1
5.5
-
Textile
1,845
0.7
3.1
26
Wood furniture
4,439
1.8
18
202
Total
250,443
100
Average = 110
Average = 358
comes from rubber and plastics (65.5%), metal fabricating (24.8%), and polymer and fibers industry (8.1%) totals 98.4% of dichloromethane. The second tier solvents are more harmful to health and environment than the first five most frequently released and transferred solvents. Ketones are suspected carcinogens by some sources. Dichloromethane is a probable human carcinogen. 1,1,1-trichloroethane is an ozone depleter. Because of their volatility, these solvents reside mostly in the lower atmosphere causing pollution and participating in radical processes. Five industries are the largest contributors to the second tier solvents: rubber and plastics (23.8%), metal fabricating (18.1%), motor vehicle assembly (11.8%), organic chemicals (11.7%), and polymer and fiber (9.4%). Their joint contribution accounts for 74.8% of the releases and transfers of these solvents. Two pertinent observations can be concluded from the above data • tight controls and applications of new technology in the top three industries using a particular solvent can make big difference • most releases and transfers stem from the use of paints and coatings. Tables 14.32.4-14.32.6 analyze releases from different industries. The polymer & fiber, pulp & paper, organic chemicals, rubber & plastics industries pollute the most from the point of view of releases and transfers of solvents.
14.32 Summary
1051
Table 14.32.6. Total releases and transfers from different industries in the USA in 1995. [Data from EPA Office of Compliance Sector Notebooks] Industry
Total ton/year
Percent %
Total/employee kg/year
Total/$1000 of shipment, g
Aerospace
4,563
0.8
5.7
63
Metal fabrication
37,474
6.7
341
3785
Inorganic chemicals
967
0.2
8.8
32
Iron and steel
6,762
1.2
16.9
68
Metal casting
7,229
1.3
28.9
243
Motor vehicle
54,355
9.8
136
271
Organic chemicals
88,358
15.8
707
1359
Petroleum
20,801
3.7
277
158
Polymer & fiber
116,230
20.9
1107
2527
Printing
18,801
3.4
12.5
139
Pulp & paper
93,558
16.8
467
1169
Rubber & plastics
72,334
13.0
90
803
Stone, clay, ...
6,366
1.1
12.7
-
Textile
8,474
1.5
14.1
121
Wood furniture
21,075
3.8
84
958
Total
557,347
100
Average = 230
Average = 835
It can be speculated from the release data per employee that workers in pulp & paper, rubber & plastics, and polymer & fiber are the most exposed to solvents. Some technologies cost taxpayers more than others. Metal fabrication is one example. In order to produce goods valued at $1,000, the industry releases and transfers almost 3.8 kg of solvents. The cost in health and cleanup may exceed the value of goods manufactured. It is characteristic that traditional industries which maintain older plants (metal fabrications, polymer & fiber, rubber & plastics, wood furniture, organic chemicals) contribute more to releases and transfers of solvents (as measured by releases plus transfers per $1000 sales), than industries which invest capital in the improvement of equipment, safety, and research and development (e.g., printing, motor vehicle assembly). This suggests that the avoidance of cost of the required investment is one reason for pollution. This reason was illustrated in the discussion of the petroleum industry where one manufacturer almost eliminated pollution by fixing leaking valves. Industries consisting of smaller individual companies have fewer resources to develop new, non-polluting technologies (or even enforce safe practices). For example, the metal fabrication industry is composed of smaller plants using older technology which do not have the resources to make environmental improvements and remain competitive.
1052
George Wypych
The traditional chemical industries (pulp & paper, rubber & plastics, and polymer & fiber) will hopefully apply their knowledge and chemical know-how to eliminate most of their contribution to pollution. In the future this will become an increasingly important requirement in competing by decreasing the social cost of applied technology. Perhaps more plastics could replace metals if the polymer, rubber, and plastics industries could demonstrate that their cost of pollution per dollar of output is much smaller than that of the metal industry.
15
Methods of Solvent Detection and Testing 15.1 STANDARD METHODS OF SOLVENT ANALYSIS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
This section includes information on solvent analysis based on methods included in national and international standards. ASTM standard methods are emphasized because they contain more methods on solvent testing than any other set of methods available. ISO standards are also covered in full detail because they are used in practical applications and are a basis for preparation of national standards. References to other national standards are also given including Australian (AS), Australian/New Zealand (AS/NZS), (British (BS), Canadian (CAN/CGSB), German (DIN), Japanese (JIS), and Finish (SFS). The same number is given in references to all national and international standards for the same solvent property to facilitate searching and referencing. Methods of analysis are grouped by subject in alphabetical order. 15.1.1 ALKALINITY AND ACIDITY Amine acid acceptance by halogenated solvents is the degree to which an amine is capable of absorbing or neutralizing acid present from an external source or generated by the solvent.1 This method is useful for comparing the effect of an amine with other acid-accepting compounds. The test is performed by the titration of an alkaline water extract from the solvent with 0.2 N hydrochloric acid to pH = 3.9 as detected by pH electrode. The alkalinity of acetone is determined by a titration with 0.05 N H2SO4 in the presence of methyl red indicator.2 The test method provides a measure of acetone alkalinity calculated as ammonia. The acidity of halogenated solvents can be determined by titration with 0.01 N sodium hydroxide in the presence of a 0.1% solution of bromophenyl blue indicator.3 Similar to the determination of alkalinity,1 acidity is determined in water extract. The determination can also be done directly in solvent but the solution of sodium hydroxide should be prepared in methanol. A third option is to determine acidity by a pH-metric titration of a water extract. Prior to determination, the pH of solvent is measured. If the pH is above 7, then an alkalinity measurement is done by the above method.1 If the pH is below 7, acidity is determined by this method.3 The method is used to establish purchasing and manufacturing specifications and control the quality of solvents.
1054
George Wypych
The acidity of benzene, toluene, xylenes, naphthas, and other aromatic hydrocarbons is determined by the titration of a water extract with 0.01 N sodium hydroxide in the presence of 0.5% phenolphthalein indicator solution.4 The method is suitable for setting specifications, internal quality control, and development of solvents. The result indicates the potential corrosivity of solvent. The acidity of solvents used in paint, varnish, and lacquer is determined by the titration of solvent diluted with water (for water soluble solvents) or isopropyl alcohol or ethanol (for water insoluble solvents) in proportion of 1:1. A water solution of 0.05 N sodium hydroxide in the presence of 0.5% phenolphthalein indicator dissolved in ethanol or isopropanol is used for titration.5 The method is useful for determination of acidity below 0.05%. Acidity is a result of contamination or decomposition during storage, transportation or manufacture. The method is used to assess compliance with specification. Solvents which are depleted of stabilizers (amine or alpha epoxide) may become acidic. The following method determines the combined effect of both alkaline (amine) and neutral (usually epoxy) stabilizers.6 The determination is done in two steps. First solvent is mixed with hydrochlorination reagent (0.1 N HCl), then the excess is titrated with 0.1 N sodium hydroxide in the presence of 0.1% bromophenyl blue as an indicator. The total acidity of trichlorotrifluoroethane and other halocarbons is determined by titration of a sample diluted with isopropanol with 0.01 N sodium hydroxide in isopropanol in the presence of a 0.05% isopropanol solution of phenolphthalein as an indicator.7 The method is used for setting specifications and quality control. 15.1.2 AUTOIGNITION TEMPERATURE The autoignition temperature can be determined by the hot and cold flame method.8 Cool flames occur in vapor-rich mixtures of hydrocarbons and oxygenated hydrocarbons in air. The autoignition temperature is the spontaneous (self-ignition) temperature at which a substance will produce a hot flame without an external ignition source. Autoignition occurs when a hot flame inside a test flask suddenly appears accompanied by a sharp rise in temperature. With cold flame ignition the temperature rise is gradual. The test equipment shown in a schematic drawing in the method8 consists of a test flask, a furnace, a temperature controller, a syringe, a thermocouple and other auxiliary parts. The measurement is performed in a dark room for optimum visual detection of cool flames. The results are reported as ignition temperature, time lags (delay between sample insertion and material ignition), and reaction threshold temperature (the lowest flask temperature at which nonluminous pre-flame reactions (e.g., temperature rise) occur). The results depend on the apparatus employed. The volume of vessel is especially critical. A larger flask will tend to produce lower temperature results. The method is not designed for materials which are solid at the measurement temperature or which undergo exothermic decomposition. 15.1.3 BIODEGRADATION POTENTIAL The method covers a screening procedure which assesses the anaerobic biodegradation of organic materials.9 The procedure converts organic substances into methane and carbon dioxide which are measured by a gas volumetric pipette − part of the standard apparatus. Other parts include, a biodegradation flask, a magnetic stirrer, a pressure transducer, a syringe, and a water seal. The apparatus may be interfaced with a gas chromatograph to determine quantities of the two gases. The biodegradation process is conducted in a specially prepared medium inoculated with sludge inoculum. The process occurs under the flow of a mixture of 70% nitrogen and 30% carbon dioxide to provide anaerobic conditions.
15.1 Standard methods of solvent analysis
1055
The method was developed to screen organic substances for their potential to biodegrade. If a high degree of biodegradability is determined it provides a strong evidence that the test substance will be biodegradable in the anaerobic digestors of a waste treatment plant and in many natural environments. Other references9 give methods of determining of biological and chemical oxygen demand. 15.1.4 BOILING POINT The boiling point of solvent, its specific heat capacity, and its enthalpy of vaporization determine the energy required for solvent separation in a distillation column. They also determine numerous other properties of solvents (see Chapter 2). Several methods of determination can be used, but two, distillation and gas chromatography are the most popular. Industrial aromatic hydrocarbons are determined by distillation. The temperature is recorded for the initial boiling point, for the sample which has been distilled at 5%, 10%, then at 10% increments up to 90%, then finally at 95%. The temperature should be recorded with precision of 0.1oC.10 A general test method to determine the distillation range of volatile liquids11 outlines a similar method of measurement. In addition to the measurements at the intervals given above, the temperature of the dry point (distillation temperature of residual quantities) is also recorded. Results must be reported as specified in the method. A vacuum distillation procedure is used to determine the amount of solvents in solvent-based paints.12 The paint sample is diluted with tricresyl phosphate, distilled for a while under normal pressure to evaporate the more volatile solvents, followed by vacuum distillation at 2 mm Hg. Capillary gas chromatography is used to determine the boiling point of hydrocarbon solvents.13 The initial boiling point is defined as the point at which the cumulative area of chromatogram equals 0.5% of its final total surface area. The final boiling point is at cumulative area of 95% of the total surface area of chromatogram. The method reports boiling point distribution in 1% intervals over the 1-99% range of the total cumulative surface area of chromatogram as well as the initial and final boiling points. A flame ionization detector is used in the determination and a standard solvent containing 16 known components is used for calibration. 15.1.5 BROMINE INDEX Two methods are used to determine bromine index of aromatic hydrocarbons which contain trace amounts of olefins and are substantially free of materials lighter than isobutane and have distillation end-point lower than 288oC. The methods measure trace amounts of unsaturations in materials which have a bromine index below 500. The bromine index can be measured by electrometric14 and coulometric15 titration. In the electrometric titration method, a sample is titrated with bromide-bromate solution (0.1 N solution of mixture of potassium bromide and potassium bromate) until the end-point increase in potential remains steady for 30 s.14 In coulometric titration, a potassium bromide solution is used to titrate the solvent until the bromine concentration increases because it is no longer being consumed by the unsaturation of the solvent.15 Both methods can be used for setting specification, quality control, and testing of development solvents to find olefinic content. The methods do not differentiate between the types of unsaturations. 15.1.6 CALORIFIC VALUE The heat of combustion of liquid hydrocarbon fuels can be determined with bomb calorimeter.16 Two definitions are used in result reporting: gross heat of combustion (the quantity of energy released from fuel burned in constant volume with all products gaseous
1056
George Wypych
except water which is in liquid state) and net heat of combustion (the same but water is also in a gaseous state). These determinations are useful in assessing the thermal efficiency of equipment used for generation of power or heat. The results are used to estimate the range of an aircraft between refueling stops which is a direct function of heat of combustion. The calorimeter bomb is standardized against benzoic acid standard. Net and gross heats of combustion are reported. A specific method is used for aviation fuels.17 This method reports results in SI units and the measurements are made under constant pressure. The method is applicable for aviation gasolines or aircraft turbine and jet engine fuels. The method is used when heat of combustion data are not available. An empirical equation was developed which gives net heat of combustion based on the determined values of aniline point (ASTM D 611) and API gravity (ASTM D 287). If the fuel contains sulfur, a correction is applied for sulfur determined according to ASTM D129, D 1266, D 2622, or D 3120 (the method selected depends on the volatility of the sample). Gross calorific value and ash content of waste materials can be determined by a calorimetric method.19 After a calorimetric analysis, the bomb washing can be used to determine of mineral content by elemental analysis. The sample is burned under controlled conditions in oxygen. The calorimeter is standardized by burning known amount of benzoic acid. The formation of acids can additionally be determined by titration. 15.1.7 CLEANING SOLVENTS Several standard procedures are available for evaluation of cleaning solvents. The stability of aircraft cleaning compounds is determined after 12 months storage at controlled conditions which may include moderate temperatures, cold storage, and hot storage.20 Solvent vapor degreasing operations which use halogenated solvents follow standardized procedure.21 The standard contains information on the location and design of a degreasing installation and operation during startup, degreasing, shutdown, and solvent reclamation. The purpose of the standard is to reduce the probability of accidents and exposure to personnel. A separate standard practice22 gives reasons for and methods of preventing acid formation in degreasing solvents. The formation of acid is generally related to excessive heat, contaminations, the presence of chlorinated and sulfonated oils, admixture of acids, and solvent mixtures. 15.1.8 COLOR Impurities in benzene, toluene, xylene, naphthas, and industrial aromatic hydrocarbons are determined by a simple colorimetric analysis of an acid wash.23 A solvent is washed with sulfuric acid and the color of the acid layer is determined by a visual comparison with color standards prepared from solutions of cobalt chloride and ferric chloride. Aromatic hydrocarbons which melt below 150oC can be subjected to color analysis in the liquid (molten) state using a visual comparison with platinum-cobalt standards.24 Standards are prepared from a stock solution of K2PtCl2 to form a scale varying in color intensity. Similar procedure was developed to evaluate color of halogenated organic solvents and their admixtures.25 Objective color measurement is based on tristimulus colorimetry.26 This instrumental method measures tristimulus values of light transmitted by a sample and compares the results to the values transmitted by distilled water. The results can be recalculated to the platinum-cobalt scale referred to in the previous standards.24,25 The results can also be interpreted by normal methods of color measurement to yellowness index, color, color depth, etc.
15.1 Standard methods of solvent analysis
1057
15.1.9 CORROSION (EFFECT OF SOLVENTS) The corrosiveness of perchloroethylene to copper is determined using Soxhlet apparatus.27 Three pre-weighed strips of copper are used, one placed in the bottom flask, the second in the bottom of the Soxhlet attachment, and the third below the condenser. The specimens are exposed to refluxing solvent for 72 h after which the entire apparatus is flushed with distilled water to wash all acidic substances back to the flask. The water layer is titrated with 0.01 N NaOH to determine its acidity and the strips are weighed to determine weight loss. The results indicate quality of solvent. A different method is used to test copper corrosion by aromatic hydrocarbons.28 Here, a copper strip is immersed in a flask containing solvent and the flask is placed in boiling water for 30 min. Next, the copper strip is compared with ASTM standard corroded copper strips. If 1,1,1-trichloroethane is not properly stabilized it forms hydrochloric acid in the presence of aluminum. HCl corrodes aluminum. The presence of free water invalidates the result of this test.29 An aluminum coupon is scratched beneath the surface of a solvent. The coupon is observed for 10 min and 1 h and the degree of corrosion is recorded in form of pass (no reaction) or fail (gas bubbles, color formation, or metal corrosion). The test is important to cleaning operations because aluminum should not be used for parts of machines (pumps, tanks, valves, spray equipment) in contact with corrosive solvent. 15.1.10 DENSITY Density and specific gravity of solvents are discussed together. The difference in their definitions is that specific gravity is the density of material relative to the density of water whereas the density is the weight in vacuo of a unit volume. The density of liquids (including solvents) can be measured by a Bingham pycnometer.30 The determination includes introduction of the liquid to the tared pycnometer, equilibration of temperature, then weighing. Other standardized method31 determines the specific density of liquid industrial chemicals by two methods: hydrometer and pycnometer. The pycnometric method is essentially similar to the previously described. It differs in that the water and then solvent are weighed. Thus the density determination error may only be due to an imprecise weighing (the pycnometer calibrated volume does not enter calculations). In the hydrometer method, the calibrated hydrometer is immersed in controlled temperature liquid and direct readings are obtained. The standard method for determining the specific gravity of halogenated organic solvents32 involves the use of both a pycnometer and a hydrometer as described above but, in addition, an electronic densitometer is also used. Here, a liquid is placed in U-shaped tube and subjected to electronic excitations. The density changes the mass of tube and frequency of oscillations which is the basis for measurement and display of specific gravity readings. Two standard tables33,34 (American and metric system of units) are used to calculate weight and volume of benzene, toluene, xylenes mixture and isomers, styrene, cumene, and ethylbenzene as well as aromatic hydrocarbons and cyclohexane. Tables provide volume corrections for these solvents in a temperature range from -5 to 109oF (-20.5 to 43oC). 15.1.11 DILUTION RATIO The dilution ratio is the maximum number of units of diluent that can be added to unit volume of solvent before precipitation occurs. Cellulose nitrate dissolved in an oxygenated solvent (8 wt% resin) is the most classical method to determine dilution ratio used to evaluate toluene as a standard diluent and to compare different diluents and solvents with a standard solvent (n-butyl acetate). The standard dilution ratio of toluene by n-butyl acetate
1058
George Wypych
solution is 2.73-2.83. Two end points are determined. The first occurs when a known amount of diluent forms precipitate after 2 min of vigorous swirling. The second is determined by re-dissolving the precipitate, adding a known volume of solvent (dependent on dilution ratio) and precipitating it again with diluent.35 Similar method can be used for any resin, solvent, or diluent.36 A solution of resin is prepared by the method described in ASTM D 1725 or by dispersing in blender (results may differ). Precipitation with diluent is determined at 25oC. The diluent is added dropwise from burette or weight is controlled throughout the experiment. The resin dilutability is recorded. If more than 100 g of diluent is required for 10 g of solution then diluent is regarded as being infinitely soluble. Heptane miscibility in lacquer solvents is determined by mixing equal amounts of the specimen (lacquer) and heptane.37 If a clear solution results after mixing, it indicates good miscibility. If a turbid solution results, either the heptane is immiscible with the tested specimen or water is present in either component. Water-insoluble admixtures in solvents may affect many uses of solvent. The specimen solvent (primarily acetone, isopropanol, methanol, and many other) is diluted to 10 volumes with water and the resulting turbidity or cloudiness is recorded.38 15.1.12 DISSOLVING AND EXTRACTION A standard practice for preparing polymer solution contains information on solvents, their concentration, temperature, pressure, mixing time, and heating.39 The annex contains information on the best solvents for 75 typical polymers with different degrees of substitution or modification. Frequently, temperature and concentration of solution is also given. Solvent extraction is used on textile materials to determine naturally occurring oily and waxy materials that have not been completely removed from the fibers.40 The percentage of extracted material is given in relationship to the dry mass of fiber. Solvents used for extraction including 1,1,2-trichloro-1,2,3-trifluoroethane and dichloromethane but these may be replaced by other solvents by mutual agreement. The Soxhlet extraction and gravimetric determination are used. 15.1.13 ELECTRIC PROPERTIES Specific resistivity is numerically equivalent to the volume resistance between opposite faces of one centimeter cube. The specific resistivity of electronic grade solvents is measured by a low-voltage a-c bridge operating at 1000 Hz providing the specific resistance does not exceed 109 Ω-cm.41 If the specific resistance is higher than 109 Ω-cm, d-c equipment is used. D-c equipment is capable of measuring up to at least 1012 Ω-cm. The electrical conductivity of liquid hydrocarbons can be measured by a precision meter.42 The method is used to evaluate aviation fuels and other low-conductivity hydrocarbons. The generation and dissipation of electrostatic charges in liquids during handling depends on ionic species which may be characterized by electrical conductivity at rest. The dissipation time of the charges is inversely proportional to the conductivity. The measurement is done by a conductivity meter. Two standard methods43,44 were designed to determine the electric breakdown voltage of insulating oils of petroleum origin. These are VDE electrodes method43 and the method under impulse conditions.44 The dielectric breakdown voltage measurement allows to estimate the ability of an insulating liquid to withstand electric stress without failure. In the presence of contaminations such as water, dirt, or cellulosic fibers low breakdown voltages are obtained. VDE stands for Verband Deutscher Electrotechniker, an organization which designed brass electrodes used for measurement. The equipment uses a transformer, voltage
15.1 Standard methods of solvent analysis
1059
control equipment, and a voltmeter. The impulse method44 uses highly divergent field under impulse conditions. The breakdown voltage of fresh oil decreases as the concentration of aromatic hydrocarbons increases. The method can be used for quality control of fresh oil, and determining the effect of service aging, and effect of impurities. The material is placed in test cell containing electrodes which are supplied from an impulse generator controlled by voltage control equipment. 15.1.14 ENVIRONMENTAL STRESS CRAZING Crazing is a group of surface fissures which appear as small cracks after the material has been exposed to solvent and stress. Crazes are usually oriented perpendicular to stress and their appearance depends on the index of refraction and on the angle of viewing. A suitable light source must be used. Transparent plastic materials can be directly tested for crazing.45 Two variations of method are used to determine stress crazing of transparent plastics: determination of stress required to cause cracking or determination of craze development along the time of stress application. In the first method specimens are exposed to solvent by the direct contact of specimen surface with filter paper wetted with solvent. By selecting different values of stress (using each time new specimen) the range of two stress forces is searched for the largest stress under which specimen does not craze and the smallest stress under which it does craze. In the second variation of the method, sample is tested first without solvent to assure that it does not craze under the selected load. Solvent is then applied and specimen inspected in 15 min time intervals taking note of location of craze front as crazing progresses. If the sample does not craze higher load is selected and vice versa. Similar to the first variation the range is determined within which specimen crazes and does not craze. The crazing effect can be indirectly determined by testing chip impact resistance of specimens which crack either because of weathering or environmental stress cracking.46 15.1.15 EVAPORATION RATE The evaporation rate of a solvent is determined to obtain relative value to some standard, selected solvent. The solvent selection depends on reasons for solvent use and the type of solvent and it is usually agreed upon between interested parties. In Europe, diethyl ether is the most frequently used reference solvent and in the US butyl acetate. The evaporation rate of other solvents is determined under identical conditions and the resultant values are used to rank solvents. The most obvious requirement is that the determination is done without excessive drafts and air currents. The evaporation rate is the ratio of the time required to evaporate a test solvent to the time required to evaporate the reference solvent under identical conditions. The results can be expressed either as the percentage evaporated within certain time frame, the time to evaporate a specified amount, or a relative rate. Relative rate is the most common. For halogenated solvents used in cleaning applications, the relative evaporation rate is compared either to xylene or perchloroethylene.47 The determination is done on a test panel using 10 ml of solvent. The relative evaporation rate is calculated. 15.1.16 FLAMMABILITY LIMITS These tests cover the methods of determination of the minimum temperature at which vapors in equilibrium with liquid solvent are sufficiently concentrated to form flammable mixtures with air at atmospheric pressure and concentration limits of chemicals. Flammable (explosive) limits are the percent levels, volume by volume, of a flammable vapor or gas mixed in air between which the propagation of a flame or an explosion will occur upon the
1060
George Wypych
presence of ignition. The leanest mixture at which this will occur is called the lower flammable limit. The richest mixture at which it will occur is the upper flammable limit. The percent of vapor mixture between the lower and upper limits is known as the flammable range. The temperature limits of flammability can be determined in an air/vapor mixture above a flammable liquid in a closed vessel.48 The temperature in vessel is varied until a minimum temperature is reached at which the flame will propagate away from the ignition source. A glass vessel is equipped with an ignition device, a magnetic stirrer, a clamping devices, and a safety glass window. The initial temperature of determination is estimated from closed-cup flash point measurement. If the flash point is below 38oC, the initial temperature should be 8oC below flash point temperature. If the flash point is between 38 and 96oC, the starting temperature should be at least 14oC below the flash point. If the flash point temperature is above 96oC then the initial temperature should be 22 to 44oC below the flash point temperature. Selecting higher initial temperature may result in explosion. The lower temperature limit of flammability is obtained from the test which can be used to determine guidelines for the safe handling of solvents in closed process and storage vessels. The concentration limits of flammability are determined using another method.49 The method is limited to atmospheric pressure and temperature of 150oC. Equipment is similar to that used in the previous method. A uniform mixture of vapor and air is ignited and flame propagation from ignition source is noted. The concentration of flammable components is varied until a composition is found which is capable to propagate flame. 15.1.17 FLASH POINT A variety of apparatus such as the small scale closed tester,50 the tag open cup,51 the tag closed tester,52 the Cleveland open cup,53 the Pensky-Martens closed cup,54 and the equilibrium method55 are used to determine the flash point. The selection of method is based suggestions included in separate standard.56 The small scale closed tester50 is a metal cup with a thermometer fitted below the bottom of the internal chamber with a hinge mounted cover having filling orifice. The sample is introduced to the cup and the cup is maintained at a constant temperature by means of temperature controller. After a specific time, a test flame is applied for 2.5 s and an observation is made whether or not flash has occurred. If flash did not occur the cup is cleaned, a new sample is introduced and the temperature is increased by 5oC. The measurements are repeated until the flash point is determined with accuracy of 1oC. The tag open cup51 is a larger unit equipped with water-glycol bath for temperature control or a solid carbon dioxide-acetone bath for lower flash points. A much larger sample is used with this equipment and the temperature is gradually increased at a rate of 1oC/min. A taper flame is passed for 1 s in 1 min intervals until the flash point is detected. The tag closed tester52 can use either a manual or an automated procedure. A sample volume of 50 ml is used. Either a gas flame or an electric ignitor is used. In the automated mode, the equipment is programmed to perform the standard procedure. The Cleveland open cup53 is placed on a heated plate which increases temperature at a rate of 5-6oC/min. This method can also be automated. The method is designed for testing petroleum products which are viscous and have flash point above 79oC. The Pensky-Martens54 closed cup tester was also designed for petroleum products but for those with flash points from 40 to 360oC. This apparatus has its own heating source, stirrer, and cover by which it differs from Cleveland cup. It can be either manual or automated. The equilibrium method55 uses either a modified tag close cup or the Pensky-Martens apparatus. The modification intended to keep the vapor/air temperature in equilibrium with the liquid temperature. The method is limited to the temperature range from 0 to 110oC.
15.1 Standard methods of solvent analysis
1061
Depending on the viscosity of liquid and its expected flash point range, one of the above methods is chosen as described in detail elsewhere.56 It should be additionally noted that if the flash point method uses continuous heating, it is not suitable for testing mixtures of flammable substances because their vapor concentrations are not representative of equilibrium conditions. One of the weaknesses of flash point analysis is that the flame is well above the liquid surface therefore full vapor concentration is not attained. Many cases exist where a flash point cannot be detected but the material does form flammable mixtures. Before a method is chosen and a data interpretation made full information on the test procedure should be studied in detail and the proper authorities should be consulted to define safe practices for a particular material. 15.1.18 FREEZING POINT Freezing point apparatus consists of freezing tube, Dewar flasks to act as cooling and warming baths, stirring mechanisms, absorption tubes, clamps and other auxiliary parts.57 Freezing point can be obtained precisely from interpretation of time-temperature freezing and melting curves. The determination is made by measuring the electrical resistance of liquid which decreases on cooling and becomes constant when it freezes. This method in conjunction with the testing details described in a separate standard58 can be used to determine the purity of many hydrocarbon solvents. The data given in the last standard58 allow for a precise determination of the purity of solvent in percent of pure compound. A simple method was designed to determine solidification point of benzene based on visual observation of formation of solid phase.59 15.1.19 FREE HALOGENS IN HALOGENATED SOLVENTS This simple qualitative test involves the extraction by water of free halogens, followed by the reaction of the halogens with potassium iodide in the presence of a starch indicator.60 The solution color changes to blue in the presence of free halogens. 15.1.20 GAS CHROMATOGRAPHY Gas chromatography provides many tools for the analysis of solvents. In section 15.1.4, a method was discussed which determines the distribution of boiling points of hydrocarbon mixtures. Many uses of gas chromatography for the determination of purity of different solvents are discussed in Section 15.1.25. In this section, some examples of of gas chromatography are included to show its usefulness in the qualitative determination of solvents mixtures61 the analysis of solvent impurities,62-64 the determination of solvents in a product by direct injection,65 and the generation of data to evaluate waste materials to determine their hazardous content.66 The relative distribution of aromatic hydrocarbons in xylene products can be quantitatively determined by gas chromatography.61 A flame ionization or thermal conductivity detector is used with a capillary or packed column containing crosslinked polyethylene glycol as the stationary phase. The peak area of each component is measured and the weight percentage concentration is calculated by dividing the peak area of the component by the sum of the areas of all peaks. Ortho-xylene concentration and the concentrations of its admixtures and impurities are measured using a flame ionization detector and a polar fused-silica capillary column.62 An internal standard (iso-octane) is used to increase precision and a standard mixture is used for calibration. A similar method is used with p-xylene but either n-undecane or n-octane are used as the internal standards.63 In both methods, peak areas are interpreted relative to the peak area of internal standard. The main impurities in benzene are non-aromatics with less than 10 carbons, toluene, 1,4-dioxane and aromatics containing 8 carbon atoms. The
1062
George Wypych
method of determination of benzene impurities64 is similar to two methods described above.62,63 Normal-nonane is used as an internal standard. In all three methods the internal standards must be at least 99% pure component. A gas chromatograph equipped with a thermal conductivity or a flame ionization detector and capillary or packed columns is used for direct determination of solvents in paints.65 Columns are usually packed with either polyethylene glycol (molecular weight 20,000) or a diisodecyl phthalate as liquid phase on diatomaceous earth (60-80 mesh) used as a solid support. Low viscosity paints are drawn into a syringe and injected through the injection port. High viscosity paints are diluted with solvent that does not interfere with the analysis, usually ethyl ether or dichloromethane. A standardized gas chromatographic technique66 is capable of determining the 67 solvents most frequently found in hazardous wastes. Several detectors are suggested for analysis such as flame ionization, electron capture, thermal conductivity, photoionization, or mass selective. Each waste mixture may contain a large number of solvents. Their detection is facilitated by the use of gas chromatograph interfaced with mass spectrometer. The method is designed to facilitate site assessment, recycling operations, plant control, and pollution programs. 15.1.21 LABELING Warnings, first aid measures and operating instructions are standardized for vapor degreasers.67 Placards containing this information should be placed close to the degreaser in an area accessible to employees. The placards should contain the information required by applicable federal and local laws and regulations. The placard should include name of the solvent used, and warnings indicating that the vapor is harmful, that breathing the vapor should be avoided, that the machine should only be used with proper ventilation, that swallowing and contact with the skin should be avoided, that cutting or welding should not be performed close to the machine, and that the tank should not be entered unless a proper procedure is followed. In addition, start-up and operation procedures should be available. Minimum labeling requirements for several halogenated solvents should also follow a standard practice.68 The label should state the company name, its logo and address, emergency telephone numbers, lot number, the net weight, solvent name, its CAS number, OSHA PEL and ACGIH TLV values, and quantity. In addition, health and safety information, precautions, first aid, and handling and storage information should be provided. 15.1.22 ODOR Strips of rapid qualitative paper are dipped in a standard liquid and in the liquid under the test. Their odor is compared to establish if the odor of sample is more or less acceptable than the standard to the purchaser and the manufacturer.69 In a similar method, papers dipped in a standard and specimens are allowed to dry at room temperature and tested for residual odor at specified time intervals. Odor testing may be performed by a selected group of panelist to either determine the effect of various additives on the odor or taste of a medium or to determine the odor or taste sensitivity of a particular group of people.70 For this purpose, a series of samples is prepared in concentration scale which increases in geometric increments. At each concentration step two samples containing the medium alone are given to a panelist. The panelist should determine which sample is different from the other two samples. The panelist should begin with the lowest concentration selected to be two or three concentration steps below estimated threshold. The method description contains information on sample selection and preparation, result, and precision determination. A method of evaluation of denatured and undenatured alcohols to assess their acceptability is used.71 This method is developed specifically to compare methanol, ethanol,
15.1 Standard methods of solvent analysis
1063
isopropanol, and n-propanol. The group of panelist is asked to compare the characteristic and residual odors of evaporation, its intensity by dilution, and its concentration. A similar method was developed for a series of glycols.72 Here, odor character and intensity are evaluated. The residual odor of a drycleaning grade of perchloroethylene is determined by comparing treated and untreated samples of bleached cotton fabric. The treated fabric is soaked for 5 min in perchloroethylene and dried for 4 hours at room temperature.73 Good quality perchloroethylene should leave no odor. 15.1.23 PAINTS − STANDARDS RELATED TO SOLVENTS The paint industry, a major user of solvents, has developed numerous standards. Some are included in Sections 15.1.25, 15.1.31, and 15.1.35. The paint industry also uses many general standards and some specific standards, which have not been included in any other section of this chapter. Details of these are given below. Sampling and testing requirements for solvents used in the paint industry are summarized in a special standard.74 This comprehensive list of standards used by paint industry also includes a brief discussion of each method of testing, including sampling, specific gravity, color, distillation range, nonvolatile matter, odor, water, acidity, alkalinity, ester value, copper corrosion test, sulfur, permanganate time test for acetone and methanol, flash point, purity of ketones, solvent power evaluation, water miscibility, analysis of methanol, analysis of ethylene and propylene glycols, acid wash color of aromatic hydrocarbons, paraffins and other nonaromatic hydrocarbons in aromatics, and aromatics in mineral spirits. The nonvolatile matter in paints is determined by a gravimetric method after drying a 100 ml sample in oven at 105oC.75 The transfer efficiency of paints is a volume or weight ratio of paint solids deposited to the volume/weight of the paint solids sprayed, expressed in percent. This method76 can be used to optimize the paint application process. The measurement is done by weighing or measuring the volume of paint used on a certain sprayed surface area and comparing this value with known or predetermined by the above method weight of solids in the paint used for spraying. In order to determine an ester value for solvents or thinners, the specimen is reacted with aqueous potassium hydroxide, using isopropanol as the mutual solvent.77 The hydrolysis is conducted at 98oC and the excess potassium hydroxide is determined by titration. From the amount of potassium hydroxide consumed, the ester value is calculated. 15.1.24 pH A method developed for halogenated solvents is applicable for determining the pH of water extracts of solvents.78 The solvent sample is shaken with distilled or deionized water and the pH is determined either by comparing color upon the addition of Gramercy universal indicator or by using a glass electrode pH meter. 15.1.25 PURITY Several techniques are used to determine purity of solvents. Gas chromatography is the most common and this and other methods are discussed first followed by other analytical methods which include instrumental and simple methods. The aim of these tests is to determine the concentration of the main component but more frequently qualitative and quantitative determination of admixtures. Some methods have already been discussed in Sections 15.1.1, 15.1.5, 15.1.8, 15.1.9, 15.1.18, 15.1.20 and 15.1.23. Alcohol content and purity of an acetate ester is determined by gas chromatography.79 The method was applied to ethyl, n-propyl, isopropyl, n-butyl, isobutyl, and 2-ethylhexyl acetates. Water and acetic acid cannot be measured by this method and other methods are
1064
George Wypych
used. A thermal conductivity or flame ionization detector is used. A stainless steel column with 80-100 mesh Chromosorb G-HP is used with 9.05% Dow Corning QF-1 and 0.45% Igepal CO-990. The concentration of the main component and the amount of free alcohol are measured by the method. Traces of benzene in hydrocarbon solvents are measured by capillary gas chromatography.80 Because of the hazardous nature of benzene, the method was introduced to ensure compliance with the stringent regulations. A flame ionization detector is used with 0.53 mm fused silica capillary columns with bound methyl silicone or polyethylene glycol. Similar method is used to determine benzene content in cyclic products (cyclohexane, toluene, cumene, styrene, etc.).81 This method does not specify any particular column but the column used should be able to resolve benzene from other components. The method can determine benzene in concentrations of 5 to 300 mg/kg. Traces of thiophene in refined benzene are determined by a flame photometric detector.82 Several column types given in standard are found satisfactory to overcome potential problem of quenching effects of hydrocarbons on the light emissions from thiophene. High purity benzene for cyclohexane feedstock is tested for several known impurities by capillary gas chromatography.83 The gas chromatograph is equipped with a flame ionization detector and a splitter injector suitable for fused silica capillary column internally coated with crosslinked methyl silicone. The concentration of benzene and the concentrations of impurities can be adequately determined. The purity of halogenated solvents is determined using a thermal conductivity or flame ionization detector, a column made from 3.2 mm stainless tubing packed with 30 wt% silicone fluid on 80-100 mesh diatomaceous earth or using capillary column.84 Admixtures in 1,1,1-trichloroethane are determined using a thermal conductivity or hydrogen flame detector. Column from copper or stainless steel is packed with Chromosorb W HP with 20% polydimethylsiloxane.85 Various impurities, such as hydrocarbons, acetone, alcohols and other can be determined using a thermal conductivity or a flame detector.86 Several columns are specified in the standard mostly using polyethylene glycol on diatomaceous earth. For determination of purity of methyl isobutyl ketone different method is used.87 In both cases the amounts of determined impurities are subtracted from total mass to give purity of specimen. The determination of aromatics in mineral spirits is another method that has been developed to ensure compliance with regulations restricting aromatic content.88 Three methods are given, each capable to determine ethylbenzene and total aromatic content. The methods differ in column type and packing. A spectrophotometric method89 for determining thiophene in benzene is available as alternative to gas chromatography.82 The spectrometer used is capable of detecting absorbance in the range from 400 to 700 nm with a repeatability of 0.005 absorbance units. Thiophene is reacted with isatin to form a colored compound. The quantitative determination is based on reading concentrations from master curve. The presence of reducing substances in pyridine can be detected by a simple visual observation.90 The sample of pyridine is mixed with 0.32% potassium permanganate solution and color is observed after 30 min. If color of the potassium permanganate is retained, the sample is free of reducing agents. Oxidative microcoulometry is used to determine trace quantities of sulphur in aromatic hydrocarbons.91 An oxidative pyrolysis converts sulfur to sulfur dioxide which is titrated in titration cell with the triiodide ion present in the electrolyte. The total chloride (organic and inorganic) in aromatic hydrocarbons and cyclohexane can be determined by titration.92 Bromides and iodides present are recorded as chlorides. The sample is mixed with toluene in a proportion which depends on the expected concentration of chloride. The reagent sodium biphenyl is added to convert organic halogens into in-
15.1 Standard methods of solvent analysis
1065
organic halides. After decomposing the excess reagent with water, the separated aqueous phase is titrated in presence of acetone with a silver nitrate solution. Organic chlorides present in aromatic hydrocarbons can also be determined by microcoulometry.93 The presence of chlorine compounds may adversely affect equipment and/or reaction therefore their concentration is frequently controlled. A liquid specimen is injected into a combustion tube maintained at 900oC, converted to hydrogen halides, and carried by a carrier gas (50% oxygen, 50% argon) to a titration cell where it reacts with silver ions in the electrolyte. The determination of peroxides has two goals: one is to monitor peroxide concentration used as initiator and catalysts and the other is to detect formation of hazardous peroxides formed as autoxidation products in ethers, acetals, dienes, and alkylaromatic hydrocarbons. A sample is dissolved in a mixture of acetic acid and chloroform. The solution is deaerated and potassium iodide reagent is added and let to react for 1 h in darkness.94 The iodine formed in reaction is measured by absorbance at 470 nm and result calculated to active oxygen in the sample. The method can determine hydroperoxides, peroxides, peresters, and ketone peroxides. Oxidizing and reducing agents interfere with the determination. Mercaptans in motor fuels, kerosene and other petroleum products can be detected by shaking the liquid sample with sodium plumbite solution, adding powdered sulfur and shaking again.95 If a mercaptan or hydrogen sulfide is present, discoloration of the floating sulfur or liquid phase occurs. The nonvolatile content of a halogenated solvent is determined by drying the sample in a platinum evaporating dish at 105oC.96 Depending on boiling point and the concentration of nonvolatile matter three alternate procedures are proposed. The Karl Fisher method is recommended for general use in solvents to determine the water content.97 It is not suitable if mercaptans, peroxides, or appreciable quantities of aldehydes and amines are present. Water in halogenated solvents may cause corrosion, spotting, reduce shelf-life of aerosols, or inhibit chemical reactions, thus special method, also based on the Karl Fischer titration, was developed for halogenated solvents.98 The titration of ionizable chlorides with mercuric acetate solution in the presence of s-diphenylcarbazone as an indicator is used to determine chloride in trichlorotrifluoroethane.99 A visual appearance test to detect admixtures in halogenated solvents is based on the observation of suspended particles, sediment, turbidity and free floating water.100 The aromatic content of hydrocarbon mixture is estimated from the determination of aniline point.101 Aromatic hydrocarbons have the lowest and paraffins the highest aniline points. Cycloparaffins and olefins are between the two. Aniline point increases as the molecular weight increases. A mixture of specific aniline and solvent is heated at a controlled rate until it forms one phase. The mixture is then cooled and the temperature at which the miscible liquid separates into two phases is determined. Four methods are discussed in the standard101 suitable for transparent, non-transparent, easily vaporizing, and measured in small quantities. The presence of oxidizable materials in acetone and methanol that are associated with contaminations during manufacture and distribution can be evaluated by permanganate time.102 Oxidizable contaminants may adversely affect catalysts or ligand complexes which are sensitive to oxidation. Oxidizable substances reduce potassium permanganate to manganese oxide which is yellow. The method is designed to measure the time required to change color to the color of a standard. Small admixtures of acetone in methanol (more than 0.003 wt%) can be detected after a reaction with Nessler’s reagent.103 The reacted sample is compared with a standard, which contains 0.003 wt% acetone and the difference in turbidity is reported.
1066
George Wypych
15.1.26 REFRACTIVE INDEX Refractive index is measured by a standard method.104 It covers transparent and light colored liquids having a refractive index in the range from 1.33 to 1.50. The refractive index is the ratio of light velocity of a specified wavelength in air to its velocity in the substance under evaluation. The refractive dispersion is the difference between refractive indices for light of two different wavelengths. This value is usually multiplied by 10,000. The method uses a Bausch & Lomb refractometer equipped with a thermostat and with a circulating bath to control the sample temperature with a precision of 0.2oC. Several light sources can be used, including the sodium arc lamp, mercury light lamp, and hydrogen or helium discharge lamp. Light sources are equipped with filters which transmit a specific spectral line. Standardization of equipment is done using a solid reference standard or using liquids standards such as 2,2,4-trimethylpentane (1.39), methylcyclohexane (1.42), or toluene (1.49). 15.1.27 RESIDUAL SOLVENTS Residual solvents may cause odor, off-taste, blocking, and an increased degradation rate in outdoor exposures. A single standard test method has been developed to determine residual solvent levels and it is primarily used for the evaluation of flexible barrier materials.105 The method is based on gas chromatography. The specimen of the barrier material is enclosed in a container and heated to vaporize the retained solvents into the head space. The vapor from the head space is taken by a gas syringe and injected into a gas chromatograph. The recovery of solvents is compared by the means of response factor which is a peak intensity of the detector in response to a given volume of injected sample. Response factors of different solutions vaporized in the test containers are compared. Round robin tests have demonstrated that this method has a coefficient of variation between laboratories of ±15%. The method does not specify the detector or the columns. Before a flexible barrier material is analyzed, the optimum heating time to recover volatilized solvents is determined. 15.1.28 SOLUBILITY Solvent power of a hydrocarbon solvents is determined by kauri-butanol value.106 The method applies to solvents having a boiling point above 40oC and a dry point below 300oC. The method is most frequently used to evaluate solvents for applications in paints and lacquers. The kauri-butanol value is the volume of solvent required to produce a defined degree of turbidity when added to 20 g of a standard solution of kauri resin in n-butanol (400 g of resin in 2000 g of solvent). High kauri value indicates a relatively strong dissolving power. The method is standardized using 105±5 ml of toluene. The solubility of common gases in hydrocarbon liquids is determined to meet requirements of aerospace industry.107 This test method is based on the Clausius-Clapeyron equation, Henry’s law, and the perfect gas law. The results are important in the lubrication of gas compressors where dissolved gas may cause erosion due to cavitation. In fuels, dissolved gases may cause interruption of fuel supply and foaming in tank. The liquid density is determined experimentally. Using this density, the Ostwald coefficient is taken from a chart and used for the calculation of the Bunsen coefficient (solubility of gas). The solubility of the gas or mixture of gases and Henry’s law constant are also calculated. 15.1.29 SOLVENT PARTITIONING IN SOILS A procedure is available to determine partitioning organic chemicals between water and soil or sediment.108 By measuring sorption coefficients for specific solids, a single value is obtained which can be used to predict partitioning under a variety of conditions. The
15.1 Standard methods of solvent analysis
1067
underlining principle of the method is that organic chemicals bind with the surfaces of solids through chemical and physical interactions. However, the sorption coefficient is based on organic carbon content which does not apply to all solvents or all soils. The sorption coefficient of a particular solvent is measured by equilibrating its aqueous solution, containing a realistic concentration similar to that found in the environment, with a known quantity of soil or sediment. After equilibrium is reached, the concentration of solvent in the water and the soil is measured by a suitable analytical technique. The sorption constants for all solids tested are averaged and reported as a single value. The standard does not define the actual method of determining the concentrations but strategy that should be followed. The data are useful in predicting the migration of chemicals in soil, in estimating their volatility from water and soil, determining their concentration in water, and their propensity to leach through the soil profile. 15.1.30 SOLVENT EXTRACTION Often materials must be extracted from a compounded product to perform testing. Extracted material must then be recovered from solution without degradation to be subjected to testing. The method discussed here was developed as means of recovering asphalt from pavement samples.109 The solution of extracted asphalt in solvent is distilled by rotating the distillation flask of a rotary evaporator in a heated oil bath. The distillation is carried out under partial pressure in the presence of nitrogen to prevent degradation. The asphalt recovered by this method can be tested in the same manner as were the original asphalt samples. 15.1.31 SPECIFICATIONS Standard specifications are designed to set criteria for commercial solvents which can be used to determine the compliance of a solvent sample. Because applications of solvents differ very widely, the selected criteria are also different for different groups of solvents. Table 15.1.1 is a compiled list of parameters all of which can be found in solvent specifications. The most common parameters used to characterize solvent include acidity, appearance, color, concentration of main component, distillation range, dry point, initial boiling point, and specific gravity. The methods of determining these parameters are found in this chapter. The list of references includes information on specifications for various solvents. Solvents in this list are arranged into groups: alcohols,110-120 aromatic hydrocarbons,121-128 other hydrocarbons,129-132 ketones,133-138 esters,139-145 glycol,146 and chlorine-containing solvents.147-152 15.1.32 SUSTAINED BURNING The sustained burning test was originally developed for British Standard BS-3900 and adapted by ASTM.153 The purpose of the test is to determine the sustained burning characteristics of solvents by direct experiment rather than by deducing characteristics from flash point data. Mixtures of some flammable liquids (e.g., alcohol and water) are classified as flammable based on the closed-cup flash point method. Some mixtures may be classed as flammable even though they do not sustain burning. The test is performed in a block of aluminum with a concave depression called a well. The liquid under test is heated to a temperature of 49oC and a flame is passed over the well and held in position for 15 s. The specimen is observed to determine if it can sustain burning.
1068
George Wypych
Table 15.1.1. Typical chemical and physical properties of solvents included in their specifications Acid acceptance
Distillation range
Acidity
Doctor test
Acid wash color
Dry point
Alkalinity
Flash point
Aluminum scratch
Free halogens
Aniline point
Initial boiling point
Appearance
Iron concentration
Bromine index
Kauri-butanol value
Color
Non-volatile matter
Concentration of admixtures
Permanganate time
Concentration of isomers
Residual odor
Concentration of main component (or purity)
Specific gravity
Concentration of sulfur
Water concentration
Copper corrosion
Water miscibility
15.1.33 VAPOR PRESSURE The vapor pressure is the pressure of the vapor of a substance in equilibrium with the pure liquid at a given temperature. Two procedures are used: an isoteniscope procedure (standard) for measuring vapor pressures from 1x10-1 to 100 kPa and a gas-saturation pressure for measuring vapor pressures from 1x10-11 to 1 kPa.154-155 In the isoteniscope method, a sample is deaerated by heating under reduced pressure. The vapor pressure is then determined by balancing the pressure of the vapor against a known pressure of an inert gas. The vapor pressure is determined at minimum three different temperatures. In the gas-saturation method, an inert gas is passed through a sufficient amount of compound to maintain saturation. The vapor is then removed from the gas by a sorbent or a cold trap and quantitatively determined by gas chromatography or other suitable technique. 15.1.34 VISCOSITY The viscosity of solvents can be determined by one of three methods: glass viscometer,156 Saybolt viscometer,157 and bubble time method.158 Glass viscometry is applicable to Newtonian, transparent liquids which because of volatility cannot be measured in conventional capillary viscometers. The viscometer uses a purge gas which helps to transfer the measuring liquid from a lower reservoir to the sample bulb. The time of flow is measured for a fixed volume of liquid at a temperature controlled with a precision of 0.01oC. A set of liquids is available as viscosity standards in order to select the standard having closest viscosity to the measured sample. The Saybolt viscometer was developed for petroleum products. A sample of 60 ml flows through a calibrated orifice and the time of flow is measured at a controlled tempera-
15.1 Standard methods of solvent analysis
1069
ture. Measurements are made at a temperature selected from the range 21-99oC. The instrument is standardized by measuring the flow of a standard oil at both 37.8 and 50oC. In the bubble method, a standard viscosity tube is filled with a specimen liquid, the temperature of liquid is equilibrated to 25oC in a bath, the level of the meniscus is adjusted to 100 mm line, cork is inserted to end on 108 mm line, sample is hold in thermostating bath for another 20 min. Then the tube is inverted and the time for the bubble to flow from a mark at 27 mm to 100 mm is measured. 15.1.35 VOLATILE ORGANIC COMPOUND CONTENT, VOC Several terms are used in the paint industry to provide data on the VOC content of paints.159 The percent of solids in paints is calculated either per unit weight or per unit volume. The weight solids content is the weight of non-volatile materials divided by the total weight of the coating. In practice, it is calculated by subtracting the total weight of volatile solvents from the total weight of the coating and dividing the result by the total weight of coating. The final result is multiplied by 100 to express it in percent. Percent of solids by weight is calculated in similar manner. A solvent, as defined by paint standards, is a volatile liquid that is incorporated primarily for vehicle solvency and control of the application characteristics. The solvent content is the calculated weight of solvents in a specific volume of paint. This definition is not equivalent to the definition of VOC compound because it does include compounds which are excluded by EPA (see more on this subject in Section 18.1). Volatile organic compound by EPA definition means any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates, and ammonium carbonate, which participates in atmospheric photochemical reactions. This includes any such organic compound other than the following, which have been determined to have negligible photochemical reactivity: methane; ethane; methylene chloride; 1,1,1-trichloroethane; 1,1,2-trichloro-1,2,2-trifluoroethane; trichlorofluoromethane; dichlorodifluoromethane; chlorodifluoromethane; trifluoromethane; dichlorotetrafluoroethane; chloropentafluoroethane; dichlorotrifluoroethane; tetrafluoroethane; dichlorofluoroethane; chlorodifluoroethane; 2-chloro-1,1,1,2-tetrafluoroethane; pentafluoroethane; 1,1,2,2-tetrafluoroethane; 1,1,1-trifluoroethane; 1,1-difluoroethane; perchloroethylene; acetone; parachlorobenzotrifluoride; cyclic, branched, or linear completely methylated siloxanes; and perfluorocarbon compounds which fall into these classes: cyclic, branched, or linear, completely fluorinated alkanes; cyclic, branched, or linear, completely fluorinated ethers with no unsaturations; cyclic, branched, or linear, completely fluorinated tertiary amines with no unsaturations; and sulfur containing perfluorocarbons with no unsaturations and with sulfur bonds only to carbon and fluorine. For purposes of determining compliance with emissions limits, VOC is measured by the test methods in the approved State implementation plan (SIP) or 40 CFR Part 60, Appendix A, as published in (7/1/91) edition, as applicable. Based on the above definition of VOC, the VOC content is calculated from a formula by excluding from the total solvent content of the paint, the content of water and solvents excluded by the above regulation. The following formula can be used for VOC calculation: VOC =
W total solvents − Ww − W excluded solvents Vpaint − Vw − Vexcluded solvents
where: Wtotal solvents Ww
total weight of solvents weight of water present in formulation
[15.1.1]
1070
George Wypych
Wexcluded solvents Vpaint Vw Vexcluded solvents
total weight of solvents excluded by EPA regulation total volume of paint volume of water present in formulation total volume of solvents excluded by EPA regulation
The above definition may vary in various countries thus the proper definition should be obtained from the appropriate authorities. Examples of other definitions are included in Chapter 18.1. The weight percent of the volatile content in waterborne aerosol paints is determined by releasing the propellant from can, testing the remaining paint for water content using the Karl Fischer method, and determining non-volatiles.160 The VOC content in a solventborne automotive paints is determined by the simulation of VOC loss during baking using laboratory panels.161
REFERENCES 1 2
3
4 5 6 7 8 9
ASTM D 2106-95. Standard test method for determination of amine acid acceptance (alkalinity) of halogenated organic solvent. ASTM D 1614-95. Standard test method for alkalinity in acetone. ISO 755-2-81. Butan-1-ol for industrial use - Methods of test - Part 2: Determination of acidity Titrimetric method. ISO 756-2-81. Propan-2-ol for industrial use - Methods of test - Part 2: Determination of acidity Titrimetric method. ISO 757-2-82. Acetone for industrial use - Methods of test - Part 2: Determination of acidity to phenolphthalein - Titrimetric method. ASTM D 2989-97. Standard test method for acidity-alkalinity of halogenated solvents and their admixtures. ISO 1393-77. Liquid halogenated hydrocarbons for industrial use - Determination of acidity - Titrimetric method. ISO 3363-76. Fluorochlorinated hydrocarbons for industrial use - Determination of acidity - Titrimetric method. ASTM D 847-96. Standard test method for acidity of benzene, toluene, xylenes, solvent naphthas, and similar industrial aromatic hydrocarbons. ASTM D 1613-96. Standard test method for acidity in volatile solvents and chemical intermediates used in paint, varnish, lacquer, and related products. ASTM D 2942-96. Standard test method for total acid acceptance of halogenated organic solvents (nonreflux method). ASTM D 3444-95. Standard test method for total acid number of trichlorotrifluoroethane. ASTM E 659-94. Standard test method for autoignition temperature of liquid chemicals. AS 1896-1976. Method of test for ignition temperature of gases and vapors. ASTM E 1196-92. Standard test method for determining the anaerobic biodegradation potential of organic chemicals. BS 6068-2.34-88. Water quality. Physical, chemical and biochemical methods. Method for the determination of the chemical oxygen demand. DIN 38409-41-80. German standard methods for examination of water, waste water and sludge. Summary action and material characteristic parameters (group H). Determination of the chemical oxygen demand (COD) in the range over 15 mg/l. DIN 38409-43-81. German standard methods for the analysis of water, waste water and sludge. Summary action and material characteristic parameters (group H). Determination of the chemical oxygen demand (COD). Short duration method. ISO 5815-89. Water quality - Determination of biochemical oxygen demand after 5 days - Dilution and seeding method. ISO 6060-89. Water quality - Determination of the chemical oxygen demand. JIS K 0400-20-10-99. Water quality - Determination of the chemical oxygen demand. JIS K 0400-21-10-99. Water quality - Determination of biochemical oxygen demand after 5 days. Dilution seeding method. JIS K 3602-90. Apparatus for the estimation of biochemical oxygen demand with microbial sensor.
15.1 Standard methods of solvent analysis
10
11
12 13
14
15 16
17 18 19 20 21 22 23
24
1071
ASTM D 850-93. Standard test method for distillation of industrial aromatic hydrocarbons and related materials. BS 7392-90. Method for determination of distillation characteristics of petroleum products (ISO 3405). ISO 918-83. Volatile organic liquids for industrial use - Determination of distillation characteristics. ISO 3405-88. Petroleum products - Determination of distillation characteristics. ISO 4626-80. Volatile organic liquids - determination of boiling range of organic solvents used as raw materials. JIS K 2254-98. Petroleum products - Determination of distillation characteristics. ASTM D 1078-97. Standard test method for distillation range of volatile organic liquids. BS 4591-90. Method of determination of distillation characteristics of organic liquids (other than petroleum products) (ISO 918). JIS K 0066-92. Test methods for distillation of chemical products. ASTM D 3272-98. Standard practice for vacuum distillation of solvents from solvent-reducible paints for analysis. ASTM D 5399-95. Standard test method for boiling point distribution of hydrocarbon solvents by gas chromatography. ISO 3924-99. Petroleum products - Determination of boiling range distribution - Gas chromatography method. ASTM D 5776-98. Standard method for bromine index of aromatic hydrocarbons by electrometric titration. ISO 3839-96. Petroleum products - Determination of bromine number of distillates and aliphatic olefins Electrometric method. JIS K 2605-96. Petroleum distillates and commercial aliphatic olefins - Determination of bromine number - Electrometric method. ASTM D 1492-96. Standard test method for bromine index of aromatic hydrocarbons by coulometric titration. ASTM D 240-97. Standard test method for heat of combustion of liquid hydrocarbon fuels by bomb calorimeter. BS 7420-91. Guide for determination of calorific values of solid, liquid and gaseous fuels (including definitions). DIN 5499-72. Gross and net Calorific values. Terms. DIN 51900-1-89. Determination of gross calorific value of solid and liquid fuels by the bomb calorimeter and calculation of net calorific value. Part 1: Principles, apparatus, methods. DIN 51900-2-77. Testing of solid and liquid fuels. Determination of gross calorific value by bomb calorimeter and calculation of net calorific value. Method of using isothermal water jacket. DIN 51900-3-77. Testing of solid and liquid fuels. Determination of gross calorific value by bomb calorimeter and calculation of net calorific value. Method of using adiabatic jacket. JIS K 2279-93. Crude petroleum and petroleum products - Determination and estimation of heat of combustion. ASTM D 1405-95a. Standard test method for estimation of net heat of combustion of aviation fuels. ASTM D 5468-95. Standard test method for gross calorific value of waste materials. ASTM D 5468-95. Standard test method for gross calorific and ash value of waste materials. ASTM F 1105-95. Standard test method for preparing aircraft cleaning compounds, liquid-type, temperature sensitive, or solvent-based, for storage stability testing. ASTM D 3698-92. Standard practices for solvent vapor degreasing operations. ASTM D 4579-96. Standard practice for handling an acid degreaser or still. ASTM D 848-97. Standard test method for acid wash color of industrial aromatic hydrocarbons. ISO 5274-79. Aromatic hydrocarbons - Acid wash test. ISO 755-3-81. Butan-1-ol for industrial use - Methods of test - Part 3: Sulfuric acid color test. JIS K 0072-98. Testing method of color after treatment with sulfuric acid. ASTM D 1686-96. Standard test method for color of solid aromatic hydrocarbons and related materials in the molten state (platinum-cobalt scale). BS 5339-76. Method of measurement of color in Hazen units (platinum-cobalt scale) of liquid chemical products (ISO 2211). ISO 2211-73. Liquid chemical products - Measurement of color in Hazen units (platinum-cobalt scale). ISO 6271-97. Clear liquids - Estimation of color by the platinum-cobalt scale. JIS K 0071-98. Testing methods for color of chemical products - Part 1. Estimation of color in Hazen units (platinum-cobalt scale).
1072
25 26 27 28
29 30 31
32 33 34
35 36 37 38 39 40 41 42 43 44 45 46 47
48 49
50
51 52 53
George Wypych
ASTM D 2108-97. Standard test method for color of halogenated organic solvents and their admixtures (platinum-cobalt scale). ASTM D 5386-93b. Standard test method for color of liquids using tristimulus colorimetry. ASTM D 3316-96. Standard test method for stability of perchloroethylene with copper. ASTM D 849-97. Standard test method for copper strip corrosion by industrial aromatic hydrocarbons. ISO 2160-98. Petroleum products - Corrosiveness of copper - Copper strip test. JIS K 2513-91. Petroleum products - Corrosiveness to copper - Copper strip test. SFS-EN ISO 2160. Petroleum products. Corrosiveness to copper. Copper strip test. ASTM D 2943-96. Standard test method for aluminum scratch of 1,1,1-trichloroethane to determine stability. ASTM D 1217-98. Standard test method for density and relative density (specific gravity) of liquids by Bingham pycnometer. ASTM D 891-95. Standard test method for specific gravity, apparent, of liquid industrial chemicals. BS 4522-88. Method for determination of absolute density at 20oC of liquid chemical products for industrial use (ISO 758). JIS K 0061-92. Test methods for density and relative density of chemical products. ASTM D 2111-95. Standard test methods for specific gravity of halogenated organic solvents and their admixtures. ASTM D 1555-95. Standard test method for calculation of volume and weight of industrial aromatic hydrocarbons. ASTM D 1555M-95. Standard test method for calculation of volume and weight of industrial aromatic hydrocarbons (metric). ISO 5281-80. Aromatic hydrocarbons - Benzene, xylene and toluene - Determination of density at 20oC. SFS 3773-76. Determination of density of liquids at 20oC. ASTM D 1720-96. Standard test method for dilution ratio of active solvents in cellulose nitrate solutions. ASTM D 5062-96. Standard test method for resin solution dilutability by volumetric/gravimetric determination. ASTM D 1476-96. Standard test method for heptane miscibility of lacquer solvents ASTM D 1722-94. Standard test method for water miscibility of water-soluble solvents. ASTM D 5226-98. Standard practice for dissolving polymer materials. ASTM D 2257-96. Standard test method for extractable mater in textiles. ASTM F 58-96. Standard test method for measuring specific resistivity of electronic grade solvents. JIS C 0052-95. Environmental testing procedure of electronic and electrical resistance to solvents. ASTM D 4308-95. Standard test method for electrical conductivity of liquid hydrocarbons by precision meter. ASTM D 1816-97. Standard test method for dielectric breakdown voltage of insulating oils of petroleum origin using VDE electrodes. ASTM D 3300-94. Standard test method for dielectric breakdown voltage of insulating oils of petroleum origin under impulse conditions. ASTM F 791-96. Standard test method for stress crazing of transparent plastics. ASTM D 4508-98. Standard test method for chip impact strength of plastics. ASTM D 1901-95. Standard test method for relative evaporation time of halogenated organic solvents and their admixtures. DIN 53170-91. Solvents for paints and similar coating materials. Determination of the evaporation rate. ASTM E 1232-96. Standard test method for temperature limit of flammability of chemicals. ASTM E 681-98. Test method for concentration limits of flammability of chemicals. BS 6713-3-86. Explosion protection systems. Method for determination of explosion indices of fuel/air mixtures other than dust/air and gas/air mixtures (ISO 6184-3). ASTM D 3828-97. Standard test method for flash point by small scale closed tester. AS/NZS 2106.0-99. Methods for determination of flash point of flammable liquids (closed cup) General. AS/NZS 2106.1-99. Methods for the determination of the flash point of flammable liquids (closed cup) Abel closed cup method. ASTM D 1310-97. Standard test method for flash point and fire point of liquids by tag open-cup apparatus. ASTM D 56-97a. Standard test method for flash point by tag closed tester. ASTM D 92-97. Standard test method for flash and fire points by Cleveland open cup. ISO 2592-73. Petroleum product - Determination of flash and fire points - Cleveland open cup method.
15.1 Standard methods of solvent analysis
54
55
56 57 58 59 60 61
62 63 64 65
66 67
1073
SFS-EN 22592. Petroleum products. Determination of flash and fire points. Cleveland open cup method (ISO 2592). ASTM D 93. Standard test methods for flash-point by Pensky-Martens closed cup tester. AS/NZS 2106.2-99. Methods for determination of the flash point of flammable liquids (closed cup) Pensky-Martens closed cup method. DIN EN 22719-93. Petroleum products and lubricants. Determination of flash point. Pensky-Martens closed cup method (ISO 2719). ISO 2719-88. Petroleum products and lubricants - Determination of flash point - Pensky-Martens closed cup method. SFS-EN 22719. Petroleum products and lubricants. Determination of flash point. Pensky-Martens closed cup method (ISO 2719) ASTM D 3941-96. Standard test method for flash point by the equilibrium method with a closed-cup apparatus. AS/NZS 2106.3-99. Methods for the determination of the flash point of flammable liquids (closed cup) Flash/no flash test - Rapid equilibrium method. AS/NZS 2106.4-99. Methods for the determination of the flash point of flammable liquids (closed cup) Determination of flash point - Rapid equilibrium method. AS/NZS 2106.5-99. Methods for the determination of the flash point of flammable liquids (closed cup) Flash/no flash test - Closed cup equilibrium method. AS/NZS 2106.6-99. Methods for the determination of the flash point of flammable liquids (closed cup) Determination of flash point - Closed cup equilibrium method. DIN EN 456-91. Paints, varnishes and related products. Determination of flash point. Rapid equilibrium method (ISO 3679). ISO 1516-81. Paints, varnishes, petroleum and related products - Flash/no flash test - Closed cup equilibrium method. ISO 1523-83. Paints, varnishes, petroleum and related products - Determination of flash point - Closed cup equilibrium method. ISO 3679-83. Paints, varnishes, petroleum and related products - Determination of flash point - Rapid equilibrium method. ISO 3680-83. Paints, varnishes, petroleum and related products - Flash/no flash test - Rapid equilibrium method. SFS-EN 456-93. Paints, varnishes, and related products. Determination of flash point. Rapid equilibrium method (ISO 3679). ASTM E 502-94. Standard test method for selection and use of ASTM standards for the determination of flash point of chemicals by closed cup methods. ASTM D 1015-94. Standard test method for freezing points of high-purity hydrocarbons. JIS K 0065-92. Test method for freezing point of chemical products. ASTM D 1016-94. Standard test method for purity of hydrocarbons from freezing points. ASTM D 852-97. Standard test method for solidification point of benzene. ISO 5278-80. Benzene - Determination of crystallizing point. ASTM D 4755-95. Standard test method for free halogenated organic solvents and their admixtures. ASTM D 2306-96. Standard test method for C8 hydrocarbon analysis by gas chromatography. DIN 51437-89. Testing of benzene and benzene homologues. Determination of the content of non-aromatics, toluene and C8-aromatics in benzene. Gas chromatography. DIN 51448-1-97. Testing of liquid petroleum hydrocarbons. Determination of hydrocarbon types. Part 1: Gas chromatographic analysis by column switching procedure. ASTM D 3797-96. Standard test method for analysis of o-xylene by gas chromatography. ASTM D 3798-96b. Standard test method for analysis of p-xylene by gas chromatography. ASTM D 4492-98. Standard test method for analysis of benzene by gas chromatography. ASTM D 3271-87-93. Standard test method for direct injection of solvent reducible paints into a gas chromatograph for solvent analysis. DIN 55682-94. Solvents for paints and varnishes. Determination of solvents in water-thinnable coating materials. Gas chromatographic method. (Amendment A1 - DIN 55682/A1-98). DIN 55683-94. Solvents for paints and varnishes. Determination of solvents in coating materials containing organic solvents only. Gas chromatographic method. ASTM D 5830-95. Standard test method for solvent analysis in hazardous waste using gas chromatography. ASTM D 4757-97. Standard practice for placarding solvent vapor degreasers.
1074
68 69 70 71 72 73 74 75 76 77 78 79
80 81 82
83 84
85 86 87 88 89 90 91
92 93 94 95
George Wypych
ASTM D 3844-96. Standard guide for labeling halogenated hydrocarbon solvent containers. ASTM D 1296-96. Standard test method for odor of volatile solvents and diluents. ISO 2498-74. Methyl ethyl ketone for industrial use - Examination for residual odor. ASTM E 679-97. Standard practice for determination of odor and taste thresholds by forced-choice ascending concentration series method. ASTM E-769-97. Standard test methods for odor of methanol, ethanol, n-propanol, and isopropanol. ASTM E 1075-97. Standard test methods for odor of ethylene glycol, diethylene glycol, triethylene glycol, propylene, glycol, and dipropylene glycol and taste of propylene glycol. ASTM D 4494-95. Standard test method for detecting residual odor of drycleaning grade perchloroethylene. ASTM D 268-96. Standard guide for sampling and testing volatile solvents and chemical intermediates for use in paint and related coatings and material. ASTM D 1353-96. Standard test method for nonvolatile matter in volatile solvents for use in paint, varnish, lacquer, and related products. ASTM D 5286-95. Standard test methods for determination of transfer efficiency under general conditions for spray application of paints. ASTM D 1617-96. Standard test method for ester value of solvents and thinners. ASTM D 2110-96. Standard test method for pH of water extractions of halogenated organic solvents and their admixtures. ASTM D 3545-95. Standard test method for alcohol content and purity of acetate esters by gas chromatography. DIN 55686-92. Solvents for paints and varnishes. Acetic esters. Gas chromatographic determination of the degree of purity. ASTM D 6229-98. Standard test method for trace benzene in hydrocarbon solvents by capillary gas chromatography. ASTM D 4534-93. Standard test method for benzene content of cyclic products by gas chromatography. ASTM D 435-96. Standard test method for determination of trace thiophene in refined benzene by gas chromatography. DIN 51438-91. Testing of benzene and benzene homologues. Determination of the content of thiophene in benzene. Gas chromatography. ASTM D 5713-96. Standard test method for analysis of high purity benzene for cyclohexane feedstock by capillary gas chromatography. ASTM D 3447-96. Standard test method for purity of halogenated organic solvents. CAN/CGSB 1-GP-71 No. 73-96. Methods of testing paints and pigments. Beilstein test for chlorinated solvents. ASTM D 3742-94. Standard test method for 1,1,1-trichloroethane content. ASTM D 2804-98. Standard test method for purity of methyl ethyl ketone by gas chromatography. ASTM D 3329-94a. Standard test method for purity of methyl isobutyl ketone by gas chromatography. ASTM D 3257-97. Standard test methods for aromatics in mineral spirits by gas chromatography. ASTM D 1685-95. Standard test method for traces of thiophene in benzene by spectrophotometry. ASTM D 2031-97. Standard test method for reducing substances in refined pyridine. ASTM D 3961-98. Standard test method for trace quantities of sulfur in liquid aromatic hydrocarbons by oxidative microcoulometry. ISO 5282-82. Determination of sulfur content - Pitt-Ruprecht reduction and spectrophotometric determination method. ISO 8754-92. Petroleum products - Determination of sulfur content - Energy-dispersive X-ray fluorescence method. SFS-EN ISO 8754. Petroleum products. Determination of sulfur content. Energy-dispersive X-ray fluorescence method. ISO 14596-98. Petroleum products - Determination of sulfur content - Wavelength-dispersive X-ray fluorescence method. ASTM D5194-96. Standard test method for trace chloride in liquid aromatic hydrocarbons. ASTM D 5808-95. Standard test method for determining organic chloride in aromatic hydrocarbons and related chemicals by microcoulometry. ASTM E 299-97. Standard test method for trace amounts of peroxides in organic solvents. ASTM D 4952-97. Standard test method for qualitative analysis for active sulfur species in fuels and solvents (Doctor Test). ISO 5275-79. Aromatic hydrocarbons - Test for presence of mercaptans (thiols) - Doctor test.
15.1 Standard methods of solvent analysis
96
97
98 99 100 101
102 103 104
105 106 107 108 109 110
111 112 113
114
115
1075
ASTM D 2109-96. Standard test methods for nonvolatile matter in halogenated organic solvents and their admixtures. BS 5598-3-79. Methods of sampling and test for halogenated hydrocarbons. Determination of residue on evaporation (ISO 2210). ISO 759-81. Volatile organic liquids for industrial use - Determination of dry residue after evaporation on water bath - General method. ISO 2210-72. Liquid halogenated hydrocarbons for industrial use - Determination of residue on evaporation. ISO 5277-81. Aromatic hydrocarbons - Determination of residue on evaporation of products having boiling points up to 150oC. JIS K 0067-92. Test methods for loss and residue of chemical products. SFS 3773-77. Determination of residue on evaporation. ASTM D 1364-95. Standard test method for water in volatile solvents (Karl Fischer reagent titration method). BS 2511-70. Methods for determination of water (Karl Fischer method). ISO 6191-81. Light olefins for industrial use - Determination of traces of water - Karl Fischer method. ISO 8917-88. Light olefins for industrial use - Determination of water - Guidelines for use of in-line analyzers. SFS 3774-76. Determination of water by the Karl Fischer method. ASTM D 3401-97. Standard test methods for water in halogenated organic solvents and their admixtures. ASTM D 3443-95. Standard test method for chloride in trichlorotrifluoroethane. ASTM D 3741-95. Standard test methods for appearance of admixtures containing halogenated organic solvents. ASTM D 611-98. Standard test methods for aniline point and mixed aniline point of petroleum products and hydrocarbon solvents. ISO 2977-97. Petroleum products and hydrocarbon solvents - Determination of aniline point and mixed aniline point. ASTM D 1363-97. Standard test method for permanganate time of acetone and methanol. ISO 757-7-83. Acetone for industrial use - Methods of test - Part 4: Determination of permanganate time. ASTM D 1612-95. Standard test method for acetone in methanol. ASTM D 1218-96. Standard test method for refractive index and refractive dispersion of hydrocarbon liquids. ISO 5661-83. Petroleum products - Hydrocarbon liquids - Determination of refractive index. JIS K 0062-92. Test methods for refractive index of chemical products. ASTM F 151-97. Standard test method for residual solvents in flexible barrier materials. ASTM D 1133-97. Standard test method for Kauri-butanol value of hydrocarbon solvents. ASTM D 2779-97. Standard test method for estimation of solubility of gases in petroleum liquids. ASTM E 1195-93. Standard test method for determining a sorption constant, Koc, for an organic chemical in soil and sediments. ASTM D 5404-97. Standard practice for recovery of asphalt from solution using the rotary evaporator. ASTM D 304-95. Standard specification for n-butyl alcohol (butanol). BS 508-1-86. Butan-1-ol for industrial use. Specification for butan-1-ol. BS 508-2-84. Butan-1-ol for industrial use. Methods of test. JIS K 1504-93. Butanol. JIS K 8810-96. 1-butanol. ASTM D 319-95. Standard specification for amyl alcohol (synthetic). ASTM D 331-95. Standard specification for 2-ethoxyethanol. ASTM D 770-95. Standard specification for isopropyl alcohol. BS 1595-1-86. Propan-2-ol (isopropyl alcohol) for industrial use. Specification for propan-2-ol (isopropyl alcohol). BS 1595-2-84. Propan-2-ol (isopropyl alcohol) for industrial use. Methods of test. ISO 756-1-81. Propan-2-ol for industrial use - Methods of test - Part 1: General. JIS K 1522-78. Isopropyl alcohol. ASTM D 1007-95. Standard specification for sec-butyl alcohol. ISO 2496-73. sec-butyl alcohol for industrial use - List of methods of test. JIS K 1523-78. 2-butanol. ASTM D 1152-97. Standard specification for methanol (methyl alcohol). BS 506-1-87. Methanol for industrial use. Specification for methanol.
1076
116 117 118
119 120 121
122
123 124 125 126 127 128 129 130
131 132 133 134 135
136
137 138 139 140 141
George Wypych
BS 506-2.84. Methanol for industrial use Methods of test. CAN/CGSB-3-GP-531M. Methanol, technical. ISO 1387-82. Methanol for industrial use - Methods of test. JIS K 1501-93. Methanol. ASTM D 1719-95. Standard specification for isobutyl alcohol. ASTM D 1969-96. Standard specification for 2-ethylhexanol (synthetic). BS 1835-91. Specification for 2-ethylhexan-1-ol for industrial use. ASTM D 2627-97. Standard specification for diacetone alcohol. BS 549-70. Specification for diacetone alcohol. ISO 2517-74. Diacetone alcohol for industrial use - List of methods of test. ASTM D 3128-97. Standard specification for 2-methoxyethanol. ASTM D 3622-95. Standard specification for 1-propanol (n-propyl alcohol). JIS K 8838-95. 1-propanol. ASTM D 841-97. Standard specification for nitration grade toluene. AS 3529-1988. Solvent - Toluene. ISO 5272-79. Toluene for industrial use - Specification. JIS K 8680-96. Toluene. ASTM D 843-97. Standard specification for nitration grade xylene. ISO 5280-79. Xylene for industrial use - specification. JIS K 8271-96. Xylene. ASTM D 4734-98. Standard specification for refined benzene-545. ASTM D 5136-96. Standard specification for high purity p-xylene. ASTM 5211-97. Standard specification for xylenes for p-xylene feedstock. ASTM D 5471-97. Standard specification for o-xylene 980. ASTM D 5871-98. Standard specification for benzene for cyclohexane feedstock. ISO 5721-79. Benzene for industrial use - Specification. ASTM D 3734-96. Standard specification for high-flash aromatic naphthas. ASTM D 5309-97. Standard specification for cyclohexane 999. ASTM D 235-95. Standard specification for mineral spirits (petroleum spirits) (hydrocarbon dry cleaning solvent). AS 35-1988. Solvents - Mineral turpentine and white spirit. BS 245-76. Specification for mineral solvents (white spirit and related hydrocarbon solvents) for paints and other purposes. CAN/CGSB-53.7-92. Duplicating liquid, Direct process, Spirit type. ASTM D 1836-94. Standard specification for commercial hexanes. ASTM D 3735-96. Standard specification for VM&P naphthas. ASTM D 740-97. Standard specification for methyl ethyl ketone. ASTM D 4360-90. Standard specification for n-amyl ketone. ASTM D 329-95. Standard specification for acetone. BS 509-1-87. Acetone for industrial use. Specification for acetone. BS 509-2-84. Acetone for industrial use. Methods of test. CAN/CGSB-15.50-2. Technical grade acetone. ISO 757-1-82. Acetone for industrial use - Methods of test - Part 1: General. JIS K 1503-59. Acetone. JIS K 8035-95. Acetone. SFS 3775 ISO/R 757. Acetone for industrial use. Methods of test. ASTM D 740-94. Standard specification for methyl ethyl ketone. CAN/CGSB-15.52-92. Technical grade methyl ethyl ketone. ISO 2597-73. Methyl ethyl ketone for industrial use - List of methods of test. JIS K 1524-78. Methyl ethyl ketone. ASTM D 1153-94. Standard specification for methyl isobutyl ketone. ISO 2499-74. Methyl isobutyl ketone for industrial use - List of methods of test. ASTM D 2917-94. Standard specification for methyl isoamyl ketone. ISO 2500-74. Ethyl isoamyl ketone for industrial use - List of methods of test. ASTM D 5137-95. Standard specification for hexyl acetate. ASTM D 1718-94. Standard specification for isobutyl acetate (95% grade). BS 551-90. Specification for butyl acetate for industrial use. ASTM D 2634-98. Standard specification for methyl amyl acetate (95% grade).
15.1 Standard methods of solvent analysis
1077
142 ASTM D 3130-95. Standard specification for n-propyl acetate (96% grade). 143 ASTM D 3131-97. Standard specification for isopropyl acetate (99% grade). BS1834-68. Specification for isopropyl acetate. 144 ASTM D 3540-98. Standard specification for primary amyl acetate, synthetic (98% grade). 145 ASTM D 3728-97. Standard specification for 2-ethoxyethyl acetate (99% grade). 146 ASTM D 5164-96. Standard specification for propylene glycol and dipropylene glycol. JIS K 1530-78. Propylene glycol. 147 ASTM D 4701-95. Standard specification for technical grade methylene chloride. ISO 1869-77. Methylene chloride for industrial use - List of methods of test. JIS K 1516-84. Methyl chloride. 148 ASTM D 4079-95. Standard specification for vapor-degreasing grade methylene chloride. 149 ASTM D 4080-96. Standard specification for trichloroethylene, technical and vapor-degreasing grade. ISO 2212-72. Trichloroethylene for industrial use - Methods of test. JIS K 1508-82. Trichloroethylene. 150 ASTM D 4081-95. Standard specification for drycleaning-grade perchloroethylene. ISO 2213-72. Perchloroethylene for industrial use - Methods of test. JIS K 1521-82. Perchloroethylene. 151 ASTM D 4126-95. Standard specification for vapor-degreasing grade and general solvent grade 1,1,1-trichloroethane. AS 2871-1988. Solvents. 1,1,1-trichloroethane (inhibited). ISO 2755-73. 1,1,1-trichloroethane for industrial use - List of methods of test. JIS K 1600-81. 1,1,1-trichloroethane. 152 ASTM D 4376-94. Standard specification for vapor-degreasing grade perchloroethylene. 153 ASTM D 4206-96. Standard test method for sustained burning of liquid mixtures using small scale open-cup apparatus. ISO/TR 9038-91. Paints and varnishes - Determination of the ability of liquid paints to sustain combustion. 154 ASTM E 1194-93. Standard test method for vapor pressure. DIN EN 12-93. Petroleum products. Determination of Reid vapor pressure. Wet method. ISO 3007-99. Petroleum products and crude petroleum - Determination of vapor pressure - Reid method. 155 ASTM D 2879-96. Standard test method for vapor pressure-temperature relationship and initial decomposition temperature of liquids by isoteniscope. 156 ASTM D 4486-96. Standard test method for kinematic viscosity of volatile and reactive liquids. DIN EN ISO 3104. Petroleum products. Transparent and opaque liquids. Determination of kinematic viscosity and calculation of dynamic viscosity. ISO 3104-94. Petroleum products - Transparent and opaque liquids - Determination of kinematic viscosity and calculation of dynamic viscosity. SFS-EN ISO 3104. Petroleum products. Transparent and opaque liquids. Determination of kinematic viscosity and calculation of dynamic viscosity. 157 ASTM D 88-94. Standard test method for Saybolt viscosity. 158 ASTM D 1545-93. Standard test method for viscosity of transparent liquids by bubble time method. 159 ASTM D 5201-97. Standard practice for calculating formulation physical constants of paints and coatings. ISO/FDIS 11890-1-99. Paints and varnishes - Determination of volatile organic compound content (VOC) - Part 1: Difference method. ISO/FDIS 11890-2-99. Paints and varnishes - Determination of volatile organic compound (VOC) content - Part 2: Gas-chromatographic method. 160 ASTM D 5325-97. Standard test method for determination of weight percent volatile content of water-borne aerosol paints. 161 ASTM D 5087-94. Standard test method for determining amount of volatile organic compound (VOC) released from solventborne automotive coatings and available for removal in a VOC control device (abatement).
1078
Myrto Petreas
15.2 SPECIAL METHODS OF SOLVENT ANALYSIS 15.2.1 USE OF BREATH MONITORING TO ASSESS EXPOSURES TO VOLATILE ORGANIC SOLVENTS
Myrto Petreas Hazardous Materials Laboratory, Department of Toxic Substances Control, California Environmental Protection Agency, Berkeley, CA, USA
15.2.1.1 Principles of breath monitoring Exposure to organic solvents may occur as a result of occupation, diet, lifestyles, hobbies, etc., in a variety of environments (occupational, residential, ambient). Solvents reach the organism through inhalation, ingestion and dermal exposure. The magnitude of exposures, however, may be modulated by factors such as the use of protective equipment (decrease) or physical exertion (increase). All sources and routes of exposure are integrated in the resulting internal dose. The dose at the target organ may trigger mechanisms that eventually may result in irreversible biological changes and disease. Cellular repair mechanisms and individual susceptibility have a significant effect on the onset of disease, resulting in a broad distribution of outcomes. This continuum between exposure to organic solvents and disease can be depicted, in a simplified way, in Figure 15.2.1.1. Biological monitoring focuses on the elucidation of the first step of this process; i.e., the relationship between environmental exposure and internal dose. Whereas the dose at the target organ is the biologically important, concentrations of the solvent (or its metabolites) may be more easily measured in other fluid or tissue samples, such as blood, breath, urine, etc. Traditional workplace monitoring relied on measurements of airborne contaminant concentrations at the breathing zone, and comparisons to reference values or regulatory benchmarks, such as the Threshold Limit Values (TLVs) and the Permissible Exposure Limits (PELs) in the USA, and similar values promulgated by the European Union and other organizations. Air measurements, however, may not always represent actual personal exposures since workers may be exposed dermally and orally, or they may be using protective equipment. Over the last decade, work-place monitoring programs have expanded to include biological monitoring. Biological monitoring gives an estimate of the dose inside the body (or specific target organ) rather than the concentration of the solvent in the external environment. Internal dose is defined as the amount of the solvent taken up by the whole organism (or a body compartment) over a specified period of time.1 Depending on the chemical, the time of sampling and the tissue analyzed, the dose may reflect the amount of the chemical recently absorbed, the amount of the chemical stored (body burden), or the amount of the chemical bound to the active sites of a receptor. It is Figure 15.2.1.1. Relationship between exposure to volatile organic the dose at the receptor or sensitive tissue (the biologically effecsolvents and potential outcomes.
15.2.1 Use of breath monitoring
1079
tive dose) that is related to toxicity and/or damage. In general, the dose at the receptor can not be measured because, at this time, available methodology is unacceptably invasive to be employed. On the other hand, biological monitoring allows measurements to be made on body fluids or tissues that would serve as surrogate measurements of the dose at the receptor. Certain conditions, however, are required for meaningful interpretations of the analytical results. 1. There should be adequate information on the absorption, distribution, biotransformation, metabolism and elimination of the chemical in the body. This type of information will indicate what tissues to sample, for what compound (parent chemical or a metabolite) and at what time. The latter issue is very important with chemicals that are eliminated rapidly (i.e., those with short half-lives). 2. The concentration of the chemical (or its metabolite) to be measured in the selected medium (breath, blood, urine) must be in equilibrium with the concentration of the chemical at the target organ. 3. The assay has to be sensitive, in order to detect low levels of the compound before any adverse effects take place; specific, in order to link exposure to dose; and accurate and precise, in order to be reliable. Very few chemicals have been adequately studied to allow biological monitoring according to the above criteria. This is, however, an expanding area, as indicated by the number of substances for which Biological Exposure Indices (BEIs) have been proposed by the American Conference of Governmental Industrial Hygienists (ACGIH). In the 1984-85 Threshold Limit Value (TLV) booklet there were BEIs for 10 chemicals; in the 1998 booklet there were 37 compounds with adopted BEIs.2 BEIs are considered reference values intended as guidelines and do not determine whether the worker is at risk of disease. They are supplementary to TLVs in evaluating workers’ exposures to workplace hazardous agents. The recommended values of BEIs are based on data correlating exposure intensity and biological effects from field studies and/or on pharmacokinetic treatment of data obtained from controlled human exposures. The BEIs developed by the ACGIH, assume 8-hour exposures for 5 days, followed by 2 days of non-exposure, and they strictly specify the time of sample collection. A clear advantage of biological monitoring over environmental or personal monitoring is that the estimate of the internal dose takes into account all possible routes of entry (inhalation, ingestion and dermal exposure). In addition, exposures other than occupational, such as through diet, hobbies or residential air, all contribute to the internal dose, which is related to adverse effects. When the effect of occupational exposure alone needs to be evaluated, non-occupational exposures need to be assessed and controlled to allow interpretation of biological monitoring results. A special case is the evaluation of protective equipment in reducing workplace exposures. Biological monitoring is limited by parameters that can affect the exposure and dose relationship. Such parameters include the following: biological conditions (age, sex, obesity, pregnancy, disease), physical workload, or exposure to other agents (other chemicals, tobacco smoking, dietary components) that in some way interfere with the uptake, distribution, metabolism or elimination of the chemical in the body.3-6 Such parameters may introduce large inter-individual variation in the resulting dose. When simulation techniques were used to evaluate the influence of these parameters, it was shown6 that up to 100% bias may result from inter-individual variation in the metabolism of aromatic compounds. If these differences occur randomly, they will widen the confidence limits of the biological monitoring results, decreasing the statistical power of a study. This is just one example illustrating the care that must be taken when interpreting biological monitoring results.
1080
Myrto Petreas
In short, biological monitoring should be used with specific objectives, such as to evaluate exceedence of a reference or regulatory value; dermal or oral exposure not assessed by personal monitoring; efficacy of protective equipment; etc. Accordingly, current, recent or cumulative exposure may need to be assessed and, therefore, the appropriate chemical (parent or metabolite) should be monitored in the appropriate specimen, collected at the appropriate time, from the appropriate worker(s). 15.2.1.2 Types of samples used for biological monitoring Of the various tissues/fluids that can be used for biological monitoring, the most common are blood, urine and exhaled air because they are relatively easy to obtain. Blood collection is the most invasive of the three and the chemical analysis may be subject to interferences because of the complex matrix. Additionally, concern over hepatitis and HIV infection, and confidentiality issues, make blood sampling less attractive for routine collection. Nevertheless, concentrations of a compound of interest in blood are easier to relate to concentrations at the target organ. Analysis of urine samples allows elimination rates of the chemical or its metabolite(s) to be ascertained. However, measurements in urine require specimen collection under a schedule (first morning urine, or 24-hour sample) that can create logistical difficulties. The concentration of the contaminant, or its metabolite(s), is usually corrected for the dilution of the urine (specific gravity or, more commonly, creatinine correction) to be used as a measure of dose. Collection of breath samples is the easiest of the three methods. It is non-invasive; it can take place almost anywhere; and the analytical matrix is usually very simple. The basic assumption underlying breath monitoring is the existence of gaseous equilibrium between the concentration of the vapor of interest in the alveolar air and in the arterial blood. This relationship must be established for every solvent for which biological monitoring by exhaled air analysis is considered. The apparent simplicity of breath sampling should not be overstated, however, and the limitations and pitfalls should be identified before any large scale, routine applications of breath monitoring can be envisioned. Since breath is a non-homogeneous mixture of air coming from different regions of the lung with varying ventilation, perfusion, and diffusion characteristics, some portion of the breath should be consistently identified and tested for its relationship to the blood concentration. The ACGIH only recommends measurements in exhaled air when:7 • The chemical is poorly metabolized and, therefore, exhalation is the primary route of elimination; • Sampling can occur when the effect of time and other circumstantial factors can be controlled; • The sampling method is well defined. 15.2.1.3 Fundamentals of respiratory physiology In order for any inhaled solvent vapors to enter the blood circulation, they need to reach the alveoli, cross the gas-blood interface and dissolve in the blood. Gas is carried to the gas-blood interface by airways, while blood is carried by blood vessels. Since these functions take place in the lung, the essentials of respiratory physiology will be briefly reviewed. The primary function of the lungs is to allow oxygen to move from the inhaled air into the arterial blood and to allow carbon dioxide to move from the venous blood to the exhaled air. Both oxygen and carbon dioxide move between air and blood by simple diffusion from an area of high partial pressure to an area of low partial pressure. The barrier between air and blood is less than 0.5 µm in thickness and has an area of between 50 to 100 m2.8 This barrier is very efficient for gas exchange by molecular diffusion, according to Fick’s 2nd law.
15.2.1 Use of breath monitoring
1081
The airways consist of a series of tubes that become narrower and shorter as they extend deeper into the lung. The first segment, the trachea, divides into two main bronchi, which in turn divide into lobar and segmental bronchi, which turn into terminal bronchioles. All these segments collectively constitute the anatomical dead space with a volume of about 150 mL for an adult male. The function of this conducting portion of the lung is to transport the inspired air into the gas exchange regions. The terminal bronchioles divide within a distance of 5 mm into respiratory bronchioles and alveolar ducts, which are completely lined with alveoli. This region of the lung is the respiratory zone where the gas exchange takes place. The volume of this zone (about 2,500 mL) makes up most of the lung volume. 15.2.1.3.1 Ventilation Air is drawn into the lungs by contractions of the diaphragm and the intercostal muscles, which raise the rib cage, and flows to the terminal bronchioles by bulk flow. After that point, the velocity of the inspired air diminishes as the cross sectional area of the airways increases dramatically. Thereafter, ventilation is carried out by molecular diffusion, which results in rapid exchange of gases. Similarly to the airways, the pulmonary blood vessels form a series of branching tubes from the pulmonary artery to the capillaries and back to the pulmonary veins. The diameter of a capillary segment is about 10 µm, just large enough for a red blood cell. The capillaries form a dense network in the walls of the alveoli, with the individual capillary segments so short that the blood forms an almost continuous sheet around the alveoli, providing ideal conditions for gas exchange. In about one second, each red blood cell transverses two or three alveoli in the capillary network, achieving complete equilibration of oxygen and carbon dioxide between alveolar gas and capillary blood. For an adult male at rest, a typical breathing frequency is 15 breaths per minute and a typical exhaled air volume (tidal volume), at rest, is about 500 mL. The total volume of air exhaled in 1 min is, therefore, about 7,500 mL. This is the minute volume total ventilation. The inhaled volume during the same time is slightly greater, since more oxygen is taken up than carbon dioxide released. Not all the inhaled air reaches the alveoli; the anatomical dead space (volume of conducting airways where no gas exchange takes place) contains approximately 150 mL of each breath. Therefore, the volume of fresh inhaled air that reaches the alveoli in 1 min is 5,250 mL [(500-150) mL/breath * 15 breaths/min = 5,250 mL/min].8 This is the alveolar ventilation and it represents the volume of fresh air available for gas exchange in 1 min. The alveolar ventilation cannot be measured directly, but it can be calculated by measuring the minute volume and subtracting the anatomical dead space. The former is easily measured by collecting all the expired air in a bag. The volume of the anatomical dead space can be assumed to be 150 mL, with minimal variation for all adults, and the dead space ventilation can be calculated given a respiratory frequency. Subtraction of the dead space ventilation from the total ventilation (or minute volume) results in the alveolar ventilation. Pulmonary ventilation may be affected by chronic diseases such as asthma or pulmonary fibrosis, or even by transient hyperventilation or hypoventilation. 15.2.1.3.2 Partition coefficients Whereas the rate of transfer of a solvent vapor between alveolar air and capillary blood is determined by its diffusivity, the equilibrium between these matrices is determined by the blood/air partition coefficient (λ). This is the ratio of the concentration of the vapor in blood and air at 37oC, at equilibrium. Partition coefficients are commonly determined in vitro, and occasionally in vivo.2,9,10 Although considered constant at a particular temperature, partition coefficients may be affected by the composition of the blood. The blood/air partition coeffi-
1082
Myrto Petreas
cients of some industrial solvents were shown to increase by up to 60% after meals,4 presumably due to an increase in lipid content of the blood. Depending on the affinity of a gas or vapor for blood, the transfer from the alveoli to the capillaries may be diffusion-limited or perfusion-limited. The difference can be illustrated by examining two solvents, styrene and methyl chloroform. Because of the high solubility of styrene (λ=52)11 large amounts of it can be taken up by the blood, and the transfer is only diffusion-limited. On the other hand, methyl chloroform is not very soluble (λ=1.4)12 and its partial pressure rises rapidly to that of the alveolar air, at which point no net transfer takes place. The amount of methyl chloroform taken up by the blood will depend exclusively on the amount of available blood flow and not on the diffusion properties of the gas-blood interface. This kind of transfer is perfusion-limited. It is the concentration in the mixed venous blood, however, that better reflects the concentration at the target organ. It has been shown that:13 P / Pv =
λ λ +V / Q
where: P Pv λ V Q
partial pressure in the alveoli and in the arterial blood leaving that region of the lung partial pressure in the venous blood coming to the lung blood/air partition coefficient alveolar ventilation blood flow of the lung region under consideration
The greater the blood/air partition coefficient, the closer the arterial concentration will be to the venous concentration. The concentration in the alveolar air will reflect both arterial and venous blood if the blood/air partition coefficient of the vapor is greater than 5,14 a criterion that applies to most industrial solvents. In a normal subject at rest, the ventilation-to-perfusion ratio (V/Q) ranges from 0.7 to 1.0, with an average of about 0.9.15 Variations in the V/Q ratio exist in various parts of the lung because not all alveoli are ventilated and perfused in ideal proportions. The V/Q ratio may vary from 0.5 at the lung base to 3.0 at the apex,16 and it becomes more homogeneous with physical exertion.17 When the partition coefficient exceeds 10, however, even large degrees of ventilation/perfusion imbalance have very little effect on the relationship between alveolar and mixed venous solvent partial pressures.14 15.2.1.3.3 Gas exchange When a person is exposed to a volatile organic solvent through inhalation, the solvent vapor diffuses very rapidly through the alveolar membranes, the connective tissues and the capillary endothelium and into the red blood cells or plasma. With respiratory gases the whole process takes less than 0.3 seconds.15 This results in almost instantaneous equilibration between the concentration in alveolar air and in blood and, therefore, the ratio of the solvent concentration in pulmonary blood to that in alveolar air should be approximately equal to the partition coefficient. As the exposure continues, the solvent concentration in the arterial blood exceeds that in the mixed venous blood.18 The partial pressures in alveolar air, arterial blood, venous blood and body tissues reach equilibrium at steady state. When the exposure stops, any unmetabolized solvent vapors are removed from the systemic circulation through pulmonary clearance. During that period the concentration in the arterial blood is lower than in the mixed venous blood18 and the solvent concentration in alveolar air will depend on the pulmonary ventilation, the blood flow, the solubility in blood and the concentration in the
15.2.1 Use of breath monitoring
1083
mixed venous blood. These are the prevailing conditions when breath monitoring takes place after the end of the exposure17 and therefore, samples of alveolar air after exposure should reflect concentrations in mixed venous blood. During exposure, however, alveolar air should reflect concentrations in arterial blood. Solvent concentrations in arterial and venous blood diverge even more with physical exertion. To summarize, the uptake of solvent vapors through inhalation will depend on the following factors:3 1. Pulmonary ventilation, i.e., the rate at which fresh air (and solvent vapor) enters the lungs. This is determined by the metabolic rate and therefore depends on physical exertion. The concentration in alveolar air approaches that in inhaled air when the physical exertion is great. When exertion is low, the alveolar air concentration approaches the concentration in mixed venous blood. 2. Diffusion of the solvent vapor through the gas-blood interface. According to Fick’s 2nd law, the rate of diffusion through the interface depends on the concentration gradient across the tissue membrane. Therefore, the concentration of the vapor leaving the lungs depends on the amount of vapor entering the alveoli, on the solubility in blood (blood/gas partition coefficient) and on the flow rate of the blood through the lungs. For lipophilic chemicals, such as organic solvents that readily cross cellular membranes, diffusion is not the rate-limiting factor.18 3. The solubility of the vapor in the blood. The higher the blood/gas partition coefficient, the more rapidly the vapor will diffuse into the blood, until equilibrium is achieved. At equilibrium the net diffusion between blood and air ceases, but the concentrations in air and blood may still be different. A highly soluble vapor will therefore demonstrate a lower alveolar air concentration relative to the inhaled air concentration during exposure, and a higher alveolar concentration relative to the inhaled air concentration after exposure. 4. The circulation of the blood through the lungs and tissues. This depends on the cardiac output and therefore, on physical exertion. For a soluble vapor, high exertion moves blood faster, increasing uptake. 5. The diffusion of the vapor through the tissue membranes, which depends on membrane permeability and is governed by Fick’s law, as discussed in #2 above. 6. The solubility of the solvent vapor in the tissue. This reflects the tissue/blood partition coefficient of the solvent vapor and depends largely on the lipid composition of the particular tissue. Lipophilic vapors will exhibit high solubility in tissues with high fat content, freeing blood of vapors and increasing uptake. Substances that are not well metabolized are eliminated primarily through the lungs3 and to a lesser extent through the kidneys. During elimination through the lungs the same physiological principles apply in a reverse sequence. 15.2.1.4 Types of exhaled air samples Samples obtained from even a single breath may contain different concentrations of the solvent vapor of interest depending on the way the sample was collected. This is a result of the non-homogeneity of the exhaled air discussed previously. Figure 15.2.1.2 depicts a concentration curve of a solvent in exhaled air. When measurement takes place in an environment free of that solvent, the first part of the exhalation, corresponding to air in the anatomical dead space, will be solvent-free (Phase 1). Then the solvent concentration curve rises sharply reflecting exhalation of air from the respiratory zone of the lungs (Phase 2). This phase is followed by a slowly increasing plateau, indicating an alveolar steady state (Phase 3). The level of this plateau depends on the preceding ventilatory state of the subject, being higher for a hypoventilating subject and lower for a hyperventilating subject.6 Sample col-
1084
Myrto Petreas
lection at the alveolar plateau constitutes end-exhaled air. When the total expiration is collected (combination of all three phases) the sample is considered mixed-exhaled air. Simultaneous monitoring of CO2,6,19-21 or oxygen and nitrogen22 has been suggested to normalize the measured solvent vapor concentrations in breath samples in order to facilitate data interpretation and comparisons. The various sampling techniques must be evaluated and the one most applicable to the purpose of sampling chosen. The most commonly used sampling techniques based on physiological principles include mixed-exhaled air and end-exhaled air: Figure 15.2.2. Solvent concentration in breath during exhalation. Mixed-exhaled air. This technique involves the collection of the entire volume of exhaled air. It corresponds to a mixture of the alveolar air with air from the dead space. The collection apparatus may also contribute to the dead space. Total dead space should be considered and the concentration adjusted, either by subtraction, or by regression against some other technique unaffected by the dead space. Timing of the breath collection is important here since the concentration of the air in the dead space may equal that of the air in the workroom if the sample is taken during exposure, or it may equal zero if taken after the end of exposure. End-exhaled air. This technique excludes air from the dead space and collects only the last part of the breath, in order to estimate the concentration in the alveolar air which is in equilibrium with the arterial blood. Use of a Haldane-Priestley tube,23 or simultaneous monitoring of the peaking of the temperature of the breath19 or its CO2 concentration,6,19,24 will assure a valid alveolar air sample.6,25,26 Two less frequently used techniques include breath holding or rebreathing to homogenize the breath sample: Breath holding. For compounds that achieve equilibrium slowly, some researchers have used the breath holding technique, which involves holding the breath for 5-30 seconds prior to exhaling into the collection device. This results in a more constant concentration in the exhaled breath. The extent to which the lungs are filled with air (as with a deep inhalation) will affect the results. Kelman14 showed, however, that breath holding is not necessary for solvents with blood/gas partition coefficients greater than 10. Rebreathing. With this technique the subject rebreathes his/her exhaled air to provide a more homogenized breath sample. This technique is applicable to solvents that reach steady state very slowly. For prolonged rebreathing, removal of CO2 to avoid pH changes and/or supply of oxygen may be necessary.27 15.2.1.5 Breath sampling methodology The theoretical principles discussed above lead to two choices for collecting breath samples: Should a mixed-exhaled air or an end-exhaled air sample be collected?
15.2.1 Use of breath monitoring
1085
End-exhaled air is becoming the standard for breath monitoring. Of the 36 chemicals with established BEIs,7 end-exhaled air is recommended for 6, while earlier BEIs recommending mixed-exhaled air have been withdrawn. Beyond the regulatory arena, however, mixed-exhaled air continues to be used with satisfactory results. Methodologically, the difference between these two types of samples is the control over which portion of the breath is collected. Many techniques28-32 make use of the Haldane-Prestley tube to collect the last portion of the expiration. Earlier work6,19 used the simultaneous monitoring of CO2 or temperature in the exhaled breath to identify the moment (CO2 or temperature reaching a maximum) when air from the anatomical dead space has been purged and alveolar air can be sampled. A variety of sampling techniques has been tried for measuring organic vapors in exhaled air. Table 15.2.1.1 shows a classification of these techniques. The main distinction is between direct reading instruments and transportable samples that require analysis in the laboratory. The former group covers the widest range, from simple detector tubes to High Resolution Mass Spectrometry. Table 15.2.1.1. Techniques for sampling organic vapors in exhaled air Technique A. SAMPLE COLLECTION A.1 END-EXHALED AIR Use of Haldane-Prestley tube Simultaneous CO2 monitoring Simultaneous temperature monitoring B. SAMPLE ANALYSIS B.1 DIRECT READING INSTRUMENTS B.1.1 DETECTOR TUBES Colorimetric B.1.2 GAS CHROMATOGRAPHS with various detectors and modified injector/inlet with mouthpiece GC/FID GC/PID GC/MS B.2 TRANSPORTABLE SAMPLE COLLECTORS B.2.1 CONTAINERS B.2.1.1 Rigid containers Stainless steel tubes Glass tubes Aluminum tubes B.2.1.2 Flexible containers Tedlar bags Saran bags Mylar bags B.2.2 ADSORBENTS B.2.2.1 Tubes/cartridges containing: Charcoal Tenax Silica gel B.2.2.2 Felt/cloth
References 23, 25, 26, 28, 29, 31 6, 20, 21, 24 19
33 52 25, 26 24, 41, 42, 43
24 9, 53, 54, 55, 61, 64 28, 29, 47 6, 49 54, 55, 60 25, 26, 28, 29, 40 31, 49 6 48
Specifically, detector tubes operating on colorimetric reaction principles have been used to determine the concentration of alcohol in breath for traffic safety.33 Alcohol measurements in breath have been thoroughly studied and scrutinized because of the forensic
1086
Myrto Petreas
and legal repercussions associated with the results. Portable devices have been developed for field use with on-site analysis of alcohol.34-36 These portable devices could be adapted for use with other organic solvents of interest in industrial hygiene and occupational health. Gas chromatographs (GCs) equipped with suitable detectors allow analysis of more than one solvent vapor at a time. The subject exhales through a valve system and the exhaled volume of air, or a fraction of it, is drawn into the GC.21,26,37-39 Depending on the detector used, one or more cylinders containing appropriate gases under pressure are required to operate the instrument, limiting its field use. An Electron Capture Detector (ECD), which is the most sensitive for polyhalogenated solvents, requires helium and/or nitrogen. A Flame Ionization Detector (FID), which is a universal detector for aliphatic and aromatic hydrocarbons, requires nitrogen, air and hydrogen. The advent of the Photoionization Detector (PID) offered a compromise, in that it can detect all the above classes of organic solvents (with varying sensitivity) requiring only one gas (nitrogen, air or helium) for its operation. A simple system employed a portable GC/PID modified to sample 1 mL of air from a Haldane-Priestley tube into which the subject would exhale.25,40 The latter assured collection of end-exhaled air and this device was successfully used to assess exposures to styrene26 and PERC.40 Mass spectrometry (MS) offers the advantages of high sensitivity and specificity at the cost of portability and the need of highly trained operators. Direct exhalation into the MS is not possible because the ion source of the detector operates under high vacuum. Two approaches have been taken to overcome this limitation. In the first, an inlet system was designed where the high pressure breath sample was directed over a methyl silicone membrane, permeable to organic molecules but not to air, that served as the entry port to the ion source operating under high vacuum.41 With the second approach, the inlet system was attached to the source of an Atmospheric Pressure Ionization mass spectrometer (API/MS), that operates under atmospheric pressure.42,43 In summary, the basic advantage of the direct reading/analyzing instruments is that there is no sample collection, transportation, storage, handling and analysis where the sample integrity may be compromised. The disadvantages are the need for expensive, specialized equipment, usually requiring highly trained operators and the difficulty, or some times impossibility, of field use. The other category of breath sampling methodology involves sample collection in the field with subsequent analysis in the laboratory. There are two groups of devices in this category: 1. Sample containers that hold the total volume or part of the exhalation in a gaseous form. 2. Sample adsorbents that trap the chemicals of interest from the exhaled breath. In the first group belong rigid containers such as stainless steel, aluminum, or glass vials of various sizes, open at both ends. The subject exhales into the vial and then the two openings are closed, either with caps or with stopcocks. Vials of this kind have been used extensively in experimental studies involving exposures to industrial chemicals.24,28,29,44-47,53-55,61,64 The advantages of these containers are inertness of the walls, low cost, reusability and simplicity of the technique. The disadvantages are the potential for contamination, photodegradation, or dilution of the collected breath sample with atmospheric air, sorption and/or reaction of the vapor with surfaces and breakage during transportation or storage. The problem of condensation of water vapor has been addressed by heating the tubes,28,29,47 but this approach increases the complexity and counters the main advantages. Another limitation is that only a small volume of sample can be withdrawn for analysis from a rigid container, with ensuing restrictions on the analytical sensitivity.
15.2.1 Use of breath monitoring
1087
The last limitation is overcome when collapsible bags are used for sample collection. The bags are commonly made of Tedlar (clear) or Mylar (aluminized) material. The subject exhales directly into the bag through a valve and when an appropriate volume has been collected, the valve is closed. The bag is transported to the laboratory where an aliquot is withdrawn for analysis through a septum. The collapsibility of the bag allows any volume to be withdrawn without air infiltration. Disadvantages, however, include losses because of permeation through the walls, sorption and/or reaction of the vapor with the wall surfaces and photodegradation. The latter concern is eliminated with the aluminized Mylar bags, which also minimize sorption and permeation through the walls. The second group of field sampling devices with subsequent analysis involves the use of adsorbents. Extensively used in personal monitoring in industrial hygiene surveys, these devices were naturally considered for exhaled breath sampling. Commercially available glass tubes containing granular adsorbent material such as activated charcoal, Tenax, molecular sieve, etc. could be used with selected organic vapors. However, because of the high pressure drop during exhalation through the adsorbent bed, and the condensation of water vapor, these materials have not been used much for direct sampling of exhaled breath. One device48 is, basically, a respirator mask with two commercial charcoal inhalation canisters and an exhalation port fitted with a wafer of charcoal cloth. A Wright respirometer connected to the exhalation port records the volume of exhaled air. Another hand-held device employed a custom-made glass tube containing charcoal to collect organic vapors in exhaled air, while the volume collected was measured with a Wright respirometer and adjusted for back-pressure measured with an attached pressure gauge.25,40 Commercial tubes or specially made cartridges containing the appropriate adsorbent have been used in conjunction with bag sampling for exhaled breath monitoring.49 The subject exhales into the bag, the contents of which are subsequently pumped through the tube or canister containing the adsorbent. This combination offers the advantages of bag sampling (low cost, low resistance), while eliminating the disadvantages (losses, bulkiness). The organic vapors can be thermally desorbed from the adsorbent material and the concentrations measured with polymer-coated surface-acoustic-wave detector,50 GC or GC/MS analysis.31,49 Alternatively, solvent vapors can be extracted from the adsorbent with an appropriate solvent prior to conventional analysis. 15.2.1.6 When is breath monitoring appropriate? Breath monitoring may be used to assess exposures to solvents either qualitatively or quantitatively. Qualitative information may be useful when exposure to a number of solvents is suspected. For qualitative purposes, either mixed-exhaled or end-exhaled air samples may be used. Provided that the sample collection and the analytical techniques are appropriate, a list of solvents may be thus identified and quantitative measurements planned. Quantitative information is needed for regulatory enforcement situations, where appropriately collected breath samples are compared to BEIs to assess whether exposure limits have been exceeded. BEIs have been developed after thorough studies of pharmacokinetics for a limited number of occupational contaminants. Exhaled air is used as a surrogate for blood when the relationships between exhaled air, alveolar air and (arterial or mixed venous) blood have been established. Besides regulatory purposes, breath monitoring can provide valid information on solvent exposures when: • The pharmacokinetics (absorption, distribution, metabolism, elimination) of the solvent are well established. • The solvent is not soluble in the conducting portion of the lung (anatomical dead space).
1088
Myrto Petreas
• The solvent is not metabolized to a great extent, and it is mainly eliminated via exhalation.
• Sample collection is conducted in a location unaffected by ambient levels of the solvent(s) of concern that may interfere and bias the measurements.
• The concentration in exhaled air is compatible with the analytical method allowing adequate sensitivity.
• Appropriate timing is strictly followed. 15.2.1.7 Examples of breath monitoring Two solvents are used to illustrate the use of breath monitoring in assessing solvent exposure: A solvent with minimal metabolism, tetrachloroethylene (PERC) and a solvent with significant metabolism, styrene. Both solvents are used extensively in industry with a great number of workers potentially exposed. As shown in Table 15.2.1.2, styrene is only minimally excreted through the lungs, whereas exhaled air is the major elimination route for PERC. Accordingly, end-exhaled air is recommended as a BEI for PERC, but not for styrene. Nevertheless, end-exhaled and mixed-exhaled air techniques have been studied in experimental settings and in real-life occupational environments for both solvents. A compilation of such studies is presented below. Table 15.2.1.2. Characteristics of styrene and tetrachloroethylene (PERC). [Information compiled from ACGIH, Documentation of Threshold Limit Values and Biological Limit Values, 19917] STYRENE Eliminated via lungs Eliminated in the urine Urinary metabolite(s) Blood/air partition coefficient (λ) Pulmonary elimination half-lives (T1/2)
~5% ~90% Mandelic acid (MA), Phenylglyoxylic acid (PA) 32-55
PERC ~80-100% ~2% Trichloroacetic acid (TCA) 9-15
13-52 min 4-20 hr 3 days
15 min 4 hrs 4 days
TLV (1998) STEL (1998) BEI (1998) End-exhaled air
50 ppm 100 ppm
25 ppm 100 ppm
Not recommended
Blood
0.55 mg/L end-of-shift 0.02 mg/L prior-to-next-shift
10 ppm end-of-shift, after at least 2 shifts 1 mg/L end-of-shift, after at least 2 shifts
Urine
Major Industry
TCA=7 mg/L MA=800 mg/g creatinine, end-of-shift MA=300 mg/g creatinine, prior-to-next-shift PA=240 mg/g creatinine, end-of-shift PA=100 mg/g creatinine, prior-to-next-shift Reinforced Plastics
Dry Cleaning
15.2.1 Use of breath monitoring
1089
15.2.1.7.1 PERC Tetrachloroethylene (PERC) is used extensively in the textile industry as a dry cleaning aid and in metal processing as a metal degreaser. The major concern is on occupational exposures, but elevated indoor air concentrations of PERC have also been documented.49 As a result, the potentially exposed population is very large, with a wide range of exposures. The uptake of PERC by the lung is high51,52 because of its high solubility in blood and adipose. Most of it is eliminated via the lung, with very small amounts metabolized to trichloroacetic acid (TCA) and eliminated in the urine.51,52 Even smaller amounts of trichloroethanol in the urine have been reported.52 A number of field studies have been published with similar designs, in which breathing zone air, blood, breath and urine were measured in groups of workers. In addition, chamber studies allowed measurements under predetermined consistent conditions. The emerging patterns formed the basis for recommendations for biologic monitoring. In a study of thirty two workers,9 end-exhaled air was measured at the beginning of the first shift of the week, after 15-30 min after the end of the shift on the 3rd and 5th day of the same week, and at the beginning of the first shift of the subsequent week. The authors required the subjects to hold their breaths for 5 seconds and collected only the last part of the breath to obtain alveolar air. These measurements correlated with personal breathing zone samples collected each day. The highest correlation was found between PERC and TCA in blood at the end of the workweek and the average exposure for the entire week (R2=0.953). Among the non-invasive techniques, PERC in exhaled air collected 15-30 min after the end of the shift at the end of the week had an R2 of 0.931, and was followed by TCA in urine (R2=0.909). In a field study of workers employed in dry cleaning,53 twenty-four female and two male subjects were monitored. End-exhaled air and blood samples were collected at the beginning and at 30 minutes after the end of the 3rd workday of the week. On that same day, personal, shift-long, breathing zone samples were also collected from each worker. Urine samples were collected before and after the shift on a daily basis and analyzed for TCA. With a mean exposure of 20.8 ppm, no TCA was detected in any of the urine samples (no detection limit was reported). Among the three exposure parameters (breathing zone air, blood and end-exhaled air) only the correlation between the PERC concentration in the blood and in the breath was reported as statistically significant by the authors (R2=0.77). The authors concluded that if the concentration of PERC in blood, 16 hours after the end of exposure, was below 1000 µg/L, the average exposure was likely to have been below 50 ppm. A group of thirteen male dry cleaners40 was followed for a week with daily shift-long personal (badge) monitoring; mixed-exhaled air collected daily before- and after-shift; and mixed-exhaled air, blood and urine measurements on the morning after the last shift. Results indicated strong correlations between PERC concentrations in mixed-exhaled air and blood (R2=0.944), as well as mixed-exhaled air and personal air samples collected either after the shift (R2=0.885), or on the following morning (R2=0.770). Mixed-exhaled air measurements collected either at the end of the last shift of the week (R2=0.814) or in the morning following the last shift (R2=0.764) were better indicators of exposure over the entire week than either blood (R2=0.697) or urine (R2=0.678). The authors recommended consideration of mixed-exhaled air as a potential BEI. In addition to field studies, controlled human exposures to PERC have been reported. In this type of studies, subjects were exposed to various constant levels of PERC for different lengths of time with or without exercise. Samples of blood, exhaled air and urine were collected according to a variety of schedules both during exposure and for varying lengths
1090
Myrto Petreas
of time after exposure. One study involved single exposures, whereas in others at least some of the subjects participated more than once. In one of the first chamber studies, nine male volunteers were exposed to 100-200 ppm of PERC for 80 to 180 minutes. Concentrations of PERC in exhaled air were measured over time.54 The authors demonstrated that exposures of similar duration resulted in similar decay curves of PERC concentrations in breath after exposure. In addition, the data showed that the length of time that PERC was measurable in exhaled air was proportional to both the concentration and the duration of the exposure. The same group55 exposed sixteen males to single and repeated concentrations of 100 ppm PERC for periods of 7 hrs/day over a 5-day workweek. Alveolar air samples were collected over a period of up to 300 hours post-exposure. The authors observed higher alveolar air concentration in subjects with greater body mass. They recommended measurement of PERC in alveolar air 2-16 hours post-exposure to assess the average exposure. Six healthy males were exposed to 72 and 144 ppm, as well as 142 ppm with 2 half hours of 100 W of exercise.51 Uptake of PERC was calculated from the minute volume and the concentration of PERC in breath. PERC was measured in breath and blood and TCA was measured in blood and urine. During the first hour of exposure the uptake was higher by 25% than during the last hour and retention decreased with time and with exercise. The mean ratio of the concentration of PERC in venous blood and the concentration of PERC in mixed exhaled air was 23. Assuming a mix-exhaled air to end-exhaled air ratio of 0.71, they estimated a blood/air partition coefficient of 16. After 20 hours from the end of exposure the half-life was estimated to be 12-16 hours, after 50 hours the half-life was estimated to be 30-40 hours and after 100 hours it was estimated at approximately 55 hours. The authors estimated that 80-100% of PERC was excreted unchanged through the lung, and about 2% was excreted as TCA in the urine. Simple and multiple linear regression analyses were applied to the data collected above to predict the best estimator of exposure to PERC.56,57 PERC in blood was the best predictor followed by PERC in exhaled air. TCA per gram of creatinine in urine was the least reliable predictor. Twenty four subjects were exposed to 100, 150 and 200 ppm of PERC for 1 to 8 hours.52 The concentration of PERC in alveolar air decreased rapidly during the first hours of the post-exposure period, but 2 weeks were required for complete elimination following an exposure of 100 ppm for 8 hours. The authors were the first to demonstrate that the rate of rise and of decay of the alveolar concentration with respect to time was independent of the level of the inspired concentration during exposure. The experimental data from the above study were used in a pharmacokinetic model58 that divided the body into four tissue compartments: the Vessel Rich Group (VRG), the Muscle Group (MG), the Fat Group (FG) and the Vessel Poor Group (VPG). No metabolic pathway was considered. After 20 hours from the end of the exposure the concentration of PERC in alveolar air was related to release from the FG with an approximate half-life of 71.5 hours. The model predicted small differences in alveolar concentrations due to differences in body weight, height and fat tissue. The authors provided a nomogram to predict mean exposure given the exposure duration (0-8 hrs), time since end of exposure (0-8 hrs) and post-exposure measurement of the alveolar air concentration. A perfusion-limited, four compartment model was used to simulate the effect of several confounding factors on the levels of several contaminants in the breath, including PERC.6 The model, which considered metabolic clearance, consisted of a pulmonary compartment, a vessel-rich group, a group of low-perfused tissues containing muscles and skin, and a group of poorly perfused fatty tissues. The authors predicted that intra-day fluctuations in exposure would have a large effect on breath sampled just after the shift, but that the
15.2.1 Use of breath monitoring
1091
effect would be negligible the next morning. Breath samples collected the next morning would reflect exposure on preceding days and the measurements would be independent of level of physical exertion. Finally, individual differences in body build and metabolism would have the greatest effect on breath levels, with the largest variations expected for metabolized solvents. Experimental data were used in a model,59 which divided the body into four tissue compartments: vessel rich group, muscle group, fat group and liver. Metabolism was assumed to take place in the liver as a combination of a linear metabolic component and a Michaelis-Menten component. Metabolic parameters and partition coefficients determined for rats were scaled for body weight and were used in fitting results for humans. The model fit very well the data reported from other controlled human exposure studies.51,52,54 The BEI committee of the ACGIH recommends2 three indices of PERC exposure: 1) PERC in end-exhaled air collected prior to the shift after at least 2 consecutive workdays. 2) PERC in blood collected prior to the shift after at least 2 consecutive workdays. 3) Trichloroacetic acid (TCA) in urine voided at the end of the workweek. This index is recommended only as a screening test because of the variability associated with urinary excretion of TCA and, as such, no creatinine correction is necessary. 15.2.1.7.2 Styrene The highest exposures to styrene take place in plants manufacturing glass-reinforced plastics, particularly during lamination, where there is considerable evaporation of styrene. Most of the inhaled styrene is absorbed and retained in the body. The first step in the major metabolic pathway is the formation of styrene-7,8-oxide (phenyloxirane), a reaction catalyzed by Cytochrome P-450 in the liver. The oxide is then hydrated to styrene glycol (1-phenyl-1,2-ethanediol) by microsomal epoxide hydratase. The styrene glycol is conjugated with beta-glucuronic acid, or is oxidized to mandelic acid and further to phenylglyoxylic acid. In humans, less than 5% of absorbed styrene is eliminated via the lung, whereas 90% is eliminated in the urine as mandelic and phenylglyoxylic acids. A number of chamber and field studies have been reported, where blood, exhaled air and urine have been studied as possible biological indicators. Although conditions and study design differed among studies, certain consistent patterns emerged, which formed the basis for subsequent recommendations. In an early series of chamber exposures,60 nine healthy males were exposed to styrene concentrations of 50 to 370 ppm for periods ranging from 1 to 7 hours. Alveolar air samples were collected during and for 8 hours following exposure. The authors showed that the amounts of styrene present in the breath were directly related to the level and the duration of exposure. In another study,61 fourteen healthy males were exposed to 50 and 150 ppm of styrene in inspired air during rest and light (50 W) physical exercise. The duration of exposure for each styrene concentration was 30 minutes, with total pulmonary ventilation, cardiac output, and styrene concentration in alveolar air, arterial blood and venous blood measured during and after exposure. During exposure at rest to either 50 or 150 ppm, the alveolar air concentrations were about 20% of the inspired air concentrations. When alveolar ventilation almost tripled with exercise at 50 W, the alveolar air concentrations increased slightly, whereas arterial blood concentrations almost tripled. Alveolar air concentrations reached a plateau within a minute or so, that persisted throughout the 30-minute period. Arterial blood concentrations, however, rose rapidly during each such period and even more with increasing exercise intensity (50, 100 and 150 W). Venous blood concentrations also rose sharply during the 30-minute exposure periods without ever reaching a plateau. No linear relationship was found between exposure duration and arterial and venous blood concentrations,
1092
Myrto Petreas
even after log-log transformations. The decay of styrene concentrations in alveolar air and in venous and arterial blood was slower in the first 30 to 50 minutes than later on. The authors concluded that post-exposure alveolar air measurements were poor predictors of exposures to styrene, whereas arterial blood would be the best index of exposure. For practical reasons, they recommended capillary blood samples from fingertips as a surrogate for arterial blood. In a follow-up study62 using the same experimental set up as above, seven male subjects were exposed to 50 ppm (210 mg/m3) of styrene in inspired air for 30 minutes at rest, followed by 30 minute exposure periods under exertion at intensities of 50, 100 and 150 W. The mean alveolar air concentration at the end of the first 30-minute period was 16% of the concentration in the inspired air and 23% at the end of the final period. Both the arterial and venous blood concentrations of styrene rose with increasing workload during exposure with no equilibrium achieved between alveolar air and arterial blood during 2 hours of total exposure. Weak correlations were found for the amount of styrene taken up per kg of body weight and the concentrations of styrene in alveolar air 0.5 and 2 hours after exposure. The elimination was faster during periods of exercise than during rest periods, leading to the authors’ recommendation for leisure time physical activity for people exposed to solvents as a means to enhance elimination. In another study,63 four male volunteers were exposed to 80 ppm of styrene in a chamber for 6 hours. Venous blood samples were collected during exposure and a nearly simultaneous set of 10 blood and mixed exhaled air samples were collected following exposure for up to 40 hours. Blood levels during exposure rose rapidly and reached an almost constant level by the end of the 6 hours. Following exposure, the results showed that styrene was cleared from the blood according to a linear two-compartment pharmacokinetic model, with half-life values of 0.58 and 13 hours, for the rapid and slow elimination phases, respectively. Retention of styrene was studied in a chamber setting where healthy males were exposed to either constant or fluctuating air concentrations of styrene.26 A computer-controlled system was used to generate time-varying air concentrations over 4-5 hrs with a mean air concentration of 50 ppm. End-exhaled air measurements taken throughout the exposure period showed styrene retention of 93.5% during constant exposures, but higher retention (96-97%) during fluctuating exposures. The authors speculated that the difference in retention was related to non-steady-state behavior of styrene in the richly perfused tissues. In summary, pulmonary retention during exposure is about 60-70%62,64 with exhaled concentrations accounting for 25-35%.60 Higher retention (96-97%) was observed when volunteers were exposed to fluctuating, rather than constant, air concentrations.26 Following exposure, the desaturation curve shows two exponential decays with half-lives of 13 minutes60 to 52 minutes63 for the rapid elimination phase and 4 hours60 to 20 hours63 for the slower elimination phase. A third compartment can be defined to represent elimination from the adipose tissue with a half-life of 3 days.62 Measurements of styrene in venous blood at the end of the workshift and/or prior to the next shift have been recommended by the ACGIH.2 Monitoring of urinary metabolites of styrene has also been recommended by the ACGIH2 at the end of the week, even though they are not specific to styrene exposure. Styrene in exhaled air after the end of exposure was not recommended as a BEI because the levels would be too low to detect, leading to uncertainties. Terminology ACGIH Alveolar air
American Conference of Governmental Industrial Hygienists Gas in the alveoli
15.2.1 Use of breath monitoring
Alveolar ventilation Anatomical dead space BEI Blood/air partition coefficient Breathing frequency Conductive zone of lungs End-exhaled air Exhaled air volume GC-MS Minute Volume Ventilation Mixed-exhaled air Respiratory zone STEL
Tidal Volume TLV
1093
Inhaled air available for gas exchange in 1 min Volume (~150 mL) of conducting airways where no gas exchange occurs Biological Exposure Index. Reference value developed by ACGIH Ratio of the vapor concentration in blood and in air at 37oC 15 breaths per minute at rest, increases with physical exertion First segment of the airways, which includes the trachea, bronchi, terminal bronchioles Operationally defined as the last portion of an exhalation to simulate alveolar air Approx. 500 mL at rest, increases with workload Gas Chromatography Mass Spectrometry The volume of air exhaled in 1 min, approx. 7,500 mL Entire volume of exhaled air collected from alveolar space and anatomically dead space Consists of respiratory bronchioles and alveolar ducts where gas exchange takes place. Approx. 2,500 mL Short-Term Exposure Limit. The concentration to which most workers may be exposed continuously for a short period of time without adverse effects. Issued by ACGIH Volume of exhaled air; ~500 mL at rest Threshold Limit Value. Airborne concentration of a substance to which most workers may be exposed without adverse health effects. Established by ACGIH and intended as guideline.
References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
R Lauwerys, Industrial chemical exposure: Guidelines for biological monitoring. Biomedical Publications. Davis, California, 1983. ACGIH, American Conference of Governmental Industrial Hygienists: TLV, Threshold Limit Values and Biological Limit Values. Cincinnati, Ohio, 1998. I Astrand, Scand J Work Environ Health. 1, 199-218, 1975. V Fiserova-Bergerova, J Vlach, JC Cassady, Br J Ind Med., 37, 42-49, 1980. PJA Borm and B de Barbanson, J Occup Med. 30, 214-223, 1988. PO Droz and MP Guillemin, J Occup Med., 28, 593-602, 1986. ACGIH, American Conference of Governmental Industrial Hygienists, Documentation of Threshold Limit Values and Biological Limit Values. Cincinnati, Ohio, 1991. JB West, Respiratory Physiology - The essentials, The Williams & Wilkins Co., Baltimore, 1981. AC Monster, W Regouin-Peeters, A van Schijndel, J van der Tuin, Scand J Work Environ Health, 9, 273-281, 1983. V Fiserova-Bergerova, Modeling of inhalation exposure to vapors, Uptake, distribution and elimination, CRC Press Inc., Boca Raton, Florida, 1983. A Sato and T Nakajima, Brit. J. Ind. Med., 36, 231-234, 1979. A Morgan, A Black, DR, Belcher, Ann Occup Hyg., 13, 219-233, 1970. LE Farhi, Respir Physiol., 3, 1-11, 1967. GR Kelman, Br J. Ind Med., 39, 259-264, 1982. JE Cotes, Lung function, Assessment and application in medicine, Blackwell Scientific Publications, Oxford, 1979. J Widdicombe and A Davies, Respiratory physiology, Edward Arnold, London, 1983. HK Wilson, Breath analysis, Scand J Work Environ Health, 12, 174-192, 1986. V Fiserova-Bergerova, M Tichy, and FJ DiCarlo, Drug metabolism reviews, 15, 1033-1070, 1984. KM Dubowski, Clin. Chem., 20, 966-972, 1974. HC Niu, DA Schoeller, PD Klein, J Lab Clin Med., 91, 755-763, 1979. M Guillemin and E Guberan, Br J Ind Med., 39, 161-168, 1982. TA Robb TA and GP Davidson, Clinica Chimica Acta, 111, 281-285, 1981. JS Haldane, and JG Priestley, J Physiol., 32, 225-266. 1905. JH Raymer, KW Thomas, SD Cooper, DA Whitaker, ED Pellizzari, J Anal. Toxicol., 14, 337-344, 1990. SM Rappaport, E Kure, MX Petreas, D Ting, J Woodlee, Scand. J. Work Envir. Health, 17, 195-204, 1991.
1094
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
Myrto Petreas
M Petreas, J Woodlee, CE Becker, SM Rappaport, Int. Arch. Occup. Environ. Health, 67, 27-34, 1995. RH Hill, M Guillemin, PO Droz, in Methods for Biological Monitoring, TJ Kneip and JV Crable Eds. American Public Health Association, Washington DC, 1988. JF Periago, A Luna, A Morente, A Zambudio, J. Appl. Toxicol., 12, 91-96, 1992. JF Periago, A Cardona, D Marhuenda, J Roel, et al., Int. Arch. Occup. Environ. Health, 65, 275-278, 1993. L Campbell, AH Jones, HK Wilson, Am. J. Ind. Med., 8, 143-153, 1985. D Dyne, J Cocker, HK Wilson, The Sci. Total Envir., 199, 83-89, 1997. L Drummond, R Luck, AS Afacan, HK Wilson, Br. J. Ind. Med., 45, 256-261, 1988. Dragerwerk AG, Detector tube handbook, Lubeck, Germany, 1976. NC Jaim, J. Chromatogr. Sci., 12, 214-218, 1974. MF Mason and KM Dubowski, J. Forensic Sci., 4, 9-41, 1975. D Clasing, U Brackmeyer and G Bohn, Blutalcohol, 18, 98-102, 1981. CG Hunter and D Blair, Am. Occ. Hyg., 15, 193-199, 1972. R Teraniski, TR Mon, AB Robinson, P Cary, L Pauling, Anal. Chem., 44, 18-20, 1972. PO Droz and JG Fernandez, Br. J. Ind. Med., 35, 35-42, 1978. MX Petreas, SM Rappaport, B Materna, D Rempel, J. Exp. Analysis & Env. Epidem., Supplm. 1, 25-39, 1992 HK Wilson and TW Ottley, Biomed. Mass Spectrom., 8, 606-610, 1981. FM Benoit, WR Davidson, AM Lovett, S Nacson, and A Ngo, Anal. Chem., 55, 805-807, 1983. FM Benoit, WR Davidson, AM Lovett, S Nacson, and A Ngo, Int. Arch. Occup. Environ. Health, 55, 113-120, 1985. RJ Sherwood, and FWG Carter, The measurement of occupational exposure to benzene vapour. Ann. Occup. Hyg., 13, 125-146, 1970. DA Pasquini, Am. Ind. Hyg. Assoc. J., 39, 55-62, 1978. F Brugnone, L Perbellini, GL Faccini, F Pasini, L Romeo, M Gobbi, A Zedde, Int. Arch. Occup. Environ. Health, 61, 303-311, 1989. M Imbriani, S Ghittori, G Pezzagno, E Capodaglio, G. Ital. Med. Lav., 4, 271-278, 1982. RA Glaser, and JE Arnold, Am. Ind. Hyg. Assoc. J., 50, 112-121, 1989. L Wallace, R Zweidinger, M Erickson, S Cooper, D Whitaker, E Pellizzari, Environment International, 8, 269-282, 1982. WA Groves and ET Zellers, Am. Ind. Hyg. Assoc. J., 57, 1103-1108, 1996. AC Monster, G Boersma, H Steenweg, Int. Arch. Occup. Environ. Health, 42, 303-309, 1979. J Fernandez, E Guberan, J Caperos, Am. Ind. Hyg. Assoc. J., 37, 143-150, 1976. R Lauwerys, J Ferbrand, JP Buchet, A Bernard, J Gaussin, Int. Arch. Occup. Environ. Health, 52, 69-77, 1983. RD Stewart, HH Gay, DS Erley, CL Hake, AW Scaffer, Arch. Environ. Health, 2, 40-46, 1961. RD Stewart, ED Baretta, HC Dodd, TR Torkelson, Arch. Environ. Health, 20, 224-229, 1970. AC Monster, Int. Arch. Occup. Environ. Health, 42, 311-317, 1979. AC Monster, JM Houtkooper, Int. Arch. Occup. Environ. Health, 42, 319-323, 1979. E Guberan and J Fernandez, Br. J. Ind. Med., 31, 159-167, 1974. RC Ward, CC Travis, DM Hetrick, ME Andersen, ML Gargas, Tox. Appl. Pharmacol., 93, 108-117, 1988. RD Stewart, HC Dodd, ED Baretta, et al., Arch. Environ. Health, 16, 656-662, 1968. I Astrand, A Kilbom, P Ovrum, I Wahlberg, O Vesterberg, Work-Environm.& Health, 11, 69-85, 1974. J Engstrom, R Bjurstrom, I Astrand P Ovrum, Scand. J. Work Environ. Health, 4, 315-323, 1978. JC Ramsey, JD Young, RJ Karbowski, MB Chenoweth, LP McCarty, WH Braun, Toxicol. Appl. Pharmacol., 53, 54-63, 1980. E Wigaeus, A Lof, R Bjustrom, MB Nordqvist, Scand. J. Work. Environ. Health, 9, 479-488. 1983.
15.2.2 A simple test to determine toxicity using bacteria
1095
15.2.2 A SIMPLE TEST TO DETERMINE TOXICITY USING BACTERIA
James L. Botsford Department of Biology, New Mexico State University, Las Cruces, NM, USA
15.2.2.1 Introduction The author has developed a simple, inexpensive and rapid method to measure toxicity using a bacterial indicator. This paper describes this test in detail. It also reviews the field of alternative tests for toxicity, tests not involving viable animals. 15.2.2.2 Toxicity defined In toxicology, a compound is defined to be toxic if it damages living organisms. There is no chemical definition of toxicity. Classically, toxicity is determined from the LD50 for animals, the amount of the compound that kills 50% of the test animals (Rodericks, 1992). Typically varying concentrations of the toxin are given to groups of 10 animals and a single test can take 80 to 100 animals. Tests can take several weeks. Animal tests are expensive. The animals must be cared for and trained personnel are required. The animals are force fed the toxin so it passes through the acidic stomach or the toxins are introduced by intraperitoneal injection. This can alter the activity of the toxin. The results from animal tests are difficult to interpret since animals can die of causes other than the toxic chemical and analysis of the data usually requires sophisticated statistical analysis. In the United States it is estimated that 30 million animals die each year testing for toxic chemicals. Many question the value of animal tests (Ruelius, 1987). Results vary dramatically between strains of animals. Results with closely related species, rats and mice, often don’t agree well. The drug Thalidomide many years ago was tested in animals and was found to be harmless. The drug had tragic effects on humans. And, of course, animal tests do not note if the animal becomes ill from the toxin. They only note if the toxin kills the animal. Often it is not known why the chemical is toxic. It is simply observed that animals coming in contact with the toxin die at abnormally high rates. Obviously the first step in a study of a toxic chemical, is simply to determine if it is toxic. Most animal tests are run with rats and mice, the usual laboratory animals. Rats and mice are rodents and they have a caecum, a chamber that opens off the small intestine. The caecum contains many microorganisms. It is not known what the role of the aecum could be in handling toxic chemicals. This may influence results with rats and mice. Alternatives to animal tests are sought. The sand flea Daphnia magna is used extensively. Tests are inexpensive, most the animal rights advocates are not offended. Daphnia tests are difficult to perform and require highly skilled personnel. The tests typically take two days (Stephenson, 1990). Once it has been determined that the toxin kills the juvenile cells, it can be determined if the toxin interferes with maturation of the juveniles, with reproduction in the adults, or if the toxin simply inhibits growth. However, nearly all reports simply deal with death of the animals. Ceriodaphnia dubia offers another approach to this sort of test (Jung and Bitton,1997). There is a version of this test called CeriofastTM that takes only one hour. Daphnia testing requires laboratory skills that typically lab workers do not develop. The juvenile insects must be isolated using a dissecting microscope. The animals must be monitored to determine if they are living. It is necessary to watch the culture to know when to isolate the juveniles.
1096
James L. Botsford
In Europe many tests with animal and human cells have been developed (Clemendson et al., 1996). These tests are effective, provide values comparable to those obtained with animals. But they require that the animal cells be grown in culture. This requires a sophisticated laboratory and well trained personnel. Again the tests take several days. Some tests take up to a week. Damage to the cells is determined in a variety of ways. The ability of cells to reduce the thiazole tetrazolium dye, MTT, is noted; the activity of an enzyme, often lactic dehydrogenase, is determined; the ability of the cell to take up some dyes or to retain other dyes is observed. Often morphological changes are noted to determine if the cells are damaged. These techniques can be quite complicated and require well trained personnel and a sophisticated, well equipped, laboratory. There are several reports using animal cells to test the MEIC chemicals (Shrivastava et al., 1992; Rommert et al., 1994; Rouget et al., 1993). Shrivastava et al., used freshly isolated rat hepatocyte cells and two transformed cell lines. Comparable results were obtained with the three cell types indicating that rat hepatocytes are no more resistant to toxic chemicals than are other cells. They determined viability of the cells with 1) morphological studies; 2) lactic dehydrogenase activity in the hepatocytes; 3) the ability of the cells to take up the dye tryptan blue. They got much the same results with the three techniques. Rommert et al., simply determined the effect of the toxins on the ability of the cells to replicate. They determined cell numbers with a Coulter Counter after 48 hours. Rouget et al., followed the effect of the toxins by determining the ability of cells to reduce the tetrazolium dye MTT and the ability of cells to take up the dye neutral red. The authors maintained that the two methods provided comparable results. In Europe tests with rotifers, cysts from aquatic invertebrates (Calleja et al., 1994), development of onion roots have been tried but none have been accepted (Persoone et al., 1994; Snell and Persoone, 1989; Fresjog, 1985; Gaggi et al. 1995). Jaffe (1995) has proposed a method using the survival of a protozoan. Toxic chemicals kill the protozoa. The test requires a Coulter particle counter to determine the numbers of viable protozoa. Tests using bacteria as the indicator organism are on the market. MicrotoxTM uses a bioluminescent marine bacterium (Bullich et al., 1990). The bacteria, when growing in high numbers, emit light. Toxic chemicals inhibit the production of light. The test requires a luminometer to measure light production and a refrigerated water bath to grow the cells at 15oC. The results are difficult to interpret. The ability of the cells to emit light gradually decreases and a control to follow production of light in the absence of a toxin must be run. And the value for this enters into the calculations. Typically the calculations require at least a half hour to carry out (Ribo and Kaiser, 1987). A sophisticated computer program is used to analyze the results. The initial cost of the kit is significant, but it is inexpensive to use once this cost is absorbed (Elanabarwy et al., 1988). The test appears to be challenge for an MS level graduate student at our university. It is not certain why toxic chemicals inhibit light production but it is assumed that the chemicals damage the cytoplasmic membrane and reduce electron transport to the pigments responsible for light production. This method is very popular as a “first test”, to see if a chemical need be investigated further. The test has been used to determine the toxicity of more than a thousand chemicals (Kaiser and Palabrica, 1991). There is a European version of the same text, BiotoxTM (Kahru and Borchardt, 1994). Thomulka et al. (1995) have proposed a variation on this test. PolytoxTM uses a consortium of bacteria from sewage sludge. Toxic chemicals inhibit the oxidation of a carbon source (presumably glucose). Presumably the toxic chemicals damage the cellular membrane and the associated cytochromes involved in electron transport. Oxidation of the substrate is followed with a respirometer or with oxygen electrodes. Both these methods are complicated and expensive to set up and to maintain. Again, at our university, this technique appears to be appropriate for MS level graduate students. This test
15.2.2 A simple test to determine toxicity using bacteria
1097
is used to determine if chemicals will harm sewage treatment plants. It is rarely used to determine the toxicity of compounds. The test provides values comparable to those observed using sewage sludge as the source of bacteria (Sun et al., 1993; Toussaint et al., 1995; Elnabarawy et al., 1988). ToxitrakTM has been used to determine toxicity in water samples. There is a report (Gupta and Karuppiah, 1995) in which they tested water samples from a river in Maryland using this method, MicrotoxTM, Ceriodaphnia dubia (a variation of the standard Daphnia test), and this test. With four samples, they got comparable results. The Toxitrak test is not used extensively. Apparently the test follows the reduction of dye resazurin by cells of Bacillus cereus (Liu, 1989). Dozens of tests using bacteria as indicator organisms in toxicity studies have been proposed (Bitton and Dutka, 1986). Very few of these have been characterized extensively. Bacterial tests are inexpensive, inoffensive to animal rights groups, the cells are easy to grow and to prepare. And it is thought there is enough universality at the biochemical level, that a compound that is toxic for a bacterium will be toxic for a higher organism. Ideally an alternative test for toxic chemicals would be inexpensive. It would not require any specialized equipment, only equipment normally found in a laboratory should be required. The test should not require that the personnel be specially trained in the techniques. It should involve only the procedures that any laboratory worker would know. It should be the sort of test that any laboratory with a tangential interest in toxicology could carry out. Water quality laboratories, industrial safety laboratories, agricultural research laboratories, should be able to carry out the procedure. Small samples should be tested. Tests with fish often involve quite large volumes of toxic chemicals and this can limit their utility. This has been discussed in detail by Blaise (1991). The sensitivity of a test can be a problem. Every test for toxic chemicals has some chemicals that the test is very sensitive to. And every test has some toxic chemicals that it is not sensitive to. Often several tests will be run and the median value for toxicity is judged to be representative for the toxicity of that chemical. It can be argued that if a test system is extremely sensitive to a chemical, than it must be concluded that this chemical could be toxic for humans at this very low level. If it damages one life form, it could very well damage another. Tests with live animals, LD50 tests, are usually not as sensitive as tests with Daphnia, with animal cells, with bacterial tests. Tests with live animals usually involve injecting the toxin into the stomach and it is not always certain what effect the acidic conditions of the stomach may have on the toxin. And humans rarely come in contact with toxins in this fashion. It can be argued that this sort of animal test is not indicative of the toxicity of the chemical (Ruelius, 1987). On the other hand, chemical analysis is often much more sensitive than any toxicity test. There are reports of chemicals in the range of parts per trillion yet few tests for toxic chemicals can detect less than parts per million or occasionally, parts per billion. Yet we see public policy formulated on the basis of these chemical tests. 15.2.2.3 An alternative A simple, inexpensive and rapid test to determine the toxicity of chemicals has been developed. (Botsford, 1997, 1998, 1999). The assay uses the bacterium Rhizobium meliloti as the indicator organism. This bacterium reduces the thiazole tetrazolium dye, MTT (3-[4,5-dimethylthiazole-2-y]-2,5-biphenyl tetrazolium bromide) readily. The dye is almost completely reduced after a 20 minute incubation at 30oC. The dye turns dark blue when reduced and this can be followed readily with a simple spectrophotometer. Reduction of the dye is inhibited by toxic compounds. More than 200 compounds have been tested and the results obtained with this test are comparable to those obtained with other tests (Botsford, 2000a).
1098
James L. Botsford
Figure 15.2.2.1. Cyanazine is in units of µl. Cyanazine is a herbicide. The toxicity of cyanazine is: IC50 = [Y/2 -B]/m; Y = 0.35, 0.37 = 0.36, 0.36/2 = 0.18; B = 0.3487 m = 3.5058x10-3; IC50 = 48.1; 48.1µl = ([48.1 x 10-6 l]/test tube) x([3300 mg]/l) x [1 test tube]/3.3 ml)= 48.1 µg/ml = 48.1 mg/l = 48.1 ppm.
Since alternative tests don’t involve killing test animals, results are reported as IC50 values (inhibitory concentration, 50%), the concentration that results in death of 50% of the cells, in 50% inhibition of oxygen consumption, 50% inhibition of reduction of the dye in the Rhizobium assay. Cells are grown overnight in a simple defined medium with 0.1% casamino acids. Cells are collected by centrifugation in a refrigerated preparative centrifuge. Cells are washed once with 0.01 M (pH 7.5) potassium phosphate buffer and are then diluted in this same buffer. Volumes of the toxin (usually 5 to 1000 µl) are put into test tubes. The volume of the toxin is made up to 2.1 ml with water. 0.1 ml of 0.1 M Tris or Bicene buffer, pH 7.5, is added. The absorbancy at 550 nm is determined. 0.1 ml MTT (the tetrazolium dye) is added (3.2 mM). The tubes are incubated 20 minutes at 30oC. The absorbance is read a second time. The difference in the absorbance at time = 0 and at time = 20 min is determined. The results are plotted, the difference in absorbance of each tube with a different volume of the toxic chemical on the X axis versus the absorbance on the Y axis (Figure 15.2.2.1). A regression line is fit to the plot and if the regression coefficient is less than 0.8, the experiment is rejected. From the average value of the controls (tubes with no toxin), the Y intercept forms the regression equation and the slope, the volume of toxin inhibits reduction of the dye 50% can be readily calculated. For some toxins, better results are obtained if the log of
15.2.2 A simple test to determine toxicity using bacteria
1099
the volume of the toxin is plotted. From this the log of the volume corresponding to 50% inhibition of reduction can be determined and the antilog of this value provides the volume. In this laboratory the data is plotted both linearly and logarithmically and the best method providing the regression coefficient closest to 1.00 is chosen. From the concentration of the toxic compound, the volume found to inhibit reduction of the dye could be converted to mg/ml present in the test tube. From this the inhibitory concentration causing 50% inhibition, the IC50, can be calculated. The test has been described in detail (Botsford et al., 1997; Botsford 1998; Botsford, 1999). A method to lyophilize cells has been developed and cells are active for at least a month after being lyophilized (Robertson, 1996). This test for toxic chemicals is very inexpensive. It can be carried out by unskilled personnel. It is rapid, taking less than an hour to determine the toxicity of a compound. It provides values comparable to more conventional tests (Botsford, 2000a). It provides a sort of toxicological triage, a method to rapidly determine if something is toxic and should be investigated further. Since it uses bacteria rather than animals as an indicator, it is acceptable with animal rights advocates. The test has been patented and the author is seeking a firm to market the test. In the meantime, it can be carried out by any laboratory equipped to grow bacteria. Since this test uses a bacterium as the indicator organism. It can be argued that the results with a bacterium are not applicable to humans. But tests with rats and mice may not be applicable to humans (Ruelius, 1987). For example, in a test of 10 chemicals comparing results with mice, rats and five cellular assays, the toxicity of three of compounds differed by at least an order magnitude when rats and mice were compared (Ekwall et al., 1989). Such anomalies been observed with many drugs, that drugs were up to 12× more toxic for humans than for rats and mice (Brodie and Reid, 1967). At a biochemical level, a compound that damages the cytoplasmic membrane of a bacterium will also damage the cytoplasmic membrane of a human or any other organism. This test has been compared with 20 other tests for toxic chemicals and comparable results (similar IC50 values) are obtained (Botsford, 2000a). No one test finds all chemicals to be toxic. Every assay for toxic chemicals is blind to some compounds and conversely every assay for toxic chemicals is much more sensitive to some compounds than are other tests. Thus toxicologists desire “batteries of tests”, a compound is not considered to be toxic until it has proven to be toxic in 3 or 4 systems. 15.2.2.4 Chemicals tested In Table 15.2.2.1, the values for the more than 200 chemicals tested in this laboratory are reported. This includes a series of organic chemicals, chemicals that are reported in the literature as being toxic. It was noted that aromatics, phenol, benzene and toluene are not very toxic. But when they have substituents, a phenoxy group, chlorine, methyl or nitro they become quite toxic. Solvents were tested for toxicity and most solvents are not very toxic. We have found we can dissolve toxic compounds in solvents (methanol and DMSO) and find no change in the toxicity. Apparently the toxin is so much more toxic than the solvent that only the toxin affects the mechanism affording the assay. Many drugs were also tested and it is significant that many of these drugs are toxic for the bacterium and presumably would be toxic for patients receiving them. Often it is noted that when a drug is taken, an individual simply does not feel well. This could be due to toxicity of the drug. It was found that many alkaloids are toxic and then it was shown that many herbs are also toxic. Presumably the herbs contain toxic levels of some alkaloids. Some of these herbs are toxic enough that if mega herb therapy ever becomes popular, there could be problems from the toxicity of the herbs. It was found that divalent cations are toxic for the system. It was found that small amounts of EDTA could relieve this toxicity in soil and water samples. This will be dis-
1100
James L. Botsford
cussed. Neither lead nor arsenic acid were found to be toxic. Finally 30 herbicides were tested for toxicity (Hillaker, 1996). The toxicity determined by this method was compared with the toxicity reported by the manufacturers using rats, ducks, and quail in classical LD50 tests; and with trout fingerlings and with Daphnia. The tests with trout fingerlings and with Daphnia were much more sensitive than the tests with animals. The tests with Rhizobium were less sensitive than with Daphnia and the trout fingerlings for most of the herbicides but were always more sensitive than tests with animals. Table 15.2.2.1. Toxicity of chemicals tested n
Ave.
Var.
n
Ave.
Var.
1,1,1-trichloroethane
20
229
10
pentachlorophenol
13
0.143
50
1,1,1-tricholorethylene
13
66.3
27
2-methyl resourcinol
4
404
19
1,2-dichloroethane
2
146.5
m-cresol (mono methyl)
6
98.9
14
1,10-phenanthroline
5
42.7
o-cresol
22
190
29
acetic acid
8
429
p-cresol
4
386
15
acetonitrile
3
>3300
2,4-dinitro-o-cresol
7
12.8
39
butyl amine
2
271
2,6-dinitro-p-cresol
7
33.5
35
coumarin
7
222
21
4,6-dinitrocresol
5
6.83
31
CTAB (detergent)
7
0.90
36
toluene
4
217
8
cyclohexyl amine
9
234
5
toluidine
6
1493
26
diethyl amine
2
194
trihydroxytoluene
8
52.4
34
dimethyl amino ethanol
2
>1200
dinitrotoluene
4
66.7
6
dibromethane
8
217
7
trinitrotoluene
8
52.4
34
EDTA
6
3788
14
benzene
4
799
18
ethyl acetate
11
1107
17
benzyl chloride
8
60.6
8
ethyl benzene
8
71
6
chlorobenzene
12
23.9
26
ethanol amine
8
251
16
1,2,4-trimethyl benzene
14
37.6
40
hexachlorophene
7
0.077
9
dimethyl amino benzaldehyde
1
236
hydrogen peroxide
10
164
28
trichlorobenzene
5
14.7
18
hydroxyl amine
13
566
20
p-amino benzaldedhyde
3
156
4
ibenofuran
4
9.61
20
1,2-dichlorobenzene
11
37.7
5
indole
3
501
5
1,3-dichlorobenzene
7
41.4
11
isonicotinic acid
5
287
18
1,4-dinitrobenzene
3
313
8
octane
5
161
20
pseudocumine
14
37.6
40
α-naphthol
9
56.9
34
p-dimethyl amino benzaldehyde
7
177
5
potassium ferricyanide
5
3600
15
p-amino benzoic acid
3
291
10
pentanol
4
1627
22
p-OH benzoic acid
5
252
27
10
15.2.2 A simple test to determine toxicity using bacteria
n
Ave.
Var.
phenanthroline
5
43.3
20
Tris (Trizma, pH 7.5)
3
5447
MOPS 3-[N-morpholino] 5 propane sulfonate
1101
n
Ave.
Var.
2-OH benzoic acid (salicylic acid)
3
217
13
16
p-amino benzoic acid
4
307
2
13090
27
chlorobenzoic acid
6
233
19
tetrachloroethylene
16
91.2
21
4-chlorobenzoic acid
5
234
14
tetrachloromethane
10
1791
12
3-phenoxybenzoic acid
4
62.7
25
salicylic acid
4
198
20
3(3,5-dichlorophenoxy) benzoic acid
5
41.9
20
sulfosalicylic acid
11
1107
17
3(4-chlorophenoxy) benzoic acid
7
32.9
43
xylene
4
132
27
3(4-methyl phenoxy) benzoic acid
6
92.1
9
phenol
16
1223
28
3(3,4-dichlorophenoxy) benzoic acid
4
45.9
24
4-chlorophenol
6
186
23
3-chlorophenol
6
3.21
45
acetone
18 123000
24
3-methyl-4-nitro phenol
4
54.9
27
acetic acid
8
429
10
2,4-dichlorophenol
4
68.7
5
coumarin
7
222
21
3,4-dichlorophenol
5
82.3
74
chloroform
10
1550
22
3,5-dichlorophenol
5
174
37
ethanol
20 75581
28
2,3-dimethyl phenol
10
120
5
DMSO
13 87629
31
2,4-dimethyl phenol
11
71.4
28
dichloroethane
28
237
38
2,4,6-trimethyl phenol
10
123
9
dibromomethane
8
217
15
3,4,5-trimethyl phenol
6
49.5
4
isopropyl alcohol
3
57000
19
2,3,6-trimethyl phenol
6
151
9
methanol
4
68172
13
dimethyl amino methyl phenol
10
413
24
pentanol
4
1623
22
trichlorophenol
45
8.24
38
ethylene glycol
5
212700
12
o-nitrophenol
13
37.6
33
Solvents
Herbs
Detergents
Aloe Vera
4
219
4
CTAB
7
0.90
36
Cayenne
4
700
1
SDS
8
31.4
14
Echinacea
7
451
1
Tween 20
5
2896
22
Echinacea root
5
939
4
Elderberry
4
291
1
Lindane
8
41.1
28
Evening Primrose
11
35
17
Malathion
14
37.1
40
Myrrh
6
71.2
3
Diazanon
18
35.6
24
Insecticides
1102
James L. Botsford
n
Ave.
Var.
Herbs continuation
n
Ave.
Var.
Drugs
Gensing Drink
9
285
9
acetomenaphen
9
249
20
Royal Jelly
6
1641
22
acetylsalicylic acid
8
447
11
Saw Palmetto
21
149
24
aspirin (Bayer)
10
170
21
Chromium Picolinate
6
206
7
St. John’s Wort
20
192
12
St. John’s Wort
6
218
4
St. John’s Wort
6
639
17
Amoxicilin Cyprus Jordan Mexico America
3 5 9 4
4412 4501 4257 7393
37 51 9 52
Thyme
4
338
5
chloramphenicol
5
1400
24
Valarian Root
12
208
13
gentamycin
5
0.68
7
Valarian Root
5
200
4
GEA-1 (nitroso drug)
5
50.5
39
2
>1200
Car’s Claw, Garlic Oil, Ginko Bioloba, Ginseng, Sin-1 (nitroso drug) Golden Seal, and Herb Garlic were found not to be ibuprofen toxic (less than 2000 ppm). Herbs listed more than once were obtained from different suppliers. isoproterenol
Herbicides
6
120
1
1770
8
kanamycin
5
0.85
14
2,4-D
9
347
39
naproxin
3
296
41
Alachlor
6
111
8
neomycin
8
0.438
31
Bensulide
7
53.7
18
novobiocin
2
2063
Bromoxynil
8
26.0
19
Orudis
6
114
14
Clomazone
15
23.8
16
orphenadrine
6
472
32
Cyanizine
13
781
27
paparavine
7
95.1
15
DCPA
6
42719
streptomycin
5
60.4
28
Dicamba
6
>1200
Diuron
9
87.0
13
EPTC
9
51.4
6
Tetracycline America Mexico
4 4
672 653
7 2
Ethalfluralin
6
>1200
Fuazifop-P
6
>1200
Glyphosate
6
18.1
12
Imazapyr
6
229
1
Imazethapyr
10 35314
Isoxaben
16
365
Mepiquat-Cl
6
>1200
Metasulfuron
6
>1200
The national names after the antibiotics indicate the country the antibiotics originated in.
15
Inhibitors of electron transport dinitrophenol
4
34.3
17
FCCP
19
0.20
36
potassium cyanide
17
12.2
48
sodium azide
17
630
19
15.2.2 A simple test to determine toxicity using bacteria
n
Ave.
1103
Var.
n
Ave.
Var.
atropine
9
191
11
Herbicides continuation
Alkaloids
Metribuzin
6
>1200
Naproamide
10
289
43
caffeine
3
3700
18
Nicosulfuron
10
267
11
immitine
11
122
8
Norflurazon
8
182
14
nicotine
13
990
13
Oxadiazon
6
269
16
quinine
4
131
14
Quniclorac
6
>1200
qunidine
4
137
15
Sethoxydim
6
2.70
17
scopolomine
5
129
5
Thiazopyr
6
43.2
4
Thifensulfuron
7
928
4
Trifluralin
7
10.3
18
n = the number of samples tests; ave. = the average of the values obtained, reported as part per million (ppm), mg per liter; var. = the variance in the results, the standard deviation divided by the mean.
15.2.2.5 Comparisons with other tests The Rhizobium test has been compared with other tests. This is done by finding values in the literature for various toxins using different tests and plotting the log of these values vs. the log of values from the second assay. A regression line was plotted and the correlation coefficient calculated from the regression coefficient (the correlation coefficient is the square root of the regression coefficient). In Figure 15.2.2.2, values for chemicals from two laboratories
Figure 15.2.2.2. Representative plot of the comparison of two assays. These are data from Daphnia 1 (Calleja et al., 1993) and Daphnia 2 (Lilius et al.,1994).
1104
James L. Botsford
(Calleja et al., 1993; Lilius et al., 1994) using the assay involving the sand flea Daphnia magna are presented. If two systems give the same results, the correlation coefficient will be 1.00. In the literature there are reports of the toxicity of 50 MEIC chemicals (MEIC, multicenter evaluation of cytotoxicity) tests used in Europe to evaluate different tests for toxicity. When these two tests with Daphnia were compared, a correlation coefficient of 0.895 was obtained (Figure 15.2.2.2). This indicates that when a standardized test is run in two laboratories with different personnel, the results are not identical. This discrepancy is noted in tests involving animals. Apparently variations in the strains of animals, differences in care and other factors influence the results of toxicity tests with animals. Often variations of 10x are observed. This sort of analysis was carried using the Rhizobium assay. Published values for the chemicals using Microtox, LD50 reports for rats, IC50 (testing with animal cells), Daphnia magna, and HLD (Human Lethal Dose determined from autopsy reports) are included (Table 15.2.2.2). These values for the assay with Rhizobium will permit comparisons to be made (Table 15.2.2.3). It should be noted that the values for rats and HLD are much higher than with the other methods. With animal tests, the animals are force fed the chemical, it is injected through a tube into their stomach. Occasionally the toxin is injected intraperitoneally. It is uncertain what happens to the toxic chemical in the acidic stomach. This methodology has been criticized because humans are rarely exposed to toxic chemicals in this fashion. HLD data comes from autopsy reports and it can never be certain how much of the chemical the victim actually came in contact with, it can only be estimated. And this method only provides data for chemicals already in the environment, it is not a good method to predict toxicity.
rat hepat, m moles
HLD, m moles
Daphnia, m moles
IC50, m moles
Rat, m moles
Microtox, m moles
Chemical tested
Rhizobium, m moles
Table 15.2.2.2. MEIC chemicals tested. Comparison
acetomenaphen
1.649
2.19
15.8
1.45
0.269
1.698
10.75
acetylsalicylic acid
1.106
0.145
1.122
1.7
0.932
2.818
2.662
amitriptyline
0.0146
0.078
1.148
0.1
0.02
0.2
0.07
barium chloride
0.109
caffeine
1.02
3.388
1
1.58
3.388
1
1.596
carbon tetrachloride
1.91
4.786
15.1
22.1
126
0.447
3.984
chloroform
5.29
12.9
7.586
2.63
1.738
6.198
chlororamphenicol
4.332
1.122
7.7
copper II sulfate
0.007
0.01
1.86
0.001
0.316
0.048
dichloromethane
4.122
37.1
18.6
10.5
4.17
109.1
digoxin
0.231
ethanol
1643
ethylene glycol*
3427
1.349
0.794
0.54
0.47
5.248
0.402
0.426
0.12
12.8
0.0014
0.271
691
151
158
234
109.6
451
1778
75.8
322
1202
25.7
358
0.138
0.0068
0.00027
isopropyl alcohol
950
380
83.2
90.5
155
41.7
304
iron II sulfate
0.32
0.782
2.089
0.302
3.020
1.621
lindane
0.141
21.9
0.263
malithion
0.112
0.479
0.871
0.00006
0.0002
3.715
methanol
2130
912
178
nicotine
6.1
0.224
0.309
orphenadrine HCI*
1.168
0.005
paraquat
0.262
mercury II chloride
0.36
rat hepat, m moles
0.019
HLD, m moles
Daphnia, m moles
0.0002
Rat, m moles
hexachlorophene
Microtox, m moles
Chemical tested
IC50, m moles
1105
Rhizobium, m moles
15.2.2 A simple test to determine toxicity using bacteria
0.002
0.0056
0.144 0.741
0.00013
0.107
0.003
673
661
52.5
906
4.52
0.0.23
0.11
3.581
0.832
0.033
0.098
0.114
2.344
0.224
0.1
0.166
1.176
0.00054
0.02
0.1
0.025
0.0022
0.098
0.05
phenol
13
0.186
3.39
4.25
0.077
1.48
0.797
potassium chloride
290
493
34.6
490
0.288
92.22
potassium cyanide
0.225
0.275
0.0776
1
0.0141
0.437
0.783
quinidine sulfate
0.422
1.202
0.617
0.036
0.0797
sodium chloride
288
562
51.3
sodium oxalate
>13
5.428
theophylline
1.926
13.8
1.349
2.63
0.724
1,1,1-trichloroethane
0.583
0.342
77.6
6.71
1.48
warfarin
1.781
0.209
0.005
1.549
0.0219
0.139
xylene
1.24
0.079
34.6
0.851
0.468
17.47
pentachlorophenol
60.3
0.129 17
102 0.582
0.67
2.175
Mw = molecular weight of the compound; n = number of times compound was tested; var = variation in the results, the standard deviation /mean\ toxicity reported as mmoles toxin for the IC50 for the test. The rat test is an LD50. Rhizobium data are from this work. MicrotoxTM, HLD, and Daphnia data are from Calleja, (1993). The IC50 data (pooled data from animal cell tests) is from Halle et al., 1993. Rat hepatocyte data is from Shrivastava (1992). The IC50 samples include only 20 chemicals. The other methods involve about 34 chemicals.
In Table 15.2.2.3 the results of comparisons among these tests are summarized. The data from these 6 determinations were plotted, one assay versus another assay, the regression coefficient noted and the correlation coefficient calculated. The Rhizobium assay was the most sensitive for 12 of 33 chemicals. The Microtox assay was most sensitive for 6 of the compounds. The IC50 assay was most sensitive for 6 of the compounds. The Daphnia test was most sensitive to 10 of the compounds. The average values for the toxicity of the chemicals was lowest for the IC50, but then it was also the test with the fewest values included.
1106
James L. Botsford
Table 15.2.2.3. Comparisons for involving Rhizobium, MicrotoxTM, HLD, IC50, Daphnia magna and rats n
r2
n
r2
Rhizobium x MicrotoxTM
34
0.875
Rats x IC50
19
0.724
Rhizobium x rats
34
0.603
Rats x HLD
27
0.817
Rhizobium x IC50
20
0.906
Rats x Daphnia
32
0.685
Rhizobium x HLD
34
0.628
Rats x rat hepatocytes
34
0.748
Rhizobium x Daphnia
34
0.888
IC50 x HLD
16
0.732
Rhizobium x rat hepatocytes
34
0.764
IC50 x Daphnia
29
0.879
Microtox x rats
34
0.571
IC50 x rat hepatocytes
20
0.700
Microtox x IC50
24
0.840
HLD x rat hepatocytes
26
0.648
Microtox x HLD
31
0.728
HLD x Daphnia
29
0.702
Microtox x rat hepatocytes
33
0.700
Daphnia x rat hepatocytes
33
0.814
Microtox x Daphnia
34
0.846
n = number of samples in the test. One test was compared with another, the log of the values plotted, a regression line fit and the correlation coefficient determined from the regression coefficient, r2. The data for rats, Daphnia, HLD is from Calleja et al., 1994. The data for IC50 is from Halle et al., 1992. The data for rat hepatocytes is from Shrivastava et al., 1992.
In Table 15.2.2.4, the results of these comparisons with the results of the Rhizobium test are presented. The Rhizobium test correlates well with MicrotoxTM, Daphnia, and particularly well with the 20 samples in the IC50 test. The comparison of the Rhizobium assays with other published results has been examined in detail (Botsford, 2000a). The Rhizobium test has been compared with values from several laboratories for the MicrotoxTM assay and correlation coefficients have varied from 0.750 to 0.893 indicating the two methods provide comparable results. Four comparisons with values for Daphnia have provided correlation coefficients from 0.776 to 0.953 indicating that the Rhizobium assay agrees well with Daphnia. QSAR provides a method to predict the toxicity of a compound from the structure of the compound and the water/octanol partition coefficient (Nirmilkhandan and Speece, 1988). This method has been compared with assays using sewage sludge and PolytoxTM (Sun et al., 1993). The method compares well with both these other techniques. An examination of the data provided by tests for toxic chemicals using these techniques suggested that they are not as sensitive as more direct methods for determining toxicity. It was found that MicrotoxTM, rat hepatocytes, the Rhizobium assay, and Daphnia all provided more sensitivity to toxic chemicals than did the QSAR estimates (Botsford, 2000a). The IC50 values for these other techniques were lower than those computed using the QSAR method. QSAR data are not often found in the literature.
15.2.2 A simple test to determine toxicity using bacteria
1107
Table 15.2.2.4. The Rhizobium assay compared with other assay methods n
cc
3T3 cells x Rhizobium
9
0.908
Ekwall et al., (1989)
Asictes cells x Rhizobium
34
0.870
Romert et al., (1994)
24
0.928
Kahru and Bordchardt (1994)
B. subtilis x Rhizobium
20
0.619
Kherzmann (1993)
Daphnia 1 x Rhizobium
35
0.891
Calleja et al., (1993)
Daphnia 2 x Rhizobium
35
0.897
Lilius et al., (1994)
pooled Daphnia x Rhizobium
14
0.775
Calleja, Liilius, Munkitrick
E. coli x Rhizobium
20
0.641
Kherzman (1993)
fathead minnow x Rhizobium
15
0.942
Munkitrick et al., (1991)
guppies x Rhizobium
9
0.950
Konemann (1981)
HepG2 cells x Rhizobium
9
0.892
Ekwali et al., (1989)
Biotox
TM
x Rhizobium
Reference
TM
x Rhizobum
35
0.803
Kaiser and Palabrica (1991)
TM
x Rhizobium
15
0.758
Munkittrick et al., (1991)
mice x Rhizobium
35
0.651
Calleja et al., (1991)
TM
1 x Rhizobium
16
0.903
Sun et al., (1993)
TM
2 x Rhizobium
15
0.796
Elanabarwy et al., (1988)
sludge x Rhizobium
16
0.853
Sun et al., (1993)
trout hepatocytes x Rhizobium
35
0.760
Lilius et al. (1994)
trout figerlingx x Rhizobium
15
0.910
Munkittrick et al., (1991)
QSAR x Rhizobium
19
0.779
Sun (1994)
pooled minnow x Rhizobium
32
0.808
Geiger et al., (1991), Munkitrick et al., (1991)
Microtox
Microtox
Polytox Polytox
n = number of samples in the comparison; cc = the correlation coefficient, the square root of the regression coefficient.
15.2.2.6 Toxic herbicides When the toxic herbicides were studied, it was found that the animal tests supplied by the manufacturers indicated that most of the herbicides were not toxic. The tests run by the manufacturers with trout fingerlings and with Daphnia indicated that most of the herbicides were quite toxic. The Rhizobium work also showed that most of the herbicides were toxic at levels lower than 1000 ppm. The trout fingerlings showed all but one of the herbicides was toxic at this level and Daphnia indicated that all but 2 were toxic (Hillaker, 1998). This shows the necessity of running “batteries of tests,” multiple tests with a compound. Every method of determining toxicity has some chemicals it cannot detect as toxic and some chemicals that are detected at very low levels. For example, the Rhizobium assay does not detect phenol as being very toxic but detects pentachlorophenol as being extremely toxic,
1108
James L. Botsford
the test is at least an order of magnitude more sensitive to pentachlorophenol than any other test found. 15.2.2.7 Toxicity of divalent cations Reduction of the dye is inhibited by divalent cations in the Rhizobium system. Common ions, calcium and magnesium, inhibit the reduction of the dye. The toxicity of the ions is shown in Table 15.2.2.5. Mercury and cadmium, generally thought to be the most toxic minerals were the most toxic with this assay. Calcium and magnesium are also toxic. Water and soil samples typically contain calcium and magnesium so in order to analyze water and soil samples for toxic organic chemicals, a method to eliminate this inhibition by metal ions was sought. Table 15.2.2.5. Toxicity of divalent cations Minerals
n
var.
m moles
Ba+2
6
33
Cd Ca
+2
+2
Co
+2
Cu
+2
12 11 11 13
41 33 46 19
Minerals
n
var.
m moles
0.109
Mg+2
9
20
0.404
0.004
+2
11
35
0.045
+2
0.05 0.009 0.007
Fe
+2
7
18
0.587
Fe
+3
14
27
0.098
Mn Hg
10
23
0.0006
+2
11
11
0.452
+2
9
39
1.849
+2
10
7
0.062
Ni Se
Zn
n = number of samples tested; var. = variation, standard deviation divided by the mean. Values reported as m moles l-1. All minerals except ferrous ion as chloride salts. Several were tested as both chlorides and sulfates with little difference. When ferrous sulfate was tested, it was made up fresh each day before the assay.
EDTA is used routinely in biochemistry to chelate divalent metal ions. It was thought this might chelate the calcium and remove it from the system. EDDA and EGTA are also used and are thought to chelate calcium more effectively than EDTA. Neither of these chelators affected the reaction, both were simply slightly toxic (about 3000 ppm). A series of experiments were run and it was found that the inhibition of reduction caused by 1.4 to 1.6 µmoles of calcium was relieved by 1 µmoles EDTA. There is not a stoichiometric relationship between EDTA and the metal ion. It is not simply chelating the metal ion. It was found that 2.5 µmoles EDTA would eliminate the toxicity of all the ions at their IC50, the concentration of cation that inhibited reduction of the dye 50%. Thus 2.5 µmoles EDTA would eliminate toxicity from 5.5 µmoles calcium but only 0.006 µmoles mercury. It was observed initially that the toxicity of most organic chemicals could be determined in the presence of 2.5 µmoles EDTA. This suggested there could be two mechanisms involved in the reduction of MTT. One is inhibited by toxic organic chemicals and the second is inhibited by divalent cations. 15.2.2.8 Toxicity of organics in the presence of EDTA A series of experiments were run looking at the toxicity of organic chemicals in the presence of 0.74 µmoles calcium (25 ppm, this concentration inhibits reduction of the dye completely) and 2.5 µmoles EDTA. Four of the 35 chemicals tested had greater toxicity with EDTA and calcium than in the controls. Two chemicals were no longer toxic. The toxicity of 16 of the chemicals was not affected by the calcium and EDTA. The toxicity of 13 chemicals was decreased by at least 10% but was not eliminated by the addition of EDTA and cal-
15.2.2 A simple test to determine toxicity using bacteria
1109
cium. The toxicity of 4 chemicals was enhanced, was greater, when the calcium and EDTA were present. The addition of calcium and EDTA at these concentrations had no effect on the apparent toxicity of: 1,4-dinitrobenzene, the herbicide 2,4-D, 2,4-dinitrotoluene, 4-chloro-benzoate, carbon tetrachloride, chloroform, cynazine, hexachlorophene, isonicotinic acid the insecticide Lindane, o-nitrophenol, p-toluidine, the antibiotic Streptomycin, tetrachloroethylene and trichlorophenol. The addition of calcium and EDTA reduced the toxicity at least 10% of 2,6-dinitrocresol, 2,4-dinitrophenol, 2,6-dinitrophenol, 2,6-dinitrotoluene, 2-methyl resourcinol, 3-phenoxybenzoate, 2,4-dinitrocresol, the detergent CTAB, the antibiotic Neomycin, p-hydroxybenzoate, pentachlorophenol, salicylic acid, and trichloroethylene. p-amino benzoic acid and p-hydroxy benzoate were no longer toxic with these levels of calcium and EDTA. 3-chlorobenzoate, 3-methyl-4-nitro-phenol, and the detergents sodium lauryl sulfate and Tween 80 had increased toxicity with the addition of calcium and EDTA. No correlation between the structure of the chemical and the effect of EDTA and calcium could be ascertained. For example, streptomycin and neomycin are both amino glycoside antibiotics. Calcium and EDTA did not affect the toxicity of streptomycin but nearly eliminated the toxicity of neomycin. Several chemicals were tested with 25 µmoles EDTA and 7.4 µmoles of calcium, 10 times the amount used in the experiments reported. The results of this experiment are shown in Table 15.2.2.6. In the presence of high levels of EDTA and calcium, isonicotinic acid is no longer toxic. Dinitrophenol and 3-phenoxy benzoate, in the presence of low levels of EDTA and calcium were not as toxic, in the presence of high levels, it had the same toxicity. In presence of low levels of EDTA and calcium, Streptomycin was comparably toxic, the calcium and EDTA did not affect the toxicity. In the presence of high levels, Streptomycin was extremely toxic. Pentachlorophenol, a common soil contaminant, had reduced toxicity with low levels of EDTA and calcium, had elevated levels of toxicity with high levels of EDTA and calcium. This work shows that the effect of EDTA and specific toxins must be worked out before any conclusions as to the toxicity of the compound in the presence of EDTA and calcium can be established. Calcium and magnesium are commonly found in water. Obviously if this assay is to be used with water samples, EDTA must be added. It must be determined using water uncontaminated with organic toxins how much EDTA must be used to compensate for the divalent cations. Often the concentration of divalent cations is determined by atomic absorption spectroscopy. However, these values do not agree with the toxicity relieved by EDTA. Soil samples with as much as 5 gm calcium (45 mM) per kg soil have been assayed using 2.5 µmoles EDTA in each sample (Hillaker, 1996). The calcium is complexed with sulfate and phosphate ions and the calcium is not available to the cell, is not seen by the mechanism that reduces the dye. Levels of soluble calcium and magnesium in water are very low. We have found that 2.5 µmoles of EDTA relieves the inhibition caused by divalent cations in all water and soil samples tested thus far (Botsford, 2000b).
1110
James L. Botsford
Table 15.2.2.6. Effect of high concentrations of EDTR and calcium on toxicity (Inhibition in %) Compound
2.5 µmole EDTA, 0.74 µmole Calcium
25 µmole EDTA, 7.4 µmole Calcium
Control
+ EDTA, + Ca
Control
+ EDTA, + Ca
isonicotinic acid
27.0
25.4
30.9
105
2,4-dinitrophenol
44.7
63.0
58.2
50.0
Streptomycin
55.0
48.9
47.2
0.00
Neomycin
0.037
62.4
77.2
77.3
pentachlorophenol
48.0
63.7
26.0
13.3
3-phenoxybenzoate
54.3
74.0
34.5
37.8
15.2.2.9 Mechanism for reduction of the dye It is not known how the dye is reduced. It is not known why toxic chemicals inhibit the reduction. It is thought that tetrazolium dyes are reduced by cytochromes (Altman, 1975). But this has been questioned (Marshall et al., 1993). In eucaryotic cells, all the cytochromes are in the mitochondria. Marshall’s group has found that the dyes are reduced in preparations from cells with the mitochondria removed. It has been found that a mutant of Escherichia coli lacking one of two major cytochromes found in this bacterium is unable to reduce MTT (Botsford, unpublished). But in E. coli, reduction of the dye is not inhibited by toxic chemicals. This may not be an analogous situation. The dye could be reduced by a different mechanism. R. meliloti, like most other bacteria, has many reductases. Some of these are membrane associated and damage to the membrane could affect the reductase. One of these reductases could be responsible for reduction of the dye. It has been found that the MTT is transported into the cell before it is reduced. The reduced dye is inside the cells. Cells with the dye can be concentrated by centrifugation, the dye appears in the cell pellet. None of the dye is in the supernatant. Toxic chemicals could interfere with the transport of dye into the cell prior to reduction. Transposon insertion mutants unable to reduce the dye have been obtained and five mutants have been isolated. All grow very slowly in minimal media supplement with 0.1 % casamino acids and obviously all have lost a critical function. With these mutants it should be possible to clone and then to sequence the function responsible for reduction of the dye. From the sequence, the nature of the function can be determined. In our studies comparing the Rhizobium assay with other assays, it was observed that tests using viable animals were almost always less sensitive to toxins. Tests using Daphnia, the various animal cell tests and MicrotoxTM and the Rhizobium test seemed to be most sensitive. Tests using fish (fathead minnow, trout fingerlings) give results comparable to tests with the bacterial indicators. PolytoxTM and QSAR were less sensitive, had higher IC50 values, but were more sensitive than the tests using viable animals. Were the author asked to recommend a test procedure to indicate if a chemical were toxic, the author would recommend an animal cell test, probably using freshly isolated rat liver hepatocytes, the Rhizobium test and the MicrotoxTM test. Tests with freshly isolated rat hepatocytes would not require that the cells be grown in a laboratory situation and this would be much simpler. These three procedures are much more sensitive than tests involved live animals. These three tests would be simpler than tests with Daphnia. This would provide a “battery of
15.2.2 A simple test to determine toxicity using bacteria
1111
tests”, would not offend animal rights advocates and should indicate if the chemical is dangerous. All could be performed by personnel with chemical laboratory skills. 15.2.2.10 Summary This work shows that the Rhizobium test provides results comparable to other tests. Tests seem particularly comparable to work with Daphnia magna and with results from in vitro tests with animal cells. The test is simple, unskilled laboratory workers can master it quickly. The test is inexpensive, no specialized equipment is required, given cells, any laboratory able to carry out simple chemical analysis should be able to perform the assay. The test is rapid, a sample can be tested and analyzed in an hour, the test does not take several days. It offers an ideal first test for toxic chemicals (Blaise, 1991) References Altman, F. P. 1976. Tetrazolium salts and formazans. Progress is Histochemistry and Cytochemistry 9:6-52. Bitton, G., Dutka, B. J. (1986) Toxicity testing using microorganisms. CRC Press Inc. Boca Raton, Florida 163 pp. Blaise, C. 1991. Microbiotests in aquatic toxicology. Environmental Toxicology and Water Quality. 6:145-151. Botsford, J. L., Rivera, J., Navarez, J., Riley, R., Wright T., Baker, R. 1997. Assay for toxic chemicals using bacteria. Bulletin of Environmental Contamination and Toxicology 59:1000-1008. Botsford, J. L. 1998. A simple assay for toxic chemicals using a bacterial indicator. World Journal of Microbiology and Biotechnology. 14:369-376. Botsford, J. L. 1999. A simple method for determining toxicity of chemicals using a bacterial indicator organism. Environmental Toxicology 99:285-290. Botsford, J. L. 2000a. A comparison of alternative tests for toxic chemicals. To be submitted ATLA journal. Botsford, J. L. 2000b. Role of EDTA in a simple method for determining toxicity using a bacterial indicator organism. World Journal Microbiology and Biotechnology, in press. Brodie, B., Reid, W.D., 1967. Some pharmacological consequences of species variation in rates of metabolism. Federation Proceedings 26:1062-1070. Bullich, A. A., Tung, K-K, Scheiber, G. 1990. The luminescent bacteria toxicity test: Its potential as an in vitro alternative. Journal of Bioluminesence and Chemiluminescenmce 5:71-77. Calleja, M. C., Persoone, G., Geiadi, P. (1994) Comparative acute toxicity of the first 50 multicenter evaluation of in vitro cytotoxicity chemicals to aquatic nonvertebrates. Archives Enviornmental Contamination and Toxicology 26:69-78. Calleja, M. C., Persoone, G., Gelandi P. 1993. The predictive potential of a batter of exotoxilogical tests for human acute toxicity, as evaluated with the first 50 MEIC chemicals. ATLA 21:330-349. Clemendson, C., McFarlane-Abdulla, E., Andersson, M., Barile, FA., Calleja, M.G., Chesne, C., Clotheir, R., Cottin, M. Curren, R., Dierickx, P., Ferro, M., Fiskejo G, Garza-Ocanas, L., Gomez-Lecon, M.J., Golden, M., Isomaa, B, Janus, J., Judge, P., Kahru, A., Kemp, R.B., Kerszman, G., Kristen, U. Kunimoto, M., Kaarenlapi, S., Lavrijsen, K., Lewan, L., Lilius, H., Malmsten, A., Ohno, T., Persoone, G., Pettersson, R., Roguet, R., Romert, L., Sandberg, M., Sawyer, T.W., Seibert, H., Shrivastava, R., Sjostrom, Stammati, A., Tanaka, N., Torres-Alanis, O., Voss, J-U. Wakuri,S., Walum, E., Wang, X., Zucco, F., Ekwall, B. (1996). MEIC evaluation of acute systemic toxicity. ATLA 24:273-311. Ekwall, B, Bondesson, 1, Catell, J.V., Gomez-Lechon, M. J., Heiberg, S., Hogberg, J. Jover, R., Ponsoda, X., Rommert, L., Stenberg, KL., Walum, E. (1989) Cytoxocity evaluation of the first ten MEIC chemicals: Acute lethal toxicity in man predicted by cytotoxicity in five cellular assays and by oral LD50 tests in rodents. ATLA 17:83-100. Ekwall, B., Johansson, A. 1980. Preliminary studies on the validity of in vitro measurements of drug toxicity using HeLa cells I. Comparative in vitro cytotoxicity of 27 drugs. Toxicology Letters 5:299-307. Einabarawy, M. T., Robideau, R. R., Beach, S. A. (1988) Comparison of three rapid toxicity test procedures: MicrotoxTM, PolytoxTM and activate sludge respiration inhibition. Toxicity Assessment 3:361-370. Fresjog, G. 1985. The allium test as a standard in environmental monitoring. Hereditas, 102:99-112. Gaggi, C., Sbrilli G., A.M. Hasab El Naby, Bucci, M., Duccini, M., and Bacci, E. 1994. Toxicity and hazard raking of S-triazine herbicides using Microtox, 2 green algal systems and a marine crustacean. Environmental Toxicology and Chemistry 14:1065-1069.
1112
James L. Botsford
Geiger, D. L., Brooke, L. T., Call, D. J. editors (1990) Acute toxicities of organic chemicals to fathead minnows (Pimphates promeias). Center for Lake Superior Environmental Studies. University of Wisconsin, Superior, Wisconsin. 900 pp. Gupta, G., Karuppiah, M. 1996. Toxicity identification of Pocomoke River porewater. Chemosphere 33:939-960. Halle, W., Baeger, I., Ekwall, B., Spielmann, H. 1991. Correlation between in vitro cytotoxicity and octanol/water partition coefficient of 29 substances from the MEIC program. ATLA 19:338-343. Hillaker, T. L. 1996. An assay for toxic chemicals using Rhizobium method as the indicator: Use of this test with agricultural herbicides. MS Thesis, Biology. New Mexico State University. Jaffe, R. L., 1995. Rapid assay of cytotoxicity, using Teratamitus flagellates. Toxicology and Industrial Health 11:543-553. Jung, K., Bitton, G. 1997. Use of CeriofastTM for monitoring the toxicity of industrial effluents: Comparison with the 48-H acute Ceriodaphnia toxicity test and MicrotoxTM. Experimental Toxicology and Chemistry 16:2264-2267. Kahru, A., Borchardt, B. 1994. Toxicity of 39 MEIC chemicals to Bioluminescent photobacteria (The BiotoxTM test): Correlation with other test systems. ATLA 22:147160. Kaiser, K.L.E., Palabrica, V. S., 1991. Photobacterium phosphoreum toxicity data index. Water Pollution Research Journal of Canada. 26:361-431. Kerszman, G. 1993. Toxicity of the first ten MEIC chemicals to bacteria. ATLA 21:151155. Kerzman, G. 1993. Of bacteria and men: Toxicity of 30 MEIC chemicals to bacteria and humans. ATLA 21:233-238. Konemann, H, 1981, Quantitative structure-activity relationship in fish toxicity studies. Toxiocology 19:209-221. Lilius, H., Isomaa, B., Holstrom, T. A comparison of the toxicity of 50 reference chemicals to freshly isolated rainbow trout hepatocytes and Daphnia magna. Aquatic Toxicology 30:47-60. Liu, D. 1989. A rapid and simple biochemical test for direct determination of chemical toxicity. Toxicity Assessment 4:389-404. Marshall, N. J., Goodwin, C. J., Holt, S. J. 1995. A critical assessment of the use of microculture tetrazolium assays to measure cell growth and function. Growth Regulation 5:69-84. Mossman, T . 1983. Rapid colorimetric assay for cellular growth and survival: Application to proliferation and cytotoxicity assays. Journal of Immunological Methods 65:55-63. Munkittrick K. R., Power, E. A., Sergy, G. A. 1991. The relative sensitivity of MicrotoxTM, Daphnia, rainbow trout, fathead minnow acute lethality tests. Environmental Toxicology and Water Qualtity 6:35-62. Nirmalakhandan, N. N., Speece, R. E. 1988. Prediction of aqueous solubility of organic chemicals based on molecular structure. Environmnental Science and Technology. 22:328-338. Robertson, B. 1996. Developing a technique to lyophilize Rhizobium mefloti MS thesis, Biology, New Mexico State University. Rodericks, J. V.1992. Calculated Risks. Cambridge, Cabridge University Press 256 pp Romert, L., Jansson, T. Jenssen, D. 1994. The cytotoxicity of 50 chemicals from the NEIC study determined by growth inhibition of Ascites Sarcoma BP8 cells: A comparison with acute toxicity data in man and rodents. Toxicology Letters 71:39-46. Rouguet, R., Cotovia, J. Gaetani, Q., Dossou K. G. Rougier, A. 1993. Cytotoxicity of 28 MEIC chemicals to rat hepatocytes using two viability endpoints: correlation with acute toxicity data in rat and man. ATLA 1:216-224. Ruelius, H. W. 1987. Extrapolation from animals to man: predictions, pitfalls and perspectives. Xenobiotica 17:255-265. Shrivastava, R., Deiominie, C., Chevalier, A., John, G., Ekwall, B., Walum, E. Massingham, R. Comparison of in vitro acute lethal potency and in vitro cytotoxicity of 48 chemicals. Cell Biology and Toxicology 8:157-167. Snell, T. W., Personne, G. 1989. Acute bioassays using rotifers. II. A freshwater test with Brachionus rubens. Aquatic Toxicology 14:81-92. Stephenson, G. L., Kausik, N. K., Solomon, K. R. 1991. Chronic toxicity of a pure and technical grade pentachlorophenol to Daphnia magna. Archives Environmental Contamination and Toxicology. 21:388-394 Sun, B., Nimalakhandan, N., Hall, E., Wang, X. H., Prakash, J., Maynes, R. (1994) Estimating toxicity of organic chemicals to activated-sludge microorganisms. Journal of Environmental Engineering 120:1459-1469. Sun, B. (1993). Comparison of interspecies toxicity of organic chemicals and evaluation of QSAR approaches in toxicity prediction. MS Thesis, Environmental Engineering. New Mexico State University. Thomulka, K. W., McGee, D. J., Lange, J. H. 1993. Detection of biohazzardous materials in water by measuring bioluminescence with the marine organism Vibrio harveyi. Journal Environmental Science and Health. A28: 2153-2166. Toussaint, M. W., Shedd, RT. R., van der Schalie, W. H., Leather, G. R. (1992) A comparison of standard acute toxicity tests with rapid-screening toxicity tests. Environmental Toxicology and Chemistry 14:907-915.
15.2.3 An innovative GC method
1113
15.2.3 DESCRIPTION OF AN INNOVATIVE GC METHOD TO ASSESS THE INFLUENCE OF CRYSTAL TEXTURE AND DRYING CONDITIONS ON RESIDUAL SOLVENT CONTENT IN PHARMACEUTICAL PRODUCTS
Christine Barthélémy Laboratoire de Pharmacie Galénique et Biopharmacie Faculté des Sciences Pharmaceutiques et Biologiques Université de Lille II, Lille, France
Michel Bauer International Analytical Sciences Department Sanofi-Synthélabo Recherche, Toulouse, France
The presence of residual solvents (RS) in pharmaceutical substances occurs for various reasons. Solvents are involved in all steps of raw material synthesis and pharmaceutical productions. The search for the presence of RS in a pharmaceutical product and their concentrations are now mandatory in any new monographs (as detailed in Chapter 16.2). The RS remaining in the crystals of pharmaceutical products may be the cause of health disorders because, when a drug is taken every day, chronic toxicity may occur. The presence of RS may have other consequences, such as modifying stability, organoleptic characters, pharmacotechnical parameters (flow properties, crystalline form, compression ability) and biopharmaceutical characteristics, that may fluctuate according to RS content (as detailed in chapters 14.21.1 and 16.2). It is therefore necessary to reduce the residual solvents contained in crystalline particles as much as possible. It is well known that solvents can exist in three different states within the crystals: • Solvents adsorbed on the crystal faces: these are generally easily desorbed during conventional drying because the binding forces between solvents and crystals are very weak. • Occluded solvents such as microdroplets in the crystal: these are often difficult to extract. Generally, they can escape when the crystal is being dissociated: - during grinding, potentially leading to clodding, - during storage leading to very compact aggregates. • Solvents bound to drug molecules in the crystal and known as “solvates”. These bound solvents escape at a characteristic temperature, producing desolvated forms; the solvate and the desolvated forms are two different crystalline entities that can exhibit very different mechanical behaviors. The main objective of any chemist crystallizing pharmaceutical raw materials should be either the total elimination of the organic solvents or the significant reduction of RS level in order to be below the regulatory limits.1 15.2.3.1 Description of the RS determination method There are several analytical methods to assess the RS content of drugs. Among them, the gas chromatography (GC) is largely preferred. We refer to the chapter 16.2 for more details. In direct injection methods, the products in which RS are included are usually dissolved in an appropriate solvent and then directly injected into a gas chromatograph (GC). The main problem with these techniques is that non-volatile substances are gradually retained in the column, causing a rapid decrease in its sensitivity and efficiency; this is one of the reasons why headspace techniques are increasingly used instead.
1114
Christine Barthélémy, Michel Bauer
To avoid these problems, we have developed a method consisting initially of the complete dissolution of the substance in an appropriate solvent followed by micro-distillation and then finally of the injection of the distillate into the GC. This technique allows for the complete recovery of RS without the drawbacks mentioned above.2,3 The operating conditions are as follows: the powders to be analyzed are poured into a micro-distilling flask. An appropriate solvent is chosen to allow the product as well as the RS to be completely dissolved. The solvent Figure 15.2.3.1. Photograph of the micro-distilling deshould not interfere with the RS extracted; for vice. example, in the following studied cases: methanol for dioxane, chloroform, tetrahydrofuran, hexane and dichloromethane determinations and 1-butanol for ethanol determination. Assay solutions were obtained by completely dissolving 500 mg of the product to be tested in 5 ml of the appropriate solvent. The micro-distillation is carried out until complete dryness and is followed by condensation of the solvent vapor. The distillate is collected in a gauge glass set on ice. The assembly of the system is photographed on Figure 15.2.3.1. In this way the non-volatile substances stay entrapped in the micro-distilling flask and the totality of the RS are recovered. The distillate is then directly injected into the GC. The method was validated (specificity, linearity, repeatability, reproducibility...) by distilling, in the same operating conditions, defined standard solutions of solvents. The recovery of RS is very good as can be seen in the example reported on Table 15.2.3.1. 15.2.3.2 Application: Influence of crystal texture and drying conditions on RS content After washing the crystals, drying is the most effective way of lowering the content of organic volatile impurities. The required drying conditions differ greatly according to the solvent state in the crystalline particles, to the thermodynamic events that can occur when the substance is heated, and also to the texture of these particles. Table 15.2.3.1. Example of a calibration curve and recovery calculations of dioxane distilled in methanol Dioxane standards area under curve and variation coefficient (%)
Dioxane recovered after distillation area under curve and variation coefficient (%)
% and ppm recovered
44.8 ppm
17161.0± 705.5, v.c: 4.11 % 16819.3± 631.5, v.c: 3.75 % 98.01 %, 43.9 ppm
56 ppm
25904.7±1493.1, v.c: 5.76 %
25869.0± 748.4, v.c: 2.89 % 99.86 %, 55.9 ppm
112 ppm
66147.3±1010.4, v.c: 1.53 %
66831.7± 472.6, v.c: 0.71 % 101.03 %, 113.2 ppm
168 ppm
108157.7±2026.6, v.c: 1.53 %
107361.0±1546.4, v.c: 1.44 %
99.26 %, 166.8 ppm
224 ppm
146674.0±1490.4, v.c: 1.02 %
146722.7±2469.7, v.c: 1.68 %
100.03 %, 224.1 ppm
linear regression
Y = 724.74 X - 14855.6
Y = 724.29 X - 14889.7
correlation coeff.
0.9999
0.9999
mean : 99.64± 1.11 % v.c : 1.1 %
15.2.3 An innovative GC method
1115
Among all parameters influencing RS content, crystal texture is of utmost importance. It is evident that a crystal exhibiting a porous texture will enable the easy escape of a solvent while a compact and dense crystal will retain the solvent inside its structure whatever the type of particles: monocrystalline (i.e., monoparticular) or “polycrystalline”. The term “polycrystalline” particles will be employed to designate elementary particles that can be composed of agglomerates, spherolites or “spherical crystals” according to Kawashima.4 To illustrate the importance of the texture of particles on RS content, we can consider some examples taken from our laboratory experiments. Several crystallization processes were investigated on pharmaceutical products with different solvents leading either to compact monocrystalline particles or “sintered-like” (i.e., microcrystallites fitted into each other and partially welded involving a porous texture) or polycrystalline particles which were more or less dense.2,3 The physical study of these particles such as, optical microscopy, particle size analysis, electron scanning microscopy and thermal analysis, have been used to link the RS contents with the drying conditions of crystals and to demonstrate that the optimal drying conditions differ greatly according to the texture of the particles. As we shall see in a first example of monocrystalline particles of paracetamol recrystallized in dioxane, a wide open texture is generally favorable to a low RS content after a progressive drying at a moderate temperature to avoid the formation of a superficial crust. In fact, in this case, the drying conditions of crystals highly influence their residual solvent content. Optimal drying conditions seem to be the progressive and moderate ones. In contrast, too drastic drying conditions may hinder the solvent escape by a “crust” effect. This crust is due to a drying temperature which is too high, leading to a melt and a dissolution of the surface of the crystals. Finally, when a desolvation occurs during the drying, it can modify the texture of crystals and form a crust. The intensity of the phenomenon depends on the solvent. To remove the solvents efficiently, it is then necessary to exceed the desolvation temperature of the solvates that may be produced during crystallization. With this example we can clearly point out that the crystal texture is a determining factor in the complete escape of the solvent: when the crystalline texture is sintered-like, after the desolvation of solvate crystals, progressive drying is necessary to prevent the “crusting” phenomenon. The drying conditions should also be adapted to the area offered to the evaporation of solvent, particularly in the case of polycrystalline particles presenting a high porosity and a large surface to be dried. This important surface is due to the disordered rearrangement of very small crystals inside the particles. To illustrate this, we shall see in a second example the cases of spherical crystals of meprobamate and ibuprofen agglomerates. In both cases the RS can escape easily from the large surface of polycrystalline particles. Lastly, we shall see in a third example, that we must not forget that thermodynamic phenomenon can occur under drying. This is a very particular example because the pharmaceutical product used (paracetamol) presents a polymorphic transition that can occur during the drying phase and lead to a new organization inside the crystal, allowing the escape of the RS. This can be very interesting in the case of products with a low transition temperature. The usual polymorphic form of commercialized paracetamol is the monoclinic form; but in particular cases small amount of metastable orthorhombic form can be obtained. In the case of paracetamol, at the transition temperature (156°C) we observe a brutal solvent escape that could be due to the solid-solid transition undergone by the orthorhombic form into the monoclinic one. The disorder produced and the increase of the specific volume occurring during this first order transition allow the occluded solvents to escape more easily.
1116
Christine Barthélémy, Michel Bauer
15.2.3.2.1 First example: monocrystalline particles of paracetamol Preparation of monocrystalline particles In this first example, paracetamol was recrystallized separately in three different types of solvents (ethanol, water and dioxane) with different boiling points, molecular weights, dielectric constants and paracetamol solubilising power. After the crystallization process,5 crystals were separated by filtration under vacuum and washed with the same crystallization solvent. Each batch was divided into four fractions to be dried differently. Particle drying conditions The paracetamol crystals were submitted to four different drying conditions2 (Table 15.2.3.2): Drastic: under vacuum at 100°C for 3 hours. Drastic: in a ventilated oven at 100°C for 3 hours. Progressive: in a ventilated oven at 60°C for 1 hour, at 80°C for 1 hour and finally at 100°C for 1 hour. Very moderate: under ventilated hood at 20°C for 24 hours and then in a ventilated oven at 35°C for 48 hours. Morphological aspect of the particles Dried crystals were observed by optical and scanning electron microscopy (SEM) in order to measure their Ferret mean diameter and to determine their habit and texture. The mean diameters of crystals obtained from different solvents and submitted to different drying conditions are reported in Table 15.2.3.3. Table 15.2.3.2. Drying conditions studied on the different particles Type of particles
monocrystalline
Drugs
paracetamol
Type of drying
Temperature
Time
drastic under vacuum drastic under ventilation
100°C 100°C 60°C + 80°C + 100°C 20°C + 35°C 90°C 60°C + 75°C + 90°C 60°C 40°C 156°C 156°C 100°C 100°C + 156°C 60°C + 80°C + 100°C 60°C + 80°C + 100°C + 156°C
3h 3h 1h +1h +1h 24 h + 48 h 2h 30 min + 30min +1h 2h 2h 10 min 30 min 2h 2h + 10 min 30 min + 30 min +1h 30 min + 30 min +1h + 10 min
progressive very moderate drastic
meprobamate
ibuprofen
polycrystalline
progressive moderate very moderate drastic flash drastic flash drastic drastic + flash
paracetamol progressive
progressive + flash
15.2.3 An innovative GC method
Figure 15.2.3.2. SEM photograph of paracetamol recrystallized in dioxane and submitted to moderate drying (Photograph from reference2).
1117
Figure 15.2.3.3. SEM photograph of paracetamol recrystallized in dioxane and submitted to progressive drying (Photograph from reference2).
The optical microscopy investigation shows some significant differences in mean diameter and crystal habit due to crystallization conditions. In the case of water and ethanol, the crystals are transparent and rather regular, their habit is generally parallelepipedal. The only difference to be noted is the mean diameter of these particles; the particles recrystallized in water being 3 times larger. ESM reveals a surface that remains smooth in all drying conditions; their texture seems compact and dense. In contrast, crystals obtained from dioxane are rather different to those obtained from ethanol and water: they are opaque to transmitted light and their habit is irregular. Figure 15.2.3.4. SEM photograph of paracetamol Their mean diameter is slightly smaller than recrystallized in dioxane and submitted to drastic drying crystals from ethanol. (Photograph from reference2). It is to be noted that all the crystals obtained from the three solvents present the same polymorphic form: the monoclinic form. Table 15.2.3.3: Mean Ferret diameters of particles (Data from references2,3) Monocrystalline particles paracetamol / dioxane paracetamol / ethanol paracetamol / water
Mean diameter, µm 122 149 538
Polycrystalline particles meprobamate spherical crystals ibuprofen agglomerates paracetamol spherolites
Mean diameter, µm 145 350 113
1118
Christine Barthélémy, Michel Bauer
Scanning electron microscopy (SEM) carried out on these crystals reveals different behaviors according to the drying conditions tested. The crystals obtained from dioxane are similar to blocks of sintered particles since they consist of microcrystallites bound to each other, as if partially welded.5 In fact, the measurement of mercury porosity indicated a very high porosity. This sintered aspect concords with the non-transmission of light through the whole crystal. In fact, paracetamol forms a solvate with dioxane.5 Its departure does not modify the crystal habit but the surface can be seen as perforated. When the drying is very moderate, the crystal surface is perforated by numerous holes through which solvent escapes; these anfractuosities give the impression of a porous sintered-like texture (Figure 15.2.3.2). With progressive drying, the number of holes decreases (Figure 15.2.3.3). On the contrary, when the drying is drastic, the surface is relatively smooth (Figure 15.2.3.4). This can be explained by the too drastic drying conditions involving the formation of a crust at the surface of the crystal preventing the escape of the solvent from the crystal. Residual solvent determination Gas Phase Chromatography (for dioxane and ethanol) was performed on a Varian 1440 Chromatograph with a Flame Ionization Detector; packed column Porapack Super Q (Alltech, France), mesh range 80/100, length 1.8 m; internal diameter 2.16 mm; carrier gas was nitrogen (40 ml/min); injector: 210°C; detector: 250°C. For dioxane: column temperature: isotherm at 170°C. Injection: 10µl. Retention Times (RT): methanol 1.5 min; dioxane 16 min. For ethanol: column temperature: isotherm at 210°C. Injection: 5µl. RT: ethanol 1.5 min; 1-butanol 4 min. Determination of residual water on paracetamol crystallized in water: according to the titrimetric direct method of Karl Fischer. The residual solvent concentrations of the crystals obtained are reported in Table 15.2.3.4. Table 15.2.3.4. Residual solvent content of crystals obtained from different solvents and submitted to different drying conditions (Data from reference2) Drastic
Drastic under vacuum
Progressive
Moderate
Dioxane
126 ppm
183 ppm
25 ppm
53 ppm
Ethanol
2045 ppm
2501 ppm
1495 ppm
2072 ppm
0.59 %
0.68 %
0.51 %
0.56 %
Water
Whatever the drying conditions may be, the content of dioxane is always lower than the 380 ppm ICH limit1,6 (Table 15.2.3.5). Progressive drying always gives lower level residual solvent. In fact, drastic drying, in a ventilated oven and under vacuum, leads to the formation of a superficial “crust” which hinders the solvent escape; this can be clearly visualized on the scanning electron photomicrograph (Figure 15.2.3.4); and can explain the relatively high content of residual solvents.
15.2.3 An innovative GC method
1119
Table 15.2.3.5. Solvent class and concentration limits in pharmaceutical products (Data from references1,6) Class
Concentration limit, ppm
Chloroform
2
60
Dioxane
2
Ethanol
3
Class
Concentration limit, ppm
Hexane
2
290
380
Dichloromethane
2
600
5000
Tetrahydrofuran
3
5000
15.2.3.2.2 Second example: polycrystalline particles of meprobamate and ibuprofen Preparation of polycrystalline particles The polycrystalline particles were produced using various crystallization processes and designed to obtain directly compressible particles of pure drug, as tablets cannot be formed by direct compression of the raw materials. Spherical crystals of meprobamate These spherical particles were prepared following the usual preparation process for spherical crystals described by Guillaume.7 Spherical crystals of meprobamate appear when stirring a mixture of three liquids in the crystallizer: methanol allowing meprobamate to dissolve; water, as a non-solvent, causing meprobamate precipitation; chloroform as a bridging liquid to gather in its dispersed droplets, meprobamate microcrystallites that finally form “spherical crystals”. Ibuprofen agglomerates Ibuprofen agglomerates were prepared by a phase separation process in a mixture of ethanol and water (50/50 v/v).8 The saturated solution obtained at 60°C was constantly stirred and cooled down to room temperature. Particle drying conditions Depending on the melting point of the drugs, different drying conditions were applied in a ventilated oven. Drastic and progressive temperature conditions were studied for each type of polycrystalline particles (Table 15.2.3.2). Spherical crystals of meprobamate Taking into account the melting point of meprobamate (105°C), temperatures higher than 90°C must be avoided. The different drying conditions were: drastic drying at 90°C for 2 hours and progressive drying at 60°C for 30 minutes, then at 75°C for 30 minutes and finally at 90°C for one hour. Ibuprofen agglomerates This is a particular case; a low drying temperature must be applied because of the very low melting point of ibuprofen (76°C). The two drying conditions studied were: 40°C for two hours or 60°C for two hours. Morphological aspect of the particles Spherical crystals of meprobamate Meprobamate crystals appear as more or less rounded opaque particles (Table 15.2.3.3). Their consistence is friable. Particle size distribution is very narrow. SEM photograph of meprobamate crystals shows nearly spherical particles; their surface seems apparently smooth (Figure 15.2.3.5). However, at high magnification, the surface appears to be coated with flat crystals (Figure 15.2.3.6).
1120
Figure 15.2.3.5. SEM photograph of “spherical crystals” of meprobamate (Photograph from reference3).
Figure 15.2.3.7. SEM photograph of the cross-section of “spherical crystals” of meprobamate (Photograph from reference3).
Christine Barthélémy, Michel Bauer
Figure 15.2.3.6. SEM photograph of the surface (high magnification) of “spherical crystals” of meprobamate (Photograph from reference3).
Figure 15.2.3.8. SEM photograph of ibuprofen agglomerates (Photograph from reference3).
The cross-section surface shows a dense tangling-up of flat crystals inside the particle (Figure 15.2.3.7). Flat crystals are concentrically disposed at the periphery of the rounded particles, like a shell. The inside of the particle is a disordered arrangement of small flat crystals. Thus, the surface offered to evaporation is very important and this should lead to the easy escape of RS. Ibuprofen agglomerates Ibuprofen particles are large agglomerates of flat crystals with quite soft consistence and a nearly rounded shape (Table 15.2.3.3).
15.2.3 An innovative GC method
1121
Ibuprofen crystals observed under SEM are in total disorder (Figure 15.2.3.8). This disordered arrangement of microcrystallites inside the polycrystalline particles gives them a high isotropy of particles arrangement that should improve their compression capacity. Residual solvent determination For organic solvents, gas phase chromatography was performed on a Shimadzu GC-14B chromatograph fitted with a Flame Ionization Detector and a CR-6A Shimadzu integrator. The packed column was Porapack super Q (Alltech, France), mesh range 80/100, length 1.80 m, internal diameter 2.16 mm. Carrier gas: anhydrous nitrogen. Injector temperature: 200°C. Detector temperature: 220°C. The chromatographic conditions were: • For chloroform in meprobamate crystals: isotherm at 150°C, injection: 5 µl, RT: methanol 0.8 min, chloroform 6.7 min. • For ethanol in ibuprofen crystals: isotherm at 170°C, injection: 5 µl, RT: ethanol 1.1 min, 1-butanol 5.3 min. Determination of residual water was carried out using Karl Fischer’s titrimetric direct method after calibration with natrium tartrate and dissolution of ibuprofen or meprobamate crystals in methanol. The residual solvent concentrations of the polycrystalline particles are reported in Tables 15.2.3.6 and 15.2.3.7. Table 15.2.3.6. Residual solvent concentration of meprobamate spherical crystals submitted to different drying conditions (Data from reference3) Chloroform
Water
Progressive drying*
345±22 ppm
0.71%±0.07%
Drastic drying**
321±8 ppm
0.21%±0.01%
*30 min at 60°C + 30 min at 75°C + 1 hour at 90°C; **2 hours at 90°C
Table 15.2.3.7. Residual solvent concentration of ibuprofen agglomerates submitted to different drying conditions (Data from reference3) Ethanol
Water
2 hours at 40°C
42±3 ppm
0.76%±0.12%
2 hours at 60°C
21±4 ppm
0.44%±0.06%
As far as meprobamate spherical crystals are concerned, no significant differences are to be observed between drastic and progressive drying. No crusting phenomenon appears on meprobamate spherical crystals due to the loose tangling up of crystals, and so the solvent may escape easily between them. The residual ethanol content of ibuprofen agglomerates is very low because of the open texture of agglomerates. Moreover, the crystallization phenomenon was relatively slow, enabling the solvent to escape from crystals in formation. The higher the temperature, the lower the residual ethanol content. Both these crystals have a very porous texture. It seems that progressive drying is not essential as far as the polycrystalline particles are concerned. If we consider the official limits1,6 reported on Table 15.2.3.5 for residual solvent contents, we can note that the concentration of chloroform in meprobamate spherical crystals is much higher than the limit allowed in any drying conditions. Due to its inherent toxicity, this solvent should be avoided in the recrystallization process of meprobamate. The solvent
1122
Christine Barthélémy, Michel Bauer
content obtained for ethanol in ibuprofen agglomerates is very low, in all drying conditions; in all cases it is considerably lower than the tolerated limits. 15.2.3.2.3 Third example: polycrystalline particles of paracetamol Preparation of polycrystalline particles Paracetamol agglomerates were prepared by the spontaneous precipitation of paracetamol into a mixture containing hexane, tetrahydrofuran and dichloromethane9 under stirring. All the crystals obtained were filtered under vacuum. Particle drying conditions The melting point of paracetamol being 169°C and the transition temperature being 156°C, different drying conditions were tested (Table 15.2.3.2): • Drastic flash drying: 156°C for either 10 or 30 minutes, • Drastic drying: 100°C for 2 hours, • Progressive drying: 60°C for 30 minutes, then 80°C for 30 minutes, and finally, 100°C for 1 hour. After drastic drying at 100°C and progressive drying, a test with complementary drying at 156°C was carried out for 10 minutes. All the recovered crystals were packaged in glass flasks before gas phase chromatography and other analysis. Morphological aspect of particles Several types of texture and morphology are to be observed in polycrystalline particles, according to solvent proportions. The most interesting and particular example are spherical polycrystalline particles which have a radial texture (spherolites) and appear as urchin-like particles (Table 15.2.3.3). SEM reveals the very typical surface crystallization of the agglomerates (Figure 15.2.3.9). They are made up of parallelepipedal flat crystals arranged perpendicularly to a central nucleus and they are relatively strong. The implantation of peripherical crystallites is perpendicular to the surface. As it has been clearly demonstrated by Ettabia10 a Figure 15.2.3.9. SEM photograph of paracetamol ag- nucleus is formed first and then, in a second glomerates (Photograph from reference3). step, microcrystallites grow on it. Residual solvent determination For organic solvents, gas phase chromatography was performed on a Shimadzu GC-14B chromatograph fitted with a Flame Ionization Detector and a CR-6A Shimadzu integrator. The packed column was Porapack super Q (Alltech, France), mesh range 80/100, length 1.80 m, internal diameter 2.16 mm. Carrier gas: anhydrous nitrogen. Injector temperature: 200°C. Detector temperature: 220°C. The chromatographic conditions were: For dichloromethane, tetrahydrofuran and hexane in paracetamol crystals: isotherm at 150°C, injection: 5 µl, RT: methanol 0.8 min, dichloromethane 3 min, tetrahydrofuran 7.8 min, hexane 11.8 min.
15.2.3 An innovative GC method
1123
The residual solvent concentrations of the polycrystalline particles are reported in Table 15.2.3.8. The wide-open texture of paracetamol spherolites hinders the crusting phenomenon; it is therefore normal that there should be no difference between the effectiveness of the two types of drying: progressive and drastic (at 100°C for two hours). However, a surprisingly good result is observed when the drying temperature is 156°C whereas a prolonged time at 100°C does not improve the solvent escape in spite of the low boiling points of solvents, all much lower than 100°C (dichloromethane: 39.5°C; tetrahydrofuran: 66°C; hexane: 69°C). Using Figure 15.2.3.10. DSC and thermogravimetry curves of a DSC method, it can be observed that a paracetamol agglomerates. slight non constant exotherm at 85°C followed by a small constant endotherm at nearly 156°C occurs before the melting endotherm at 169°C (Figure 15.2.3.10). In fact, the paracetamol obtained by the crystallization process described is mainly the usual monoclinic form. However, as has been demonstrated by Ettabia10 the formation of a certain amount of the amorphous form, causes the slight exotherm. The amorphous form recrystallized at 85°C into the orthorhombic metastable form, which transits into monoclinic form at 156°C, causing the small endotherm observed at nearly 156°C. The thermogravimetric analysis shows that no solvation occurs during crystallization. Furthermore, the sudden solvent escape is not specific to one definite solvent, whereas all the contained solvent amounts dramatically decrease at 156°C. The loss of mass is high (about 0.7%) when the substance is about to melt (Figure 15.2.3.10). Table 15.2.3.8. Residual solvent concentration of paracetamol agglomerates submitted to different drying conditions (Data from reference3) "Flash drying" 156°C for 10 min for 30 min PD* PD*+ “flash” (10 min at 156°C) DD ** DD**+"flash" (10 min at 156°C)
Dichloromethane
Tetrahydrofuran
Hexane
< 200 ppm < 200 ppm 315±3 ppm < 200 ppm 309±3 ppm < 200 ppm
1095±5 ppm 630±24 ppm 2066±1 ppm 883±10 ppm 2014±30 ppm 873±2 ppm
167±1 ppm 88±4 ppm 305±3 ppm 145±5 ppm 313±4 ppm 140±1 ppm
*
PD: Progressive drying (30 min at 60°C + 30 min at 80°C + 1 h at 100°C), **DD: Drastic drying (2 h at 100°C)
Paracetamol, containing three different solvents, exhibits low dichloromethane content (200 to 320 ppm). The referencing limits remain higher than the experimental results when these particles are submitted to different drying conditions (Table 15.2.3.5). The same observation is valid for tetrahydrofuran for which the allowed upper limit is 5000 ppm. But the residue in hexane is sometimes above the regulatory threshold. The only way to be definitely below this limit is to heat paracetamol to 156°C; this confirms the advantage of flash drying this substance. As previously described, this temperature is critical for paracetamol recrystallization in the solvents used.
1124
Christine Barthélémy, Michel Bauer
This study enables us to underline some considerations about the drying conditions of crystals. It is important to note that not only should the usual parameters be taken into account, such as the melting point of the substances and the boiling point of the solvents, but also the texture of the particles and the thermodynamic events that can occur when the substance is heated. The kinetic of temperature increase must be chosen with respect to crystalline particle texture to obtain optimal drying conditions for lowest RS content. Progressive drying can give better solvent elimination when the texture is very finely sintered to avoid the crusting effect, which hinders any ulterior solvent escape.2 When a solvate is formed, the drying temperature must be either equal to or higher than the desolvation temperature. Lastly, the knowledge of the existence of solid-solid transitions could be in this respect of great interest. Thermogravimetry and DSC analysis are, among others, very efficient tools to assess these phenomena. References 1 2 3 4 5 6 7 8 9 10
European Pharmacopoeia, 3rd edition, addendum 1999, pp. 216-224. C. Barthélémy, P. Di Martino, A-M. Guyot-Hermann, Die Pharmazie, 50, 609 (1995). A. Ettabia, C. Barthélémy, M. Jbilou, A-M. Guyot-Hermann, Die Pharmazie, 53, 565 (1998). Y. Kawashima, M. Okumura, H. Takenaka, Science, 216, 1127 (1982). J-M. Fachaux, A-M. Guyot-Hermann, J-C. Guyot, P. Conflant, M. Drache, S. Veesler, R. Boistelle, Powder Techn., 82, 2, 123 (1995). Note for Guidance on impurities: Residual solvents, Drugs made in Germany, 41, 98 (1998). F. Guillaume, A-M. Guyot-Hermann, Il farmaco, 48, 473 (1993). M. Jbilou, A. Ettabia, A-M. Guyot-Hermann, J-C. Guyot , Drug Dev. & Ind. Pharm., 25, 3, 297 (1999). A. Ettabia, E. Joiris, A-M. Guyot-Hermann, J-C. Guyot, Pharm. Ind., 59, 625 (1997). A. Ettabia, European Thesis, Lille II University, France (8/10/1997).
16
Residual Solvents in Products 16.1 RESIDUAL SOLVENTS IN VARIOUS PRODUCTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada
There are physical and chemical barriers that control solvent removal from solid-solvent systems. The most basic relation is given by the following equation: W =
P1 Kw
[16.1.1]
where: W P1 Kw
equilibrium fraction of residual solvent partial pressure of solvent in vapor phase Henry’s law constant
Both the partial pressure and Henry’s law constant depend on temperature, pressure, and solvent properties. This relationship does not consider interaction between solute and solvent. In the case of polymers, the Flory-Huggins theory gives a simplified relationship for low concentrations of solvent: ln
P1
= ln φ1 + 1 + χ
P10
[16.1.2]
where: P10 φ1 χ
vapor pressure of pure solvent volume fraction of solvent Flory-Huggins interaction parameter
Vapor pressures of some solvents can be found in the referenced monograph.1 The weight fraction of residual solvent at equilibrium can be calculated from the following equation, which accounts for polymer-solvent interaction: W =
P1 ρ 1 exp− (1 + χ ) P10 ρ 2
where: ρ1 ρ2
density of solvent density of polymer
[16.1.3]
1126
George Wypych
The last equation does not give the real values of residual solvents because equilibrium is not attained in real drying processes 0.8 and the prediction of different interactions by the interaction parameter is too simplistic. The 0.6 real values are substantially higher and the real barriers of solvent 0.4 removal more complex. These are discussed below. In real systems, several phenomena take place. These include 0.2 chemical interaction between the functional groups of polymer and the solvent. These are mostly re0 0 2 4 6 8 10 lated to hydrogen bond formation. The crystalline structure of polyTime, min mer is responsible for the modifiFigure 16.1.1. Number of n-heptane molecules per one mer of poly- cation of the diffusion process. Solvent properties determine difstyrene vs. drying time. [Data from L A Errede, P J Henrich, fusion. Polymer properties are reJ N Schrolpfer, J. Appl. Polym. Sci., 54, 649 (1994).] sponsible for the macromechanism of solvent removal from a highly viscous polymer. In addition, some real systems make use of stripping solvents which are designed to help in removal of trace quantities of solvents by use of stripping solvent displacing process solvent. These various factors interplay and determine the result. Figure 16.1.1 shows the number of residual solvent molecules per one mer of polystyrene. It is evident that solvent removal has zero-kinetics until its concentration is decreased to about 0.2 molecules of solvent per mer. It is also true that some solvent remains after drying. Even after 24 h drying, 0.06% solvent remains. These data indicate that there is a different mechanism of removing residual solvent. It is not clear if this is because of interactions, a change in the glass transition temperature, or a change in crystallinity. So far the partial effects of these influences cannot be separated. It is confirmed by experiment that in the last stages of drying, glass transition temperature of polymer changes rapidly. Also, the degree of crystallinity of the polymer increases during drying.3,4 From studies on polyaniline, it is known that its conductivity depends on the concentration of adsorbed molecules of water.5 Water interacts by hydrogen bonding with the polymer chain. The activation energy of hydrogen bonding is very low at 3-5 kcal/mol. Drying at 120oC reduced the amount of water molecules from 0.75 to 0.3 molecules per aniline unit. This change in water concentration drastically alters electrical conductivity which decreases by three orders of magnitude. Drying for two hours at 120oC did not result in complete removal of water. Given that the activation energy of hydrogen bonding is very low, the process of interaction is probably not the main barrier to removal of residual moisture. Also the relationship between conductivity and number of molecules of water is linear in the range from 0.15 to 0.75 molecules of water per aniline unit which means that there is no drastic change in the mechanism by which water participates in increasing the conductivity of polyaniline. Its conductivity simply depends on the distance between neighboring adsorbed molecules of water which apparently participate in the charge migration.
n-heptane molecules/styrene unit
1
16.1 Residual solvents in various products
1127
Figure 16.1.2. Blister formation in polyethylene containing originally 4000 ppm hexane. a - lateral surface 2.2 s after extrusion, b - cross-section after 28 s. [Adapted, by permission from R J Albalak, Z Tadmor, Y Talmon, AIChE J., 36, 1313 (1990).]
SEM studies contribute to an understanding of a major obstacle to residual solvent removal. Figure 16.1.2 shows two photographs of polyethylene strands extruded from a melt which initially contained 4000 ppm hexane. After a short period of time following the extrusion, blisters form which remain in the material and become enlarged until they break and release solvent. This blistering mechanism, determines the rate of residual solvent removal from the material. The rate of removal depends on bubble nucleation, temperature, and polymer rheological properties.6 Observing such mechanisms makes it easy to understand the principle involved in stripping solvents which became popular in recent inventions.7-9 Stripping solvents were used to improve the taste and odor properties and the oxidative thermal stability of thermoplastic ethylene polymers.7 Volatile components, such as products of degradation, solvent and monomer contribute to taste problems and odor formation. Striping solvents used include highly volatile hydrocarbons (ethylene, propylene, isobutane), inert gases, and supercritical fluids. An addition of at least 0.1% stripping solvent reduces volatiles from the typical levels of between 300-950 ppm to 45 ppm with even as low as 10 ppm possible. A stripping solvent helps in the generation of bubbles and their subsequent breaking by which both the stripping and the residual solvent are removed. In cosmetics and pharmaceutical formulations traces of solvents such as benzene or dichloromethane, used in the synthesis of acrylic acid polymer, disqualify the material. It is not unusual for this polymer to contain up to 1000 ppm of dichloromethane or up to 100 ppm of benzene. The use of mixed ester solvents helps to reduce residual solvent to below 5 ppm.8 Polycarbonate pellets from normal production may contain up to 500 ppm solvent. This makes processing polycarbonate to optical products very difficult because of bubble formation. Elimination of volatiles renders the product suitable for optical grade articles.9 These inventions not only demonstrate how to eliminate solvents but also confirm that the mechanism discussed in Figure 16.1.2 operates in industrial processes. The examples also show that large quantities of residual solvents are retained by products in their normal synthesis. Many standard methods are used to devolatilize materials. Flash devolatilizer or falling strand devolatilizer are synonyms of equipment in which the falling melt is kept below the saturation pressure of volatiles. Styrene-acrylonitrile copolymers devolatilized in flash devolatilizer had a final concentration of ethylbenzene of 0.04-0.06.1 Devolatilization of LLDPE in a single-screw extruder leaves 100 ppm of hydrocarbon solvent. 500 ppm chlorobenzene remains in similarly extruded polycarbonate.1 It is estimated that if the polymer contains initially 1-2% solvent, 50-70% of that solvent will be removed through the vacuum port of an extruder.1 These data seem to corroborate the information included in the above
1128
George Wypych
discussed patents. The data show that considerable amounts of residual solvents can MEK be found in polymers and plastic materials. In the food industry, residual solvents associated with packaging odors enter food products from two sources: packing materiFat-free als and printing inks.10 It is estimated that 14% fat P concentrations of residual solvents have recently decreased (from 2000 mg/ream in past to 1000 mg/ream).10 However, a new problem has become apparent in introducPA tion of low fat or no fat food.11 It was discovered11 that more customer complaints about odor were received for these 0 2000 4000 6000 8000 1 104 low fat baked goods products. Analysis shows that fat was a good solvent for Peak area volatiles (solvents) and consequently solFigure 16.1.3. Volatility of three printing solvents (PA propyl acetate, P - n-propanol, MEK - methyl ethyl vent odor was not detected because flavor ketone) from fat-free and fat-containing cookies. 20 µg perception is developed relative to the consolvent added to 2.5 g cookies. [Data from T Clark, Paper centration of gaseous flavor compounds. Film Foil Converter, 70, 11, 48 (1996).] Figure 16.1.3 shows the concentrations of three solvents as detected by gas chromatography. Substantially higher concentrations are detected in solvent-spiked fat-free cookies than in fat-containing cookies.11 In this experiment solvents were added to the cookies. In another experiment, packaged cookies were exposed to a solvent vapor atmosphere and different trends were recorded for cookies packaged in two different films. If the film had good barrier properties, no difference was noticeable between both types of cookies and the adsorbed quantities of solvents were minimal. If the cookies were packaged in a coextruded film having lower barrier properties, no-fat cookies absorbed 42% more solvent than full fat cookies. In the construction industry, residual solvent evaporation becomes an increasingly more critical issue, especially in the case of products used for indoor applications. Sealants, adhesives, and paints are now a major focus of this concern since they contribute to indoor pollution. Similar trends are observed in the automotive industry where both solvents and plasticizers are suspected of contributing to a “plastics” odor in car interiors. In contrast, changes in the solvent evaporation rate may also contribute to product improvement in paints. Small quantities of properly selected solvents can improve physical properties and the appearance of paints. Other applications of residual solvents include time-controlled release of fertilizers and production of materials with controlled morphology.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11
R J Albalak, Ed., Polymer Devolatilization, Marcel Dekker, Inc., New York, 1996. L A Errede, P J Henrich, J N Schrolpfer, J. Appl. Polym. Sci., 54, 649 (1994). M O Ngui, S K Mallapragada, J. Polym. Sci., Polym. Phys., 36, 2771 (1998). H-T Kim, S-K Kim, J-K Park, Polym. J., 31, 154 (1999). E S Matveeva, Synthetic Metals, 79, 127 (1996). R J Albalak, Z Tadmor, Y Talmon, AIChE J., 36, 1313 (1990). M M Hughes, M E Rowland, C A Strait, US Patent 5,756,659, The Dow Chemical Company, 1998. A Bresciani, US Patent 5,342,911, 1994. T Hosomi, T Umemura, T Takata, Y Mori, US Patent 5,717,055, Mitsubishi Gas Chemical, Ltd., 1998. R M Podhajny, Paper Film Foil Converter, 72, 12, 24 (1998). T Clark, Paper Film Foil Converter, 70, 11, 48 (1996).
16.2 Residual solvents in pharmaceutical substances
1129
16.2 RESIDUAL SOLVENTS IN PHARMACEUTICAL SUBSTANCES Michel Bauer International Analytical Department, Sanofi-Synthélabo, Toulouse, France
Christine Barthélémy Laboratoire de Pharmacie Galénique et Biopharmacie, Faculté des Sciences Pharmaceutiques et Biologiques, Université de Lille 2, Lille, France
16.2.1 INTRODUCTION The need to test for residual solvents (RS) in pharmaceutical substances was recognized in the late 70’s when some pharmacopoeias like those of USA (USP XX) or Great-Britain (BP 80 + add 82) introduced specific tests for RS in some monographs. But we had to wait until the early 80’s to see a rational approach for establishing specifications from toxicological data. This strategy was developed by a working group of the Italian Pharmacopoeia1 starting from the threshold limit values for Chemical Substances and Physical Agents in the Work Environment published by the American Congress of Governmental Experts for Industrial hygiene.2 In the late 80’s, RS were definitively classified as impurities per se. Methods and specifications appeared in different issues of Pharmacopeial Forum and were submitted for discussion and finally integrated in the USP. At the same time only few such monographs could be found in the European Pharmacopoeia (Eur. Ph.) or British Pharmacopoeia (BP). Interestingly, the notion of a content limit for residual solvents in relation to the daily intake of the drug was introduced (see Eur. Ph. 2nd edition), a concept which was taken up later in the ICH Guideline. Numerous publications have been devoted to this subject.3-9 Although the number of papers on RS is immense, they are very often limited to specialized areas such as regulatory aspects or methodology aspects. In this chapter, different topics will be considered, starting with the fundamental question: why look for RS in pharmaceutical products? It is worth noting that Witschi and Doelker10 published in 1997 a very detailed and up-to-date review stressing the importance of this subject in the pharmaceutical field. 16.2.2 WHY SHOULD WE LOOK FOR RS? As we already have seen in Chapter 14.21.1, RS could have various effects on the drug substances, excipients and drug products. 16.2.2.1 Modifying the acceptability of the drug product The presence of RS could seriously impair customer compliance because of the odor or the taste they can cause in the final pharmaceutical preparation. Rabiant2 quotes the case of a drug substance having undergone, for technical reasons, a washing with isopropanol not planned in the manufacturing protocol. The oral solution prepared from this batch contained 100 ppm of this solvent and consequently had an odor that the majority of the patients accepted only with reluctance; the batch concerned was finally removed from the market.
1130
Michel Bauer, Christine Barthélémy
16.2.2.2 Modifying the physico-chemical properties of drug substances (DS) and drug products (DP) The role of the quality of the solvents on the stability e.g. of the raw materials, DS and DP (see Section 14.21.1) has already been discussed. It must be remembered that RS (including water) can show different kinds of interactions with solid substances:11 • Solvents adsorbed on the crystal surfaces, which generally are easily removed because of the existence of weak physical interactions. • Occluded solvents and clathrates which are more difficult to extract without impairing the quality of the drug by, for example, excessive drying. • Solvents bound to drug molecules in the crystal lattice. These solvents present as solvates (hydrates) are lost at a characteristic temperature and may be stable only over a limited range of relative humidity. The solvates and desolvated solvates, whilst being two different chemical entities, can retain the same crystalline structure (similarity of x-ray diffraction pattern) but show different physico-chemical properties.12-14 One of the most important effects of organic solvents absorbed on crystal surfaces is the ability to reduce the wettability of the crystals, especially if the solvent concerned is hydrophobic.15 Another interesting aspect, particularly in the case of residual water, is its role as an agent of recrystallization of poorly crystalline substances. It is well known that amorphous or partially amorphous products can undergo a recrystallization16,17 process over time in presence of water. If amorphous phases are an interesting way to promote the dissolution rate of poorly soluble drugs, their main drawback is their physical instability triggering the possible crystallization of the drug and leading to a decrease in dissolution rate and possibly of bioavailability, over a period of time. Furthermore the residual adsorbed water may have an impact on the flowability of a powder, which is linked to the solubility of the substance and the hydrophilicity of the crystal faces.15 Other physico-chemical parameters are influenced by RS, like particle size and dissolution properties. For more information we refer to the publications of Doelker10 and Guyot-Hermann15 and references quoted therein. Nevertheless, one example deserves to be mentioned here.10,16 Residual isopropyl alcohol enhanced the water permeability of Eudragit® L films used as tablet coating for protecting water-sensitive drugs as demonstrated by List and Laun.16 This implies particular conditions for the storage of coated tablets during the film-drying process. The atmosphere should be as dry as possible. The need to keep the RS level as low as possible can lead to some problems. It has been reported15,17 that a drug substance displaying a strong odor of residual solvent was submitted to reprocessing, consisting of the displacement of the residual solvent by a stream of water vapor. During this process which slightly modified the surface crystallinity of the particles, a small amount of an impurity was produced. The consequence was an increase of the surface solubility of this drug substance. During the preparation of the DP using an aqueous wet granulation, a liquification of the granulate was observed making the manufacture impossible. After having removed the impurity by purifying the DS a successful manufacture of the DP was achieved. There are other aspects linked to the manufacturing process and drying conditions which impact on the final RS content. They are discussed in Chapter 15.2.3 of this book (and references cited therein). Nevertheless before closing this paragraph, another example of the relationship between the manufacturing process and RS is worth mentioning here. It relates to the formation of volatile compounds produced by radiolysis and which could induce odor. Barbarin et
16.2 Residual solvents in pharmaceutical substances
1131
al.18 have investigated this subject in different antibiotics belonging to the cephalosporin group (Cefotaxime, Cefuroxime and Ceftazidine). Using GC-MS and GC-IR they were able to identify carbon monoxide, nitric oxide, carbon disulfide, methanol, acetaldehyde, ethyl formate, methyl acetate and acetaldehyde O-methyloxime. In a subsequent publication19 on cefotaxime, they demonstrated that some of the radio-induced compounds (such as carbon monoxide sulfide (COS) and carbon disulfide) came from the degradation of the drug itself whereas the formation of others required the RS, present before the irradiation. For instance, acetaldehyde arises from the irradiation of methanol. Incidentally, this is a good way to differentiate between radio-sterilized and non-radio-sterilized products. It is worth remarking that residual humidity can favor15 microbiological growth especially in some natural products used as excipients (starch, gelatine, etc.). From a physico-chemical point of view, residual humidity may have an impact on the hardness of the tablets as shown by Chowhan,20 Down and McMullen.21 16.2.2.3 Implications of possible drug/container interactions It is possible that the RS contained in a powder may migrate up to the interface between the contents and the container facilitating the extraction and migration into the drug of additives used during the container manufacture. On the contrary, solvents may be used during the packaging of a drug. An example of this situation is described by Letavernier et al.22 Cyclohexanone was used for sealing PVC blisters containing suppositories. After 42 months of storage at ambient temperature up to 0.2 mg to 0.3 mg of cyclohexanone was found per gram of suppository. 16.2.2.4 As a tool for forensic applications Forensic laboratories are interested in identifying and assessing trace impurities in bulk pharmaceutical products with the idea of using the impurity profile as a “fingerprint” of the manufacturer. Of course, RS can be an important aid in this process.23,24 It has been demonstrated,25 for instance, that static headspace GS coupled with mass spectrometry (MS) was able to detect and identify volatile impurities, making possible the characterization of illicit heroin or cocaine samples. 16.2.2.5 As a source of toxicity 16.2.2.5.1 General points Toxicity is obviously the main reasons for testing for RS. Besides the toxicity of the drug itself, the related impurities,26 degradants and the RS obviously can each bring their own contribution. A drug could be prescribed to a patient either for a short period of time or a long one. To minimize either acute toxicity or chronic toxicity resulting from some accumulation process, RS have to be kept at the lowest achievable level. When developing new chemical entities, the presence of RS could bias toxicity studies including mutagenicity and carcinogenicity and cause a risk of wrongly ascribing to the drug substance or the formulation, side effects which are actually due to volatile impurities. Knowing the cost of such studies, it is preferable to have the RS under control. The long-term exposure to solvents has been recognized for a long time as a possible cause of serious adverse effects in human. Tables of maximal tolerated solvent concentrations in air for defined exposures have been published and used to set limits for residual organic solvents in pharmaceuticals.1-10 In the late 90’s, a set of articles appeared in the Pharmacopeial Forum proposing RS limits from toxicological data,27 including carcinogenicity, mutagenicity, teratogenicity and neurotoxicity.
1132
Michel Bauer, Christine Barthélémy
Table 16.2.1. (After from reference 28)
In the early 80’s, the “International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human use” (ICH) was created. Among the different topics deserving to be harmonized, the need to have an agreement amongst Europe, Japan and the USA on the ways to limit the RS in pharmaceuticals was clearly identified and the topic adopted in June 1994. The final guide was finally adopted in 1998 by the Health Authorities of the three zones and is now in force. This Guideline will be examined in detail in the last paragraph of this part. 16.2.2.5.2 Brief overview of the toxicology of solvents28,29 16.2.2.5.2.1 Aspects concerning metabolism There are four main routes by which solvents can interact with the human body: • O.R.L. (Otorhinolaryngology) • transdermal (including the ocular area) • oral • injection Once in the organism, solvents will undergo biotransformation which essentially takes place in the liver. This metabolism very often leads to more water-soluble products than the parent compound30,31 and, as such, more easily excreted by the kidneys. This detoxication process is beneficial for the individual but varies greatly from one subject to another. Unfortunately, this metabolic detoxication can be complicated by the appearance of reactive intermediates which, if not rapidly inactivated, will destroy the essential constituents of the cells (proteins, nucleic acids, unsaturated lipids) and cause INTOXICATION. The latter ranges from a simple allergic reaction to tissue necrosis or, at worst, to cancer. Table 16.2.1 summarizes the different events which can occur in the organism. 16.2.2.5.2.2 Solvent-related pathology 16.2.2.5.2.2.1 Acute toxicity It is especially the affinity of the solvents for lipid-rich organs which triggers problems of acute toxicity and these concern primarily the nervous system, the heart, the liver and kidneys. In this acute toxicological process, the molecules act per se without any previous biotransformation. The acute toxicity encompasses:
16.2 Residual solvents in pharmaceutical substances
1133
• nervous toxicity (headache, somnolence, coma, more or less deep, which can extend to death)
• cardiac toxicity32 • action on skin and mucous membranes generating irritation (including ocular area)33 16.2.2.5.2.2.2 Long-term toxicity A prolonged exposure, even at low doses, to several liposoluble solvents leads sooner or later to irreversible effects on different organs: • central nervous system (e.g., toluene could lead to degeneration of the brain) • peripheral nervous system (e.g., methanol shows a peculiar affinity for the optic nerve leading possibly to blindness) • liver and kidneys; as solvents are metabolized in the liver and excreted by kidneys, these two organs are, of course, particular targets for these products • skin and mucous membranes (e.g., dermatitis) • blood (cyanosis, anaemia, chromosomal abnormality) With regard to carcinogenicity, benzene has long been recognized as carcinogen in man. It is the reason why its use is strictly limited and not recommended (ICH class I/specifications 2 ppm). Carbon tetrachloride and 1,2-dichloroethane have been demonstrated to be carcinogenic in animals and potentially carcinogenic in man. The embryotoxicity of solvents must be taken into account. As solvents can cross the placenta, pregnant women should be especially protected. 16.2.2.5.2.2.3 Metabolism of benzene By way of illustration, Figure 16.2.1 summarizes the metabolic pathway of benzene.34 Numerous publications dealing with the metabolism of benzene are given in reference 27.2. 16.2.3 HOW TO IDENTIFY AND CONTROL RS IN PHARMACEUTICAL SUBSTANCES? 16.2.3.1 Loss of weight Historically this is the first method which appeared in the pharmacopoeias, performed either at normal pressure or under vacuum. This is, of course, an easy method, particularly for routine control but there are several drawbacks: • lack of specificity • it is product demanding (1-2 g) • the limit of detection (LOD) is currently about 0.1 % This determination can now be done by thermogravimetric analysis (TGA) which makes the method more sensitive (possible LOD 100 ppm) and less product demanding (5 20 mg). It can also be used as a hyphenated method linking TGA to a mass spectrometer, allowing the identification of the desorbed solvents (specificity). However, whilst this kind of equipment exists, there are no signs of it replacing gas chromatography (GC) in the near future. 16.2.3.2 Miscellaneous methods Infrared spectroscopy35 and 1H-NMR36 have been used occasionally to identify and to quantify residual solvents, but their sensitivity is rather limited if compared with GC. On the other hand specificity is not always assured. The solvent should display signals well separated from those arising from the product, which is not always the case.
1134
Michel Bauer, Christine Barthélémy
Figure 16.2.1. Overview of benzene metabolism (after reference 34).
16.2.3.3 Gas chromatography (GC) 16.2.3.3.1 General points This is, of course, the method of choice which has long been used to determine RS whatever the area of application (pharmaceuticals, polymers, water analysis, etc.).37-40 From the late 70’s to the beginning of the 80’s, there was a large number (or flood) of publications dealing with different possible GC techniques which could be applied for detecting or analyzing residual solvents especially in pharmaceuticals (intermediates of synthesis, drug sub-
16.2 Residual solvents in pharmaceutical substances
1135
stances, excipients and drug products). So much has been written that it is difficult to be original. Methodological aspects will be briefly covered in the next paragraph. The paper of Witschi and Doelker10 is particularly recommended. With 171 references, it represents a worthwhile, up-to-date review of the different GC techniques available. The last part of this chapter concerns pharmacopoeial methods. 16.2.3.3.2 Review of methods Regarding GC methodology, four aspects must be examined: • injection systems • columns • detectors • method validation 16.2.3.3.2.1 Injection systems 16.2.3.3.2.1.1 Direct injection After having dissolved the substance containing the RS to be looked for in an appropriate solvent, it is possible to directly inject into the system 100% of an aliquot of the solution (if packed columns are used) or partially through a split system (if capillary, narrow-bore and wide-bore columns are used). It is simple, accurate and repeatable (with an internal standard). The main drawback is that samples very often contain non-volatile substances which are retained by the column, leading rapidly to a loss of efficiency and a dramatic decrease in sensitivity.41 In the current literature dealing with the RS, the direct injection process is less frequently used. Nevertheless, publications have appeared until recently using split/ splitless injection.42,43 16.2.3.3.2.1.2 Static headspace injector The solubilized or suspended sample in an appropriate vehicle is heated at a defined temperature in a tightly closed vial until thermodynamic equilibrium is reached between the liquid phase and the gas phase. A known aliquot is then transferred either with the aid of a syringe or by an automatic transfer system onto the column. The main advantage is that only volatile products including solvents are injected into the column greatly improving its lifetime. The sensitivity is good and the system is easily automated. The main drawback is the existence of matrix effects and the possible non- ideality of the solvents mixture. These imply that ideally one should determine the calibration curve by adding standard solutions of the solvents of interest to the sample matrix, free of solvents. Because it is very difficult to obtain such a sample matrix, the classical standard addition method is recommended. It consists of adding to the sample matrix to be analyzed a known amount of the solvents to be determined. This method requires two analyses for the final calculation but the main advantage is that the matrix effect is overcome. The linearity of the response has, of course, to be demonstrated before the use of the simplified version mentioned above. Nevertheless, if based on a sound validation, external calibration can be used.41,44 The nature of the solvent used to prepare the solution (e.g., water, dimethylformamide, dimethylacetamide, 1,3-dimethyl-2-imidazolidinone (DMI)), equilibration temperature, the ratio between the gas phase and the liquid phase and the possible need to promote a salting out effect by adding mineral salts are the parameters, amongst others, which should be investigated and optimized to improve the sensitivity (LOD, LOQ) of the method.45-47 Another version of this static headspace chromatography is what has been called by Kolb48 multiple headspace extraction (MHE) chromatography. This is a multi-step injection
1136
Michel Bauer, Christine Barthélémy
technique which was alluded to in the Suzuki publication39 and more openly developed by MacAuliffe.49 The principle of this method is the following.50,51 After the first extraction has been made and the aliquot injected, the gas phase is removed by ventilating the vial and re-establishing the thermodynamic equilibrium. The equilibrium between the analyte in the solid or liquid phase and the gas phase will be displaced each time. After n extractions, the analyte content in the liquid or solid phase becomes negligible. It is then sufficient to sum the peak areas obtained for each extraction (which decrease exponentially) and, from an external calibration curve determine the amount of RS in the substance. This method is particularly useful for insoluble products or in cases where the partition coefficient of the RS is too favorable relative to the liquid phase. It has been recently successfully applied to the determination of RS in transdermal drug delivery systems.51 16.2.3.3.2.1.3 Dynamic headspace injection In the dynamic headspace method, the sample is put in a thermal desorption unit in order to desorb the RS; a continuous flow of a carrier gas pushes the RS into a trapping system which is refrigerated and where they are accumulated prior to analysis. Then the RS are rapidly desorbed by rapid heating and carried onto the column via the carrier gas. There are different ways to apply this technique.10,52 The arrangement when purge gas passes through the sample is often called the purge and trap technique (some other equipment uses the acronym DCI (desorption, concentration, injection)). This method is particularly useful for very low concentrations of RS as the total amount of a substance is extracted and can be applied directly to powders without need to dissolve them. The main drawback is that the dynamic headspace methods are not readily automated.41 16.2.3.3.2.1.4 Other techniques Several others techniques dealing with the injection problems have been developed. Among them the solid-phase microextraction method52-55 (SPME) and the full evaporation technique56 must be mentioned. According to Camarasu,53 the SPME technique seems to be very promising for RS determination in pharmaceuticals, with much better sensitivity than the static headspace technique. 16.2.3.3.2.2 Columns The wealth of publications dealing with RS determination by GC is so impressive that it is difficult to provide an exhaustive review. The interested reader will find plenty of information in the references quoted so far and in others recently published, mentioned below. Packed columns,57 wide or narrow bore columns, capillary columns, etc. have been used for RS determination. It is true to say that capillary columns and narrow bore columns are the most often mentioned techniques. Today it is almost certain that any user can find in the literature10 the stationary phase and the relevant conditions to resolve his RS problem, at least in terms of selectivity. By way of example, it has been shown by Brinkmann and Ebel58,59 how it was possible to screen 65 of the 69 solvents mentioned in the ICH Guideline (discussed in paragraph 16.2.4) using capillary columns filled with two stationary phases (DB 624 and Stabilwax) which basically constitute the strategy proposed by the European Pharmacopoeia. 16.2.3.3.2.3 Detectors The almost universally used detector is the flame ionization detector (FID) which works with all organic solvents but which is not selective. The mass spectrometer detector (MSD) is now more and more utilized.24,53,55,60,61 It can be either universal in its electron ionization (EI) mode or selective in its selective ion monitoring mode (SIM). Other detectors, selective
16.2 Residual solvents in pharmaceutical substances
1137
and/or universal, can be used.10,62 Among them the electron capture detector has to be mentioned when looking for chlorinated solvents, even if its use is not straightforward. 16.2.3.3.2.4 Method validation Whatever the technique used for determining the RS content in pharmaceuticals, a thorough validation of the complete analytical process has to be conducted according to the ICH Guidelines [Text on “validation of analytical procedures” and “validation of analytical procedures: methodology”63]. For testing impurities in a quantitative manner the following items have to be completed: • specificity (or more appropriately selectivity) • accuracy • precision • repeatability • intermediate precision (first part of reproducibility) • limit of detection (LOD), limit of quantification (LOQ) • linearity of the response • the range which is the interval between the upper and lower concentration (amounts) of analyte in the sample (including these concentrations) for which it has been demonstrated that the analytical procedure has a suitable level of precision, accuracy and linearity. In the publications mentioned above,24,41-43,51-54,58,59,64 it is possible for reader to find experimental procedures to conduct validation efficiently. Attention is drawn to the fact that, when using the static headspace technique in particular some other parameters have to be investigated, such as:24,51 • ratio gas phase/solid or liquid phase • temperature of equilibrium between the two phases and time to reach it • temperature of the transfer line • pressurization time and sampling time when using a fully automated headspace injector Finally, some additional comments are worth making: • As regards reproducibility, it is true that the best way to assess it is to set up an inter-laboratory study. It is of course burdensome but this is probably the only way to infer reasonable suitability parameters for the routine quality control monograph (resolution, plate number, tailing factor, repeatability, LOD, LOQ) and specifications. It is worth noting that the latter should take the performance of the analytical method into account. But for those who cannot follow this approach, sound ruggedness testing has to be performed. Maris et al. have designed a method65 for evaluating the ruggedness of a gas chromatographic method for residual solvents in pharmaceutical substances. • When using the static headspace injector the possible matrix effect should be studied and can be evaluated by a statistical method.66,67 • One of the most important suitability parameters in case of RS determination is, of course, the LOQ (and LOD). In order to avoid an unrealistic value in the QC monograph, it is highly recommended to use a working limit of quantification WLOQ (and WLOD) which consists of determining a reasonable upper limit for LOQ (and LOD). In fact, the LOQ derived from the validation package is obtained in what we can call an ideal or optimized environment. A way for determining α (> 1) in the relationship:
1138
Michel Bauer, Christine Barthélémy
Figure 16.2.2. Head-space injection of the gaseous sample into the chromatographic system. Typical chromatogram of class 1 solvents using the conditions described for System A (European Pharmacopoeia method). Flame-ionization detector: calculation of H/h for 1,1,1-trichloroethane.
WLOQ = α LOQ is to calculate the LOQ according to the signal to noise ratio S/N method63,68 (see Figure 16.2.2) and to repeat this determination independently several times (e.g., n = 6) during the intermediate precision determination. If LOQm is the mean value and σ the standard deviation, a possible definition of WLOQ could be the upper limit of the one side confidence limit at a specified risk α (e.g., 0.05). WLOQ = LOQ m + t n −1, α σ where t is Student’s coefficient.
16.2 Residual solvents in pharmaceutical substances
1139
16.2.3.3.3 Official GC methods for RS determination Current official GC methods are described in USP XXIII under chapter 467 “Organic volatile impurities”. Four methods (I, IV, V, VI) are mentioned. Methods I, V and VI are based on direct injection. They are suitable for water-soluble drugs and V for water insoluble drugs. Method IV describes the static headspace technique and is used for water soluble drugs. Method VI is very general and refers to the individual monograph which describes the chromatographic conditions ( injection, column, conditions) which should be used. The main characteristics of these four methods are summarized in Table 16.2.2. The European Pharmacopoeia69 used a two-tiered process based on two different columns: System A. Fused silica capillary or semi capillary column (30 m x 0.32 mm (ID) or 30 m x 0.53 mm (ID)) DB 264 (1.8 µm or 3 µm film thickness of the phase (6 per cent cyanopropylphenyl-94 percent dimethylpolysiloxane)) which is identical to the USP method V. System B. Fused silica capillary or semi-capillary column (30 m x 0.32 mm (ID) or 30 m x 0.53 mm (ID) DB-wax (0.25 µm film thickness of Macrogol 20000R)). Table 16.2.2. Gas chromatographic methods described in USP 23 (from ref. 10) Method
USP <467> Method I: Direct GC injection
USP <467> Method IV: Static HSC
USP <467> Method V: Direct GC injection
Sample
Standardization
Column
Detector
Dissolved in water or another appropriate solvent
External
30 m x 0.53 mm ID, fused silica, with 5µm a crosslinked G27 stationary phase and A 5 m x 0.53 mm ID silica guard column, phenylmethyl siloxane deactivated
Dissolved in water containing sodium sulphate and heated for 1 h at 80°C before injection of the headspace
External
As USP <467> Method V
FID
As in USP <467> Method I
30 m x 0.53 mm ID, fused silica with 3 µm b G43 stationary phase and a 5 m x 0.5 mm ID silica guard column, phenylmethyl siloxane deactivated
FID
As in USP <467> Method I
One of 9 columns , listed under <467>, specified in the monograph
As USP Method I
<467>
FID*
c
USP <467 Method VI: Direct GC injection **
As USP Method I
<467>
FID
1140
Michel Bauer, Christine Barthélémy
Method
Sample
USP <467> Method for dichlormethane in coated tablets
The tablet water extract is heated for 20 min at 85°C before headspace injection
Standardization
Standard addition
Column
As USP Method V
<467>
Detector
FID
*To confirm the identity of a peak in the chromatogram, a mass spectrometer can be used or a second validated column, containing a different stationary phase. **Method VI presents a collective of chromatographic systems. a 5% phenyl/95% methylpolysiloxane; b6% cyanopropylphenyl/94% dimethylpolysiloxane; cS2, styrene-divinylbenzene copolymer; S3, copolymer of ethylvinylbenzene and divinylbenzene; S4, styrene-divinylbenzene; G14, polyethyleneglycol (Mw 950-1050); G16, polyethyleneglycol compound (polyethyleneglycol compound 20M or carbowax 20 %; G27, see a; G39, polyethyleneglycol (Mw 1500).
The static headspace injector has been selected and the method developed for water soluble products using water as dissolution medium or using N,N-dimethylformamide for water-insoluble products. If N,N dimethylacetamide and/or N,N-dimethylformamide are suspected in the drug under investigation 1,3-dimethyl 2-imidazolinone (DMI) is used as dissolving medium. The method has been designed: • in order to identify most class 1 and class 2 RS potentially present in drug substances, excipients or drug products. • as a limit test for class 1 and class 2 RS present in drug substances, excipients and drug products. • to quantify class 2 RS where the content is higher than 1000 ppm or class 3 RS if the need arises. Figures 16.2.3, 16.2.4, 16.2.5, and 16.2.6 illustrate the separations obtained in the two systems for solvents belonging to class 1 and class 2 solvents. This general method is the outcome of European working party which has been published in Pharmeuropa.70 It should be stressed that the use of a general pharmacopeial method is not a reason not to validate the latter when analyzing a particular substance. The matrix effect, in particular, has to be investigated when using the static headspace mode of injection. 16.2.4 HOW TO SET SPECIFICATIONS? EXAMINATION OF THE ICH GUIDELINES FOR RESIDUAL SOLVENTS The introduction briefly summarizes the strategies dealing with the setting of specifications in pharmaceuticals which appeared during the 80’s. From the early 90’s onwards, the International Conference on Harmonization was initiated in three important pharmaceutical regions (Europe, Japan, USA) in order to define a common way of preparing a registration file acceptable in the three zones. The topics included in this harmonization process are: • Quality • Safety • Efficacy Regarding Quality a set of Guidelines have been already adopted. Three of them are particularly relevant with regard to this article: • Guideline Q3A: Impurities in new active substances • Guideline Q3B: Impurities in new medicinal products • Guideline Q3C: Note for guidance on impurities: residual solvents
16.2 Residual solvents in pharmaceutical substances
1141
Figure 16.2.3. Typical chromatogram of class 1 solvents using the conditions described for system A and procedure 1. Flame-ionization detector. 4: benzene; 10: carbon tetrachloride; 14: 1,2-dichloroethane; 15: 1,1-dichloroethylene; 52: 1,1,1-trichloroethane. [Adapted, by permission, from European Pharmacopoeia, Addendum 2000, pp31-36.] [Please note that information concerning residual solvents are susceptible to be modified in the successive editions of the European Pharmacopeia.]
Figure 16.2.4. Chromatogram of class 2 solvents using the conditions described for system A and procedure 1. Flame-ionization detector. 3: acetonitrile; 11: chloroform; 13: cyclohexane; 16a: cis-1,2dichloroethylene; 17: dichloromethane; 29: hexane; 30: 2-hexanone; 34: methanol; 49: pyridine; 51: toluene; 53: 1,1,2-trichloroethylene; 54: xylene ortho, meta, para; 58: chlorobenzene; 61: tetraline; 62: methylcyclohexane: 63; nitromethane; 64: 1,2-dimethoxyethane. [Adapted, by permission, from European Pharmacopoeia, Addendum 2000, pp31-36.] [Please note that information concerning residual solvents are susceptible to be modified in the successive editions of the European Pharmacopeia.]
1142
Michel Bauer, Christine Barthélémy
Figure 16.2.5. Chromatogram of class 1 residual solvents using the conditions described for system B and procedure 1. Flame-ionization detector. 4: benzene; 10: carbon tetrachloride; 14: 1,2-dichloroethane; 15: 1,1-dichloroethylene; 52: 1,1,1-trichloroethane. [Adapted, by permission, from European Pharmacopoeia, Addendum 2000, pp31-36.] [Please note that information concerning residual solvents are susceptible to be modified in the successive editions of the European Pharmacopeia.]
Figure 16.2.6. Typical chromatogram of class 2 residual solvents using the conditions described for system B and procedure 1. Flame ionization detector. 3: acetonitrile; 11: chloroform; 13: cyclohexane; 16a: cis-1,2-dichloroethylene; 17: dichloromethane; 23: 1,4-dioxane; 29: hexane ; 30: 2-hexanone; 34: methanol; 49 : pyridine ; 51 : toluene ; 53 : 1,1,2-trichloroethylene ; 54 : xylene, ortho, meta, para; 58: chlorobenzene; 61: tetralin; 62: methylcyclohexane; 63: nitromethane; 64: 1,2-dimethoxyethane. [Adapted, by permission, from European Pharmacopoeia, Addendum 2000, pp31-36.] [Please note that information concerning residual solvents are susceptible to be modified in the successive editions of the European Pharmacopeia.]
16.2 Residual solvents in pharmaceutical substances
1143
Comments on the latter are given below. The full text can be found in the US Pharmacopoeia, in the European Pharmacopoeia or in Journals71 (there is also a website www.ifpma.org/ich1.htm). 16.2.4.1 Introduction The Guideline recommends acceptable amounts of RS in pharmaceuticals which are safe for the patient. Residual solvents in pharmaceuticals are defined as organic volatile chemicals that are used or produced in the manufacture of active substances or excipients, or in the preparation of medicinal products. It is stated that medicinal products should contain no higher levels of residual solvents than can be supported by safety data. Three classes of solvents have been defined based on risk assessment. 16.2.4.2 Classification of residual solvents by risk assessment Class 1 solvents: solvents to be avoided, known as human carcinogens or strongly suspected carcinogens and environmental hazards. Class 2 solvents: solvents to be limited. Nongenotoxic animal carcinogens or possible causative agents of other irreversible toxicities such as neurotoxicity or teratogenicity. Solvents suspected of other significant but reversible toxicities. Class 3 solvents: Solvents with low toxic potential to man: no health-based exposure limit is needed. Class 3 solvents have permitted daily exposures (PDE) of 50 mg or more per day. 16.2.4.3 Definition of PDE. Method for establishing exposure limits The PDE is defined as a pharmaceutically acceptable intake of RS. The method used to establish PDEs for RS is described in the references cited above.27,71 16.2.4.4 Limits for residual solvents Solvents of class 1 (see Table 16.2.3. Class 1 solvents in pharmaceutical Table 16.2.3) should not be products (solvents that should be avoided) employed. However, if their use is unavoidable in order to Concentration produce a significant theraSolvent Concern limit, ppm peutic advance, then their levels should be restricted as Benzene 2 Carcinogen shown in Table 16.2.3, unless Toxic and environmenotherwise justified. Carbon tetrachloride 4 tal hazard Solvents of class 2 (see Table 16.2.4). Two options 1.2-Dichloroethane 5 Toxic are available when setting 1.1-Dichloroethene 8 Toxic limits for class 2 solvents. Option 1 1.1.1-Trichloroethane 1500 Environmental hazard The concentration limits in ppm stated in Table 16.2.4 can be used. They were calculated using the equation [16.2.1] by assuming a product mass of 10 g is administered daily. concentration (ppm) =
1000 × PDE dose
where PDE is given in mg/day and dose is given in g/day.
1144
Michel Bauer, Christine Barthélémy
Table 16.2.4. Class 2 solvents in pharmaceutical products Solvent
PDE, mg/day
Concentration limit, ppm
Acetonitrile
4.1
410
Chlorobenzene
3.6
360
Chloroform
0.6
60
Cyclohexane
38.8
3880
1,2-Dichloroethene
18.7
1870
Dichloromethane
6.0
600
1,2-Dimethoxyethane
1.0
100
N,N-Dimethylacetamide
10.9
1090
N,N-Dimethylformamide
8.8
880
1,4-Dioxane
3.8
380
2-Ethoxyethanol
1.6
160
Ethylene glycol
6.2
620
Formamide
2.2
220
Hexane
2.9
290
Methanol
30.0
3000
2-Methoxyethanol
0.5
50
Methylbutylketone
0.5
50
Methylcyclohexane
11.8
1180
N-Methylpyrrolidone
48.4
4840
Nitromethane
0.5
50
Pyridine
2.0
200
Sulfolane
1.6
160
Tetralin
1.0
100
Toluene
8.9
890
1,1,2-Trichloroethene
0.8
80
Xylene
21.7
2170
Option 2 It is not considered necessary for each component of the medicinal product to comply with the limits given in option 1. The PDE in terms of mg/day as indicated in Table 16.2.4 can be used with the known maximum daily dose and equation [16.2.1] to determine the concentration of residual solvent allowed in the medicinal product. Option 2 may be applied by adding the amounts of RS present in each of the components of the pharmaceutical formulation. The sum of the amounts of solvent per day should be less than that given by the PDE.
16.2 Residual solvents in pharmaceutical substances
Table 16.2.5. Class 3 solvents which should be limited by GMP or other quality-based requirements
1145
Table 16.2.6. Solvents for which no adequate toxicological data was found
Acetic acid
Heptane
1,1-Diethoxypropane
Acetone
Isobutyl acetate
1,1-Dimethoxymethane
Anisole
Isopropyl acetate
2,2-Dimethoxypropane
1-Butanol
Methyl acetate
Isooctane
2-Butanol
3-Methyl-1-butanol
Isopropyl ether
Butyl acetate
Methyl ethyl ketone
Methylisopropyl ketone
tert-Butyl methyl ether
Methyl isobutyl ketone
Methyltetrahydrofuran
Cumene
2-Methyl-1-propanol
Petroleum ether
Dimethylsulfoxide
Pentane
Trichloroacetic acid
Ethanol
1-Pentanol
Trifluoroacetic acid
Ethyl acetate
1-Propanol
Solvents with low toxic potential solvents of class 3 (see Table 16.2.5) may be Ethyl formate Propyl acetate regarded as less toxic and of lower risk to Formic acid Tetrahydrofuran human health. It is considered that amounts of these RS of 50 mg per day or less (corresponding to 5000 ppm or 0.5 % under option 1) would be acceptable without justification. Higher amounts may be acceptable provided they do not have a negative impact on the processability and the stability of the pharmaceutical product. Solvents for which no adequate toxicological data was found (Table 16.2.6). These solvents can be used in the manufacture of drug substances, excipients and drug products, but the manufacturer should supply justification for residual levels of these solvents in pharmaceutical products. 16.2.4.5 Analytical procedures If only class 3 solvents are present, a non-specific method such as loss of drying may be used. In the other cases a selective method (e.g., GC) is required. Especially if solvents of class 2 and class 3 are present at greater than their option 1 limits or 0.5 %, respectively, they should be identified and quantified. 16.2.4.6 Conclusions regarding the ICH Guideline The lists are not exhaustive and other solvents can be used and added later to the lists. Recommended limits of class 1 and class 2 solvents or classification of solvents may change as new safety data become available. Nevertheless this Guideline is of great interest for those involved in the pharmaceutical development in order to prepare successfully a pharmaceutical dossier acceptable everywhere in the world. Ethyl ether
2-Propanol
1146
Michel Bauer, Christine Barthélémy
16.2.5 CONCLUSIONS Those who have worked for many years in pharmaceuticals will have observed continuous progress in standards of Quality and Control in manufacturing. Solvents including water are still used in almost every step of the elaboration of a drug product. Their residues could be detrimental for the processability and stability of the pharmaceutical products and the safety of patients. At the end of this millennium it can be said that the testing and control of RS has been thoroughly assessed and is based on robust and sensitive techniques, for which the limitations are known, resulting in a sound strategy accepted almost everywhere in the world. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27.1 27.2 27.3 27.4 27.5 28 29 30 31 32 33 34 35 36 37 38
Societa Italiana di Scienze Farmaceutiche, Cronache Farmaceutiche, 6, 227-229 (1980). Treshold limit values for chemical substances and physical agents in the work environment AGGIH 6500 GLENWAY. J. Rabiant, Ann. Pharm. Fr., 42, 503 (1984). J. Rabiant, S.T.P. Pharma Pratiques, 1, 278 (1991). M. Gachon, S.T.P. Pharma Pratiques, 1, 531 (1991). Procedures for setting limits for organic volatile solvents with methylene chloride as an example of the process, Pharm For., 15, 5748 (1989). Survey: residual solvents, Pharmeuropa, 2, 142 (1990). H. Stumpf, E. Spiess and M. Habs, Dtsch. Apoth. Ztg., 132, 508 (1992). A. Eichhorn, T. Gabrio and S. Plagge, Zent. Bl. Pharm. Pharmakother. Lab. Diagn., 128, 675 (1989). C. Witschi and E. Doelker, Eur. J. Pharm. Biopharm., 43, 215 (1997). E. Ettabia, C. Barthelemy, M. Jbilou and A.M. Guyot-Hermann, Pharmazie, 53, 8 (1998). G.A. Stephenson, E.G. Groleau, R.L. Kleemann, W. Xu and D.R. Rigsbee, J. Pharm. Sci., 87, 536 (1998). S.R. Byrn, C.T. Lin, P. Perrier, G.G. Clay and P.A. Sutton, J. Org. Chem., 47, 2978 (1982). J. Bauer, J. Quick, R. Oheim, J. Pharm. Sci., 74, 899 (1985). A.M. Guyot-Hermann, S.T.P. Pharma Pratiques, 1, 258 (1991). P.H. List and G. Laun, Pharm. Ind., 42, 399 (1980). C. Lefebvre Ringard, A.M. Guyot-Hermann, R. Bouche et J. Ringard, S.T.P. Pharma, 6, 228 (1990). N. Barbarin, A.S. Crucq and B. Tilquin, Radiat. Phys. Chem., 48, 787 (1996). N. Barbarin, B. Rollmann, B. Tilquin, Int. J. Pharm., 178, 203 (1999). Z.T. Chowhan, Drug Dev. Ind. Pharm, 5, 41 (1979). G.R.B. Down and J.N. McMullen, Powder Technol., 42, 169 (1985). J.F. Letavernier, M. Aubert, G. Ripoche et F. Pellerin, Ann. Pharm. Franc., 43, 117, (1985). K.J. Mulligan, T.W. Brueggemeyer, D.F. Crockett, J.B. Shepman, J. Chrom., 686, 85 (1996). K.J. Mulligan and H. McCaulay, J. Chrom. Sci., 33, 49 (1995). D.R. Morello and R.P. Meyers, J. Forensic Sci, JFSCA, 40(6), 957 (1995). A.C. Cartwright, Int. Pharm. J., 4, 146 (1990). Procedure for setting limits for organic volatile solvents with methylene chloride as an example of the process, Pharm. For., 15, 5748 (1989). D. Galer, R.H. Ku, C.S. Schwartz, Pharm. For., 17, 1443 (1991). L. Brooks, J.S. Mehring, Pharm. For., 16, 550 (1990). B.D. Naumann, E.V. Sargent, Pharm. For., 16, 573 (1990). G.L. Sprague, S. Beecham-K., D.L. Conine, Pharm. For., 16, 543 (1990). A. Picot, Information Toxicologique n° 3, Unité de prévention du risque chimique - CNRS (France) (1995). J. Belegaud, Communication personnelle (1997). H.H. Cornish, B.P. Ling and M.L. Barth, Amer. Ind. Hyg. Ass. J., 34, 487 (1973). K. Morpoth, U. Witting and M. Springorum, Int. Arch. Arveits Med., 33, 315 (1974). J.W. Hayden and E.G. Comstock, Clinical Toxicology, 9, 164 (1976). M.J. Archieri, H. Janiaut, A. Picot, L’Actualité Chimique, Mai-Juin, 241 (1992). Le Goff - Les risques physicochimiques et toxiques des solvants - Communication personnelle M.G. Vachon, J.G. Nain, J. Microencapsul., 12, 287 (1995). H.W. Ardovich, M.J. Lebelle, C. Savard, W.L. Wilson, Forensic. Sci. Int., 49, 225 (1991). J. Haslam and A.R. Jeffs, Analyst, 83, 455 (1958). J.A. Hudy, J. Gas Chromatogr., 4, 350 (1966).
16.2 Residual solvents in pharmaceutical substances
39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
1147
M. Suzuki, S. Tsuge and T. Takeuchi, Anal. Chem., 42, 1705 (1970). K. Grob, J. Chrom., 84, 255 (1973). T.K. Natishan and Y. Wu, J. Chromatogr., A800, 275 (1998). Q. Chanli, K.A. Cohen and G. Zhuang, J. Chrom. Sci., 36, 119 (1998). T.K. Chen, J.G. Phillips and W. Durr, J. Chromatogr., A811, 145 (1998). J.P. Guimbard, M. Person and J.P. Vergnaud, J. Chromatogr., 403, 109 (1987). Progress report of the working party on residual solvents (technical) of the European Pharmacopoeia Commission, Pharmeuropa, 8, 586 (1996). V.J. Naughton, Pharm. Forum, 20, 7223 (1994). M. de Smet, K. Roels, L. Vanhoof and W. Lauwers, Pharm. Forum, 21, 501 (1995). B. Kolb, Chromatographia, 10, 587 (1982). C. MacAullife, Chem. Tech., 46, 51 (1971). J.P. Guimbard, J. Besson, S. Beaufort, J. Pittie et M. Gachon, S.T.P. Pharma Pratiques, 1(3), 272 (1991). P. Klaffenbach, C. Brüse, C. Coors, D. Kronenfeld and H.G. Schulz, LC-GC, 15, 1052 (1997). P. Kuran and L. Sojak, J. Chromatogr., A773, 119 (1996). C.C. Camarasu, M. Mezei-Szüts, G. Bertok Varga, J. Pharm. Biomed. Anal., 18, 623 (1998). R.J. Bartelt, Anal. Chem., 69, 364 (1997). F.J. Santos, M.T. Galceram and D. Fraisse, J. Chromatogr., 742, 181 (1996). M. Markelov and J.P. Guzowski, Anal. Chim. Acta, 276, 235 (1993). G. Castello, S. Vezzani and T.C. Gerbino, J. Chromatogr., 585, 273 (1991). K. Brinkmann and S. Ebel, Pharm. Ind., 61, 263 (1999). K. Brinkmann and S. Ebel, Pharm. Ind., 61, 372 (1999). W.D. Bowers, M.L. Parsons, R.E. Clement, G.A. Eiceman and F.W. Karaseck, J. Chromatogr., 206, 279 (1981). C.N. Kunigami, M.S. Sanctos, M. Helena, W. Morelli-Cardoro, J. High Resol. Chromatogr., 22, 477 (1999). D.G. Westmorland, G.R. Rhodes, Pure Appl. Chem., 61, 1148 (1989). Validation of analytical methods: Methodology, Pharmeuropa, 8, 114 (1996). R.B. George and P.D. Wright, Anal. Chem., 69, 2221 (1997). G. Wynia, P. Post, J. Broersten and F.A. Maris, Chromatographia, 39, 355 (1994). R.J. Markovitch, S. Ong and J. Rosen, J. Chromatogr. Sci., 35, 584 (1997). M. Desmet, K. Roels, L. Vanhoof and W. Lauwers, Pharm. Forum, 21, 501 (1995). C.M. Riley, Statistical parameters and analytical figures of Merit in Development and Validation of Analytical Methods, C.M. Riley and T.W. Rosanske, Ed., Pergamon, 1996, pp15-72. European Pharmacopoeia, Addendum 2000, pp31-36. [Please note that information concerning residual solvents are susceptible to be modified in the successive editions of the European Pharmacopeia.] Progress report of the working party on residual solvent (Technical) of the European Pharmacopoeia Commission, Pharmeuropa, 8, 586 (1996). Note for Guidance on impurities: Residual solvents, Drugs made in Germany, 41, 98 (1998).
17
Environmental Impact of Solvents 17.1 THE ENVIRONMENTAL FATE AND MOVEMENT OF ORGANIC SOLVENTS IN WATER, SOIL, AND AIRa William R. Roy Illinois State Geological Survey, Champaign, IL, USA
17.1.1 INTRODUCTION Organic solvents are released into the environment by air emissions, industrial and waste-treatment effluents, accidental spillages, leaking tanks, and the land disposal of solvent-containing wastes. For example, the polar liquid acetone is used as a solvent and as an intermediate in chemical production. ATSDR1 estimated that about 82 million kg of acetone was released into the atmosphere from manufacturing and processing facilities in the U.S. in 1990. About 582,000 kg of acetone was discharged to water bodies from the same type of facilities in the U.S. ATSDR2 estimated that in 1988 about 48,100 kg of tetrachloroethylene was released to land by manufacturing facilities in the U.S. Once released, there are numerous physical and chemical mechanisms that will control how a solvent will move in the environment. As solvents are released into the environment, they may partition into air, water, and soil phases. While in these phases, solvents may be chemically transformed into other compounds that are less problematic to the environment. Understanding how organic solvents partition and behave in the environment has led to better management approaches to solvents and solvent-containing wastes. There are many published reference books written about the environmental fate of organic chemicals in air, water, and soil.3-7 The purpose of this section is to summarize the environmental fate of six groups of solvents (Table 17.1.1) in air, water, and soil. A knowledge of the likely pathways for the environmental fate of organic solvents can serve as the technical basis for the management of solvents and solvent-containing wastes.
a
Publication authorized by the Chief, Illinois State Geological Survey
1150
William R. Roy
Table 17.1.1. The six groups of solvents discussed in this section Alcohols n-Butyl alcohol Isobutyl alcohol Methanol Benzene Derivatives Benzene Chlorobenzene o-Cresol o-Dichlorobenzene Ethylbenzene Nitrobenzene Toluene o-Xylene Chlorinated Aliphatic Hydrocarbons Carbon tetrachloride Dichloromethane Tetrachloroethylene 1,1,1-Trichloroethane Trichloroethylene
Chlorinated Fluorocarbons Trichlorofluoromethane (F-11), 1,1,2,2-Tetrachloro-1,2-difluoroethane ( F-112) 1,1,2-Trichloro-1,2,2-trifluoroethane (F-113) 1,2,-Dichlorotetrafluoroethane (F-114) Ketones Acetone Cyclohexanone Methyl ethyl ketone Methy isobutyl ketone Others Carbon disulfide Diethyl ether Ethyl acetate Hexane Decane (a major component of mineral spirits) Pyridine Tetrahydrofuran
17.1.2 WATER 17.1.2.1 Solubility One of the most important properties of an organic solvent is its solubility in water. The greater a compound's solubility, the more likely that a solvent or a solvent-containing waste will dissolve into water and become part of the hydrological cycle. Hence, water solubility can affect the extent of leaching of solvent wastes into groundwater, and the movement of dissolved solvent into rivers and lakes. Aqueous solubility also determines the efficacy of removal from the atmosphere through dissolution into precipitation. The solubility of solvents in water may be affected by temperature, salinity, dissolved organic matter, and the presence of other organic solvents. 17.1.2.2 Volatilization Solvents dissolved in water may volatilize into the atmosphere or soil gases. A Henry's Law constant (KH) can be used to classify the behavior of dissolved solvents. Henry's Law describes the ratio of the partial pressure of the vapor phase of an ideal gas (Pi) to its mole fraction (Xi) in a dilute solution, viz., K H ( i ) = Pi / X i
[17.1.1]
In the absence of measured data, a Henry's Law constant for a given solvent may be estimated by dividing the vapor pressure of the solvent by its solubility in water (Si) at the same temperature; KH(i) = Pi (atm) / Si (mol/m3 solvent)
[17.1.2]
A KH value of less than 10-4 atm-mol/m3 suggests that volatilization would probably not be a significant fate mechanism for the dissolved solvent. The rate of volatilization is
17.1 The environmental fate and movement of organic solvents
1151
more complex, and depends on the rate of flow, depth, and turbulence of both the body of water and the atmosphere above it. In the absence of measured values, there are a number of estimation techniques to predict the rate of removal from water.8 17.1.2.3 Degradation The disappearance of a solvent from solution can also be the result of a number of abiotic and biotic processes that transform or degrade the compound into daughter compounds that may have different physicochemical properties from the parent solvent. Hydrolysis, a chemical reaction where an organic solvent reacts with water, is not one reaction, but a family of reactions that can be the most important processes that determine the fate of many organic compounds.9 Photodegradation is another family of chemical reactions where the solvent in solution may react directly under solar radiation, or with dissolved constituents that have been made reactive by solar radiation. For example, the photolysis of water yields a hydroxyl radical: H2 O + hν → HO • +H
[17.1.3]
Other oxidants such as peroxy radicals (RO2r) and ozone can react with solvents in water. The subject of photodegradation is treated in more detail under atmospheric processes (17.1.4). Biodegradation is a family of biologically mediated (typically by microorganisms) conversions or transformations of a parent compound. The ultimate end-products of biodegradation are the conversion of organic compounds to inorganic compounds associated with normal metabolic processes.10 This topic will be addressed under Soil (17.1.3.3). 17.1.2.4 Adsorption Adsorption is a physicochemical process whereby a dissolved solvent may be concentrated at solid-liquid interfaces such as water in contact with soil or sediment. In general, the extent of adsorption is inversely proportional to solubility; sparingly soluble solvents have a greater tendency to adsorb or partition to the organic matter in soil or sediment (see Soil, 17.1.3.2). 17.1.3 SOIL 17.1.3.1 Volatilization Volatilization from soil may be an important mechanism for the movement of solvents from spills or from land disposed solvent-containing wastes. The efficacy and rate of volatilization from soil depends on the solvent's vapor pressure, water solubility, and the properties of the soil such as soil-water content, airflow rate, humidity, temperature and the adsorption and diffusion characteristics of the soil. Organic-solvent vapors move through the unsaturated zone (the interval between the ground surface and the water-saturated zone) in response to two different mechanisms; convection and diffusion. The driving force for convective movement is the gradient of total gas pressure. In the case of diffusion, the driving force is the partial-pressure gradient of each gaseous component in the soil air. The rate of diffusion of a solvent in bulk air can be described by Fick's Law, viz., Q = −Df ∇ a
[17.1.4]
1152
William R. Roy
where: Q Df ∇a
diffusive flux (mass/area-time) diffusion coefficient (area/time) concentration gradient (mass/volume/distance)
Compared with the relatively unobstructed path for the diffusion of solvents in the atmosphere, diffusion coefficients for solvents in soil air will be less because of the tortuosity of the soil matrix pathways. Several functional relationships have been developed that relate the soil diffusion coefficient (Ds) to various soil properties (see Roy and Griffin11), such as the Millington Equation12 Ds = Df η a3. 3 / η t2
[17.1.5]
where: ηa ηt
the air-filled porosity, and total soil porosity
17.1.3.2 Adsorption As discussed in 17.1.2.4., adsorption by soil components can remove solvents dissolved in water. Furthermore, the rate of movement of dissolved solvents through soil may be retarded by adsorption-desorption reactions between the solvents and the solid phases. The partitioning of solvents between the liquid phase and soil is usually described by an adsorption isotherm. The adsorption of solvents may be described by the Freundlich Equation: x / m = Kf C l /n
[17.1.6]
where: x m Kf C n
the mass adsorbed mass of sorbent the Freundlich constant, a soil-specific term the equilibrium concentration of the solvent in water, and the Freundlich exponent which describes the degree of non-linearity of the isotherm
When n is equal to one, the Freundlich Equation becomes a relatively simple partition function: x/m = KC
[17.1.7]
where K is an adsorption or distribution coefficient which is sometimes written as Kd. It has been known since the 1960s that the extent of adsorption of hydrophobic (sparingly soluble in water) solvents often correlates with the amount of organic matter in the soil.13 When Kd is divided by the amount of organic carbon in the soil, the resulting coefficient is the organic carbon-water partition coefficient (Koc): Kd x 100/organic carbon(%) = Koc
[17.1.8]
The organic carbon-water partition coefficient is a compound-specific term that allows the user to estimate the mobility of a solvent in saturated-soil water systems if the amount or organic carbon is known. For hydrophilic solvents, Koc values have been measured for many compounds. Other values were derived from empirical relationships drawn between water solubility or octanol-water partition coefficients.13
17.1 The environmental fate and movement of organic solvents
1153
17.1.3.3 Degradation Solvents may be degraded in soil by the same mechanisms as those in water. In biodegradation, microorganisms utilize the carbon of the solvents for cell growth and maintenance. In general, the more similar a solvent is to one that is naturally occurring, the more likely that it can be biodegraded into other compound(s) because the carbon is more available to the microbes. Moreover, the probability of biodegradation increases with the extent of water solubility of the compound. It is difficult to make generalities about the extent or rate of solvent biodegradation that can be expected in soil. Biodegradation can depend on the concentration of the solvent itself, competing processes that can make the solvent less available to microbes (such as adsorption), the population and diversity of microorganisms, and numerous soil properties such as water content, temperature, and reduction-oxidation potential. The rate and extent of biodegradation reported in studies appears to depend on the conditions under which the measurement was made. Some results, for example, were based on sludge-treatment plant simulations or other biological treatment facilities that had been optimized in terms of nutrient content, microbial acclimation, mechanical mixing of reactants, or temperature. Hence, these results may overestimate the extent of biodegradation in ambient soil in a spill or waste-disposal scenario. First-order kinetic models are commonly used to describe biodegradation because of their mathematical simplicity. First-order biodegradation is to be expected when the organisms are not increasing in abundance. A first-order model also lends itself to calculating a half-life (t1/2) which is a convenient parameter to classify the persistence of a solvent. If a solvent has a soil half-life of 6 months, then about half of the compound will have degraded in six months. After one year, about one fourth the initial amount would still be present, and after 3 half-lives (1.5 years), about 1/8 of the initial amount would be present. Howard et al.14 estimated ranges of half-lives for solvents in soil, water, and air. For solvents in soil, the dominant mechanism in the reviewed studies may have been biodegradation, but the overall values are indicative of the general persistence of a solvent without regard to the specific degradation mechanism(s) involved. 17.1.4 AIR 17.1.4.1 Degradation As introduced in 17.1.2.3, solvents may be photodegraded in both water and air. Atmospheric chemical reactions have been studied in detail, particularly in the context of smog formation, ozone depletion, and acid rain. The absorption of light by chemical species generates free radicals which are atoms, or groups of atoms that have unpaired electrons. These free radicals are very reactive, and can degrade atmospheric solvents. Atmospheric ozone, which occurs in trace amounts in both the troposphere (sea level to about 11 km) and in the stratosphere (11 km to 50 km elevation), can degrade solvents. Ozone is produced by the photochemical reaction: O 2 + hν → O + O
[17.1.9]
O + O2 → O3 + M
[17.1.10]
where M is another species such as molecular nitrogen that absorbs the excess energy given off by the reaction. Ozone-depleting substances include the chlorofluorocarbons (CFC) and carbon tetrachloride in the stratosphere.
1154
William R. Roy
17.1.4.2 Atmospheric residence time Vapor-phase solvents can dissolve into water vapor, and be subject to hydrolysis reactions and ultimately, precipitation (wet deposition), depending on the solubility of the given solvent. The solvents may also be adsorbed by particulate matter, and be subject to dry deposition. Lyman16 asserted that atmospheric residence time cannot be directly measured; that it must be estimated using simple models of the atmosphere. Howard et al.14 calculated ranges in half-lives for various organic compounds in the troposphere, and considered reaction rates with hydroxyl radicals, ozone, and by direct photolysis. 17.1.5 THE 31 SOLVENTS IN WATER 17.1.5.1 Solubility The solubility of the solvents in Table 17.1.1 ranges from those that are miscible with water to those with solubilities that are less than 0.1 mg/L (Table 17.1.2). Acetone, methanol, pyridine and tetrahydrofuran will readily mix with water in any proportion. The solvents that have an aqueous solubility of greater than 10,000 mg/L are considered relatively hydrophillic as well. Most of the benzene derivatives and chlorinated fluorocarbons are relatively hydrophobic. Hexane and decane are the least soluble of the 31 solvents in Table 17.1.1. Most material safety data sheets for decane indicate that the n-alkane is “insoluble” and that the solubility of hexane is “negligible.” How the solubility of each solvent affects its fate in soil, water, and air is illustrated in the following sections. Table 17.1.2. The solubility of the solvents in water at 25oC Solubility, mg/L
Solvent (reference)
∞
Acetone (1) Methanol (1) Pyridine (1) Tetrahydrofuran (1)
239,000
Methyl ethyl ketone (4)
77,000 76,000 64,000 60,050 25,950 23,000 20,400 13,000
n-Butyl alcohol (4) Isobutyl alcohol (4) Ethyl acetate (4) Diethyl ether (4) o-Cresol (17) Cyclohexanone (4) Methyl isobutyl ketone (4) Dichloromethane (4)
2,100 1,900 1,780 1,495 1,100 1,080
Carbon disulfide (4) Nitrobenzene (18) Benzene (19) 1,1,1-Trichloroethane (4) Trichloroethylene (4) F-11 (4)
Miscible
Relatively hydrophillic
17.1 The environmental fate and movement of organic solvents
Solubility, mg/L
1155
Solvent (reference)
805 535 472 175 170 161 156 150 130 120
Carbon tetrachloride (4) Toluene (20) Chlorobenzene (17) o-Xylene (4) F-113 (4) Ethylbenzene (17) o-Dichlorobenzene (17) Tetrachloroethylene (4) F-114 (4) F-112 (4)
9.5
Hexane (21)
0.05
Decane (22)
Relatively hydrophobic
“Insoluble”
17.1.5.2 Volatilization from water Henry's Law constants were compiled for each of the solvents in Table 17.1.1. The numerical values ranged over 7 orders of magnitude (Table 17.1.3). Based on these values, it can be expected that volatilization from water will be a significant fate mechanism for decane, hexane, the chlorinated fluorocarbons, carbon tetrachloride, tetrachloroethylene and trichloroethylene. Many of the solvents in Table 17.1.1 are characterized by KH values of 10-3 to 10-2 atm-m3/mole; volatilization from water can be an important pathway for these solvents, depending on the specific situation. Volatilization may be a relatively slow process for the remaining solvents. The actual rate of volatilization of some solvents from water has been experimentally measured.4,17 However, experimental data are lacking for some compounds, and the diversity of experimental conditions makes generalizations difficult. Thomas8 described a two-layer model of the liquid-gas interface that is based on a Henry's Law constant and mass-transfer coefficients. To illustrate the relative volatilities of the solvents in water, the half-lives of each solvent in a shallow stream were compiled (Table 17.1.4). The stream was assumed to be 1 meter deep and flowing at a rate of 1 meter per second. With the exception of hexane, it was also assumed that there was a breeze blowing across the stream at a rate of 3 meters per second. Under these conditions, the predicted half-lives of many of the solvents in Table 16.1.1 are less than 10 hours, indicating that volatilization into the atmosphere can be a relatively rapid pathway for solvents released to surface water. The volatilization of pyridine, isobutyl alcohol, and cyclohexanone may be a slow process, and other fate processes may be more important in water. 17.1.5.3 Degradation in water As mentioned in 17.1.3.3, Howard et al.14 compiled ranges of half-life values for most of the organic solvents given in Table 17.1.1. If a “rapid” half-life is defined as in the range of 1 to 10 days, then about 12 of the solvents in Table 17.1.1 may degrade rapidly in surface water by primarily biodegradation (Figures 17.1.1 and 17.1.2). Abiotic mechanisms such as photo-oxidation, photolysis, and hydrolysis appear to be either slow or not significant. If “slow degradation” is defined as that taking longer than 100 days, then it appears that F-11 and most of the chlorinated hydrocarbons may be relatively persistent in surface water. The available data suggest that the half-life of nitrobenzene and isobutyl alcohol may be variable. Note that data were not available for all of the solvents listed in Table 17.1.1. In
1156
William R. Roy
Table 17.1.3. Henry's Law constants (KH) for the solvents at 25oC KH, atm-m3/mole
6.98 2.8 1.69
Table 17.1.4. Estimated half-lives for the solvents in water at 20oC
Solvent (reference)
Decane (22) F-114 (4) Hexane (21)
0.53
F-113 (4) -2
9.74 x 10 9.70 x 10-2 3.04 x 10-2 1.49 x 10-2 1.03 x 10-2
F-112 (4) F-11 (4) Carbon tetrachloride (4) Tetrachloroethylene (4) Trichloroethylene (4)
9.63 x 10-3 8.4 x 10-3 8.0 x 10-3 7.0 x 10-3 5.94 x 10-3 5.43 x 10-3 5.1 x 10-3 3.58 x 10-3 2.68 x 10-3 1.4 x 10-3 1.2 x 10-3
Tetrahydrofuran (4) Ethylbenzene (17) 1,1,1-Trichloroethane (4) Pyridine (4) Toluene (4) Benzene (4) o-Xylene (4) Chlorobenzene (23) Dichloromethane (4) Carbon disulfide (4) o-Dichlorobenzene (4)
7.48 x 10-4 4 x 10-4 1.35 x 10-4 1.2 x 10-4
Diethyl ether (11) Isobutyl alcohol (4) Methanol (4) Ethyl acetate (4)
Half life, h
Solvent
1.6 2.6 2.7 2.9 3.0 3.1 3.2 3.4 3.7 4.0 4.2 4.4 4.6 5.3
Tetrahydrofuran Carbon disulfide Hexanea Toluene Dichloromethane Ethylbenzene o-Xylene Trichloroethylene, F-11 Carbon tetrachloride F-112, F-113, F-114 Tetrachloroethylene o-Dichlorobenzene Chlorobenzene Methanol
10 18 45 74 80 90
Ethyl acetate Acetone Nitrobenzene Cyclohexanone Isobutyl alcohol Pyridine
a
Based on a wind speed of 1 meter per second.21
groundwater, the half-life values proposed by Howard et al.14 appear to be more variable than those for surface water. For examMethyl isobutyl ketone (4) 9.4 x 10-5 ple, the half-life of benzene ranges from 10 -5 4.26 x 10 Acetone (1) -5 days in aerobic groundwater to 2 years in 2.44 x 10 Nitrobenzene (2) -5 anaerobic groundwater.19 Such ranges in 1.2 x 10 Cyclohexanone (4) -5 half-lives make meaningful generalizations 1.05 x 10 Methyl ethyl ketone (4) difficult. However, it appears that metha-6 n-Butyl alcohol (4) 5.57 x 10 -6 nol, n-butyl alcohol, and other solvents (see 1.2 x 10 o-Cresol (4) Figures 17.1.1 and 17.1.2) may biodegrade in groundwater with a half-life that is less than 60 days. As with surface water, the chlorinated hydrocarbons may be relatively persistent in groundwater. Howard et al.14 cautioned that some of their proposed half-life generalizations were based on limited data or from screening studies that were extrapolated to surface and groundwater. Scow10 summarized that it is currently not possible to predict rates of biodegradation because of a lack of standardized experimental methods, and because the variables that control rates are not well understood. Hence, Figures 17.1.1 and 17.1.2 should be viewed as a summary of the potential for each solvent to degrade, pending more site-specific information.
17.1 The environmental fate and movement of organic solvents
Figure 17.1.1. The ranges in degradation half-lives for the alcohols and benzene derivatives in surface water, groundwater, and soil (data from Howard et al.14).
1157
Figure 17.1.2. The ranges in degradation half-lives for the chlorinated aliphatic hydrocarbons, F-11 and F-113, and ketones in surface water, groundwater, and soil (data from Howard et al.14).
17.1.6 SOIL 17.1.6.1 Volatilization Soil diffusion coefficients were estimated for most of the solvents in Table 17.1.1. Using the Millington Equation, the resulting coefficients (Table 17.1.5) ranged from 0.05 to 0.11 m2/day. Hence, there was little variation in magnitude between the values for these particular solvents. As discussed in Thomas24 the diffusion of gases and vapors in unsaturated soil is a relatively slow process. The coefficients in Table 17.1.5 do not indicate the rate at which olvents can move in soil; such rates must be either measured experimentally or predicted using models that require input data such as soil porosity, moisture content, and the concentrations of the solvents in the vapor phase to calculate fluxes based solely on advective movement. Variations in water content, for example, will control vapor-phase movement. The presence of water can reduce the air porosity of soil, thereby reducing the soil diffusion coefficient (Eq. 17.1.5). Moreover, relatively water-soluble chemicals may dissolve into water in the vadose zone. Hence, water can act as a barrier to the movement of solvent vapors from the subsurface to the surface. Solvents spilled onto the surface of soil may volatilize into the atmosphere. The Dow Method24 was used in this section to estimate half-life values of each solvent if spilled on the surface of a dry soil. The Dow Method is a simple relationship that was derived for the evaporation of pesticides from bare soil; t1/2 (days) = 1.58 x 10-8 (KocS/Pv)
[17.1.11]
1158
William R. Roy
Table 17.1.5. Estimated soil diffusion coefficients Ds (from Roy and Griffin11) Ds, m2/day
Solvent n-Butyl alcohol Isobutyl alcohol Methanol Benzene Chlorobenzene o-Cresol
Table 17.1.6. Estimated soil-evaporation half lives
o
0.062 (25 C) o
0.050 (0 C)
Solvent
Half-life, min.
o-Cresol
38
Nitrobenzene
19
o
n-Butyl alcohol
18
o
Pyridine
8
o
Decane
4
o
Isobutanol Cyclohexanone
1
All other solvents
<1
0.111 (25 C) 0.060 (15 C) 0.052 (30 C) 0.053 (15 C) o
o-Dichlorobenzene
0.049 (20 C)
Ethylbenzene
0.046 (0oC)
Nitrobenzene
0.050 (20oC)
Toluene
0.058 (25oC)
o-Xylene
0.049 (15oC)
Carbon tetrachloride
0.051 (25oC)
Dichloromethane
0.070 (15oC)
Tetrachloroethylene
0.051 (20oC) o
1,1,1-Trichloroethane
0.075 (20 C)
Trichloroethylene
0.058 (15oC)
F-11
0.060 (15oC)
F-112
-
F-113
0.053 (15oC)
F-114
0.056 (15oC)
Acetone
0.076 (0oC)
Cyclohexanone
-
Methyl ethyl ketone
-
Methyl isobutyl ketone Carbon disulfide
0.074 (25oC)
Diethyl ether
0.054 (0oC)
Ethyl acetate
0.059 (25oC)
Hexane
-
Mineral spirits
-
Pyridine
-
Tetrahydrofuran
-
where: t1/2 Koc S Pv
evaporation half-life (days) organic carbon-water partition coefficient (L/kg) solubility in water (mg/L), and vapor pressure (mm Hg at 20oC)
The resulting estimated half-life is inversely proportional to vapor pressure; the greater the vapor pressure, the greater the extent of volatilization. Conversely, the rate of volatilization will be reduced if the solvent readily dissolves into water or is adsorbed by the soil. Organic carbon-water partition coefficients were compiled for each solvent (see 17.1.6.2.), and vapor pressure data (not shown) were collected from Howard.4 The resulting half-life estimates (Table 17.1.6) indicated that volatilization would be a major pathway if the liquid solvents were spilled on soil; all of the half-life estimates were less than one hour. Thomas24 cautioned, however, that soil moisture, soil type, temperature, and wind conditions were not incorporated in the simple Dow Model.
17.1 The environmental fate and movement of organic solvents
1159
Table 17.1.7. The organic carbon-water partition coefficients (Koc) of the solvents at 25oC Koc, L/kg
Solvent (reference)
<1
Methanol (13), Tetrahydrofurana
1 4 7 8 9
Acetone (13) Methyl ethyl ketone (13) Pyridine (13) Ethyl acetate, isobutyl alcohol (13) Diethyl ether (13)
10 20 24 25 63 67 72 97
Cyclohexanone (13) o-Cresol (17) Methyl isobutyl ketone (13) Dichloromethane (13) Carbon disulfide (13) Nitrobenzene (13) n-Butyl alcohol (4) Benzene (13)
110 152 155 164 242 303 318 343 363 372 437 457 479
Carbon tetrachloride (4) Trichloroethylene (13) 1,1,1-Trichloroethane (13) Ethylbenzene (17) Toluene (26) Tetrachloroethylene (13) Chlorobenzene (13) o-Dichlorobenzene (25) o-Xylene (13) F-113 (13) F-114 (13) F-112 (13) F-11 (13)
1,950
Hexane (21) a
57,100
Decane
Mobile
Relatively mobile
Relatively Immobile
a
Calculated using the relationship logKoc = 3.95 - 0.62logS where S = water solubility in mg/L (see Hassett et al.25)
17.1.6.2 Adsorption Organic carbon-water partition coefficients were compiled (Table 17.1.7) for each of the solvents in Table 17.1.1. A Koc value is a measure of the affinity of a solvent to partition to organic matter which in turn will control the mobility of the solute in soil and groundwater under convective flow. Although the actual amount of organic matter will determine the extent of adsorption, a solvent with a Koc value of less than 100 L/kg is generally regarded as relatively mobile in saturated materials. Hence, adsorption may not be a significant fate mechanism for 16 of the solvents in Table 17.1.1. In contrast, adsorption by organic matter may be a major fate mechanism controlling the fate of three of the benzene derivatives, and most of the chlorinated compounds. Hexane and particularly decane would likely be relatively immobile. However, when the organic C content of an adsorbent is less than about 1
1160
William R. Roy
Figure 17.1.3. The ranges in atmospheric half-life of the solvents in Table 17.1.1 (data from Howard et al.14 and ATSDR21).
g/kg, the organic C fraction is not a valid predictor of the partitioning of nonpolar organic compounds,27 and other properties such as pH, surface area, or surface chemistry contribute to or dominate the extent of adsorption. Moreover, pyridine occurs at a cation (pKa = 5.25) over a wide pH range, and thus it is adsorbed by electrostatic interactions rather than by the hydrophobic mechanisms that are endemic to using Koc values to predict mobility. The desorption of solvents from soil has not been extensively measured. In the application of advection-dispersion models to predict solute movement, it is generally assumed that adsorption is reversible. However, the adsorption of the solutes in Table 17.1.1 may not be reversible. For example, hysteresis is often observed in pesticide adsorption-desorption studies with soils.28 The measurement and interpretation of desorption data for solid-liquid systems is not well understood.29,30 Once adsorbed, some adsorbates may react further to become covalently and irreversibly bound, while others may become physically trapped in the soil matrix.28 The non-singularity of adsorption-desorption may sometimes result from experimental artifacts.28,31 17.1.6.3 Degradation As discussed in 17.1.3.3., Howard et al.14 also estimated soil half-life values (Figures 17.1.1 and 17.1.2) for the degradation of most of the solvents in Table 17.1.1. Biodegradation was cited as the most rapid process available to degrade solvents in a biologically active soil. The numerical values obtained were often the same as those estimated for surface water.
17.1 The environmental fate and movement of organic solvents
1161
Consequently, it appears likely that the alcohols, ketones, o-cresol, ethyl acetate, and pyridine will degrade rapidly in soil if rapidly is defined as having a half-life of 10 days or less. Most of the benzene derivatives, F-11, and the chlorinated aliphatic hydrocarbons may be relatively persistent in soil. Analogous information was not located for diethyl ether, hexane, decane, or tetrahydrofuran. ATSDR21 for example, found that there was little information available for the degradation of n-hexane in soil. It was suggested that n-hexane can degrade to alcohols, aldehydes, and fatty acids under aerobic conditions. 17.1.7 AIR Once released into the atmosphere, the most rapid mechanism to attenuate most of the solvents in Table 17.1.1 appears to be by photo-oxidation by hydroxyl radicals in the troposphere. Based on the estimates by Howard et al.,14 it appeared that nine of the solvents can be characterized by an atmospheric residence half-life of 10 days or less (Figure 17.1.3). The photo-oxidation of solvents yields products. For example, the reaction of OH radicals with n-hexane can yield aldehydes, ketones, and nitrates.21 The reaction of some of the solvents with ozone may be much slower. For example, the half-life for the reaction of benzene with ozone may be longer than 100 years.19 Solvents such as carbon tetrachloride, 1,1,1-trichloroethane, and the chlorinated fluorocarbons may be relatively resistant to photo-oxidation. The major fate mechanism of atmospheric 1,1,1-trichloroethane, for example, may be wet deposition.32 REFERENCES 1 2 3 4 5 6 7 8
9 10
11 12 13 14 15
16
Agency for Toxic Substances and Disease Registry. Toxicological Profile for Acetone. ATSDR, Atlanta, Georgia, 1994. Agency for Toxic Substances and Disease Registry. Toxicological Profile for Tetrachloroethylene. ATSDR, Atlanta, Georgia, 1991. D. Calamari (ed.) Chemical Exposure Predictions, Lewis Publishers, 1993. P. H. Howard, Handbook of Environmental Fate and Exposure Data for Organic Chemicals. Vol. II Solvents. Lewis Publishers, Chelsea, Michigan, 1990. W. J. Lyman, W. F. Reehl, and D. H. Rosenblatt (eds). Handbook of Chemical Property Estimation Methods, American Chemical Society, Washington, D.C., 1990. R. E. Ney. Fate and Transport of Organic Chemicals in the Environment. 2nd ed. Government Institutes, Inc. Rockville, MD, 1995. B. L. Sawhney and K. Brown (eds.). Reactions and Movement of Organic Chemicals in Soils, Soil Science Society of America, Special Publication Number 22, 1989. R. G. Thomas, Volatilization From Water, W. J. Lyman, W. F. Reehl, and D. H. Rosenblatt (eds). in Handbook of Chemical Property Estimation Methods, American Chemical Society, Washington, D.C, Chap. 15, 1990. J. C. Harris. Rate of Hydrolysis, Lyman, W. J., W. F. Reehl, and D. H. Rosenblatt (eds). in Handbook of Chemical Property Estimation Methods, American Chemical Society, Washington, D.C, Chap. 7, 1990. K. M. Scow, 1990, Rate of Biodegradation, W. J. Lyman, W. F. Reehl, and D. H. Rosenblatt (eds). in Handbook of Chemical Property Estimation Methods, American Chemical Society, Washington, D.C, Chap. 9, 1990. W. R. Roy and R. A. Griffin, Environ. Geol. Water Sci., 15, 101 (1990). R. J. Millington, Science, 130, 100 (1959). W. R. Roy and R. A. Griffin, Environ. Geol. Water Sci., 7, 241 (1985). P. H. Howard, R. S. Boethling, W. F. Jarvis, W. M. Meylan, and Edward M. Michalenko. Handbook of Environmental Degradation Rates, Lewis Publishers, Chelsea, Michigan, 1991. M. Alexander and K. M. Scow, Kinetics of Biodegradation, B. L. Sawhney, and K. Brown (eds.). in Reactions and Movement of Organic Chemicals in Soils. Soil Science Society of America Special Publication, Number 22, Chap. 10, 1989. W. J. Lyman, Atmospheric Residence Time, W. J. Lyman, W. F. Reehl, and D. H. Rosenblatt (eds). in Handbook of Chemical Property Estimation Methods. American Chemical Society, Washington, D.C, Chap. 10, 1990.
1162
17 18 19 20 21 22 23 24
25
26 27 28 29
30 31 32
William R. Roy
P. H. Howard, Handbook of Environmental Fate and Exposure Data for Organic Chemicals. Vol. I. Large Production and Priority Pollutants. Lewis Publishers, Chelsea, Michigan, 1989. Agency for Toxic Substances and Disease Registry. Toxicological Profile for Nitrobenzene. ATSDR, Atlanta, Georgia, 1989. Agency for Toxic Substances and Disease Registry. Toxicological Profile for Benzene. ATSDR, Atlanta, Georgia, 1991. Agency for Toxic Substances and Disease Registry. Toxicological Profile for Toluene. ATSDR, Atlanta, Georgia, 1998. Agency for Toxic Substances and Disease Registry. Toxicological Profile for Hexane. ATSDR, Atlanta, Georgia, 1997. D. MacKay and W. Y. Shiu, J. Phys. Chem. Ref. Data, 4, 1175 (1981). Agency for Toxic Substances and Disease Registry. Toxicological Profile for Chlorobenzene. ATSDR, Atlanta, Georgia, 1989. R. G. Thomas. Volatilization from Soil in W. J. Lyman, W. F. Reehl, and D. H. Rosenblatt (eds). Handbook of Chemical Property Estimation Methods. American Chemical Society, Washington, D.C, Chap. 16, 1990. J. J. Hassett, W. L. Banwart, and R. A. Griffin. Correlation of compound properties with soil sorption characteristics of nonpolar compounds by soils and sediments; concepts and limitations In C. W. Francis and S. I. Auerback (eds), Environmental and Solid Wastes, Characterization, Treatment, and Disposal, Chap. 15, p. 161-178, Butterworth Publishers, London, 1983. J. M. Gosset, Environ. Sci. Tech., 21, 202 (1987). T. Stauffer W. G. MacIntyre. 1986, Tox. Chem., 5, 949 (1986). W. C. Koskinen and S. S. Harper. The retention process, mechanisms. p. 51-77. In Pesticides in the Soil Environment. Soil Science Society of America Book Series, no. 2, 1990. R. E. Green, J. M. Davidson, and J. W. Biggar. An assessment of methods for determining adsorption-desorption of organic chemicals. p. 73-82. In A. Bainn and U. Kafkafi (eds.), Agrochemicals in Soils, Pergamon Press, New York, 1980. R. Calvet, Environ. Health Perspectives, 83,145 (1989). B. T. Bowman and W. W. Sans, J. Environ. Qual., 14, 270 (1985). Agency for Toxic Substances and Disease Registry. Toxicological Profile for 1,1,1-Trichloroethane. ATSDR, Atlanta, Georgia, 1989.
17.2 FATE-BASED MANAGEMENT OF ORGANIC SOLVENT-CONTAINING WASTESa William R. Roy Illinois State Geological Survey, Champaign, IL, USA
17.2.1 INTRODUCTION The wide spread detection of dissolved organic compounds in groundwater is a major environmental concern, and has led to greater emphasis on incineration and waste minimization when compared with the land disposal of solvent-containing wastes. The movement and environmental fate of dissolved organic solvents from point sources can be approximated by the use of computer-assisted, solute-transport models. These models require information about the composition of leachate plumes, and site-specific hydrogeological and chemical
a
Publication authorized by the Chief, Illinois State Geological Survey
17.2 Fate-based management
1163
data for the leachate-site system. A given land-disposal site has a finite capacity to attenuate organic solvents in solution to environmentally acceptable levels. If the attenuation capacity of a site can be estimated, then the resulting information can be used as criteria to make decisions as to what wastes should be landfilled, and what quantities of solvent in a given waste can be safely accepted. The purpose of this section is to summarize studies1-3 that were conducted that illustrate how knowledge of the environmental fate and movement of the solvents in Section 17.1 can be used in managing solvent-containing wastes. These studies were conducted by using computer simulations to assess the fate of organic compounds in leachate at a waste-disposal site. 17.2.1.1 The waste disposal site There are three major factors that will ultimately determine the success of a land-disposal site in being protective of the environment with respect to groundwater contamination by organic solvents: (1) the environmental fate and toxicity of the solvent; (2) the mass loading rate, i.e., the amount of solvent entering the subsurface during a given time, and (3) the total amount of solvent available to leach into the groundwater. The environmental fate of the solvents was discussed in 17.1. The hypothetical waste-disposal site used in this evaluation (Figure 17.2.1) had a single waste trench having an area of 0.4 hectare. Although site-specific dimensions may be assigned with actual sites, this hypothetical site was considered representative of many situations found in the field. The trench was 12.2 meters (40 ft) deep and was constructed with a synthetic/compacted-soil double-liner system. The bottom of the trench was in direct contact with a sandy aquifer that was 6.1 meters (20 ft) thick. The top of the water table was defined as being at the top of the sandy aquifer. Thus, this site was designed as a worst-case scenario. The sandy aquifer directly beneath the hazardous-waste trench would offer little resistance to the movement of contaminants. To further compound a worst-case situation, it was also assumed that the entire trench was saturated with leachate, generating a 12.2 meter (40 ft) hydraulic head through the liner. This could correspond to a situation where the trench had completely filled with leachate because the leachate collection system had either failed or the site had been abandoned. The following aquifer properties, typical of sandy materials,1 were used in the study:
Figure 17.2.1. Design of the waste-disposal site model used in the simulations (Roy et al.1).
1164
William R. Roy
saturated hydraulic conductivity = 10-3 cm/sec saturated volumetric water content = 0.36 cm3/cm3 dry bulk density = 1.7 g/cm3 hydraulic gradient = 0.01 cm/cm mean organic carbon content = 0.18% These aquifer properties yield a groundwater flow rate of 9.3 meters (30 ft) per year. The direction of groundwater flow is shown in Figure 17.2.1 to be from left to right. The edge of the disposal trench was 154 meters (500 ft) from a monitoring well that was open to the entire thickness of the aquifer. This monitoring well served as a worst-case receptor because it was placed in the center of the flow path at the site boundary and it served as the compliance point for the site. The downgradient concentrations of organic solvents at the compliance well, as predicted by a solute-transport model, were used to evaluate whether the attenuation capacity of the site was adequate to reduce the contaminants to acceptable concentrations before they migrated beyond the compliance point. 17.2.1.2 The advection-dispersion model and the required input The 2-dimensional, solute-transport computer program PLUME was used to conduct contaminant migration studies. Detailed information about PLUME, including boundary conditions and quantitative estimates of dispersion and groundwater dilution, were summarized by Griffin and Roy.3 In this relatively simple and conservative approach, PLUME did not take into account volatilization from water. Volatilization is a major process for many of the solvents (see Section 17.1). Adsorption was assumed to be reversible, and soil-water partition coefficients were calculated by assuming that the aquifer contained 0.18% organic carbon (see Roy and Griffin4). A degradation half-life was assigned to each solvent (Table 17.2.1). In many cases, conservative half-life values were used. For example, all of the ketones were assigned a half-life of 5 years, which is much longer than those proposed for ketones in groundwater (see Section 17.1). The movement of each solvent was modeled separately whereas it should be recognized that solvents in mixtures may have different chemical properties that can ultimately affect their fate and movement. 17.2.1.3 Maximum permissible concentrations Central to the type of assessment is a definition of an environmentally acceptable concentration of each contaminant. These acceptable levels were defined as Maximum Permissible Concentrations (MPC), and were based on the toxicological assessments of solvents in drinking water by George and Siegel.5 These MPC levels (Table 17.2.1) are not the same levels as the current Maximum Contaminant Levels (MCL) that were promulgated by the U.S. Environmental Protection Agency for drinking water. 17.2.1.4 Distribution of organic compounds in leachate An initial solute concentration must be selected for the application of solute transport models. An initial concentration for each solvent was based on the chemical composition of leachates from hazardous-waste sites.1 Where available, the largest reported concentration was used in the modeling efforts (Table 17.2.1). No published data were located for some of the solvents such as cyclohexanone. In such cases, the initial concentration was arbitrarily assigned as 1,000 mg/L or it was equated to the compound's solubility in water. Hexane, decane, and tetrahydofuran were not included in these studies. The amount of mass of each organic compound entering the aquifer via the double-liner system was calculated using these initial leachate concentrations. There was a continuous 12.2-meter head driving the leachate through the liner. Leachate was predicted to
17.2 Fate-based management
1165
break through the liner in 30 years. Under these conditions, approximately 131,720 L/year/acre of leachate would seep through the liner. The assumptions used in deriving this flow estimate were summarized in Roy et al.1 Table 17.2.1. The six groups of solvents discussed in this section, their corresponding Maximum Permissible Concentrations (MPC), the largest reported concentrations in leachate (LC), and the assigned half-lives from Roy et al.1 MPC, µg/L
LC, mg/L
Half-life, years
n-Butyl alcohol
2,070
1,000
5
Isobutyl alcohol
2,070
1,000
5
Methanol
3,600
42.4
5
Benzene
1.6
7.37
20
Chlorobenzene
488
4.62
20
o-Cresol
304
0.21
20
o-Dichlorobenzene
400
0.67
50
Ethylbenzene
1,400
10.1
10
Nitrobenzene
19,800
0.74
20
Toluene
14,300
100
10
o-Xylene
14,300
19.7
10
Carbon tetrachloride
0.4
25.0
50
Dichloromethane
0.19
430
20
Tetrachloroethylene
0.80
8.20
20
1,1,1-Trichloroethane
6.00
590
50
Trichloroethylene
2.70
260
20
Trichlorofluoromethane (F-11)
0.19
0.14
50
1,1,2,2-Tetrachloro-1,2-difluoroethane ( F-112)
0.19
120
50
1,1,2-Trichloro-1,2,2-trifluoroethane (F-113)
0.19
170
stable
1,2,-Dichlorotetrafluoroethane (F-114)
0.19
130
stable
Acetone
35,000
62
5
Cyclohexanone
3,500
1,000
5
Methyl ethyl ketone
30,000
53.0
5
Alcohols
Benzene Derivatives
Chlorinated Aliphatic Hydrocarbons
Chlorinated Fluorocarbons
Ketones
1166
William R. Roy
MPC, µg/L
LC, mg/L
Half-life, years
143
10.0
5
830
1,000
10
Diethyl ether
55,000
1,000
5
Ethyl acetate
55,000
1,000
5
207
1,000
20
Methyl isobutyl ketone Others Carbon disulfide
Pyridine
A mass-loading rate was conservatively calculated for each solvent as, M lr = Q × Cl
[17.2.1]
where: Mlr Q Cl
the mass loading rate (mass/time/area), calculated leachate flux (131.7 kL/year/hectare), and largest concentration of the solvent in leachate (mg/L)
17.2.2 MOVEMENT OF SOLVENTS IN GROUNDWATER Ketones and alcohols have little tendency to be adsorbed by soil materials (see Section 17.1), and would appear at the compliance point only a few years after liner breakthrough (Figure 17.2.2). Because the mass loading rates were held constant, the ketones and alcohols assumed maximum steady-state concentrations after approximately 40 to 50 years (Figure 17.2.3). These two classes of organic solvents degrade readily, reducing their downgradient concentrations. The distribution of the benzene derivatives at the compliance well depended substantially on their soil-water partition coefficients, their tendencies to de-
Figure 17.2.2. The predicted distribution of methyl ethyl ketone (mg/L) in the aquifer 100 years after the leachate has broken through the liner (Griffin and Roy3).
Figure 17.2.3. The predicted concentrations of methyl ethyl ketone, methanol, benzene, 1,1,1-trichloroethane, ethyl acetate, and F-22 at the compliance point as a function of time (Roy et al.1).
17.2 Fate-based management
1167
grade, and the initial concentrations. Under the conditions described, the relative steady-state concentrations of the benzene derivatives were: toluene > benzene > chlorobenzene > p-xylene > nitrobenzene > o-dichlorobenzene > o-cresol > ethyl benzene. Methylene chloride and 1,1,1-trichlorethane would dominate the chlorinated hydrocarbons. Among the group of unrelated organic solvents, the concentration of pyridine at the well was predicted to increase rapidly. Pyridine would eventually dominate this group in the relative order: pyridine > carbon disulfide > ethyl acetate > diethyl ether. The relative order of fluorocarbons at the compliance well in terms of concentration was: F-21, F-22 >> F-12 > F-113 > F-114 > F-112 > R-112a > FC-115 >> F-11. In brief, the computer simulations predicted that all 28 organic compounds would eventually migrate from the waste trench, and be detected at the compliance well. The predicted concentrations varied by four orders of magnitude, and were largely influenced by the initial concentrations used in calculating the mass loading rate to the aquifer. 17.2.3 MASS LIMITATIONS The next step in this analysis was to determine whether these predicted concentrations would pose an environmental hazard by evaluating whether the site was capable of attenuating the concentrations of the organic compounds to levels that are protective of human health. In Figure 17.2.4 the predicted steady-state concentrations of the organic compounds
Figure 17.2.4. The predicted steady-state concentrations (Css) of each solvent in groundwater at the compliance point as a function of its Maximum Permissible (MPC) Concentration (Roy et al.1).
1168
William R. Roy
in groundwater at the compliance well were plotted against their MPCs. The boundary shown in Figure 17.2.4 represents the situation where the steady-state concentration (Css) equals the MPC. Consequently, the predicted Css is less than its corresponding MPC when the Css of a given compound plots in the lower-right side. In this situation, these organic compounds could enter the aquifer at a constant mass loading rate without exceeding the attenuation capacity of the site. The steady-state concentrations of twenty solvents exceeded their corresponding MPCs. The continuous addition of these organic compounds (i.e., a constant mass loading rate) would exceed the site's ability to attenuate them to environmentally acceptable levels in this worst-case scenario. There are two avenues for reducing the steady-state concentrations downgradient from the trench: (1) reduce the mass loading rate, and/or (2) reduce the mass of organic compound available to leach into the aquifer. Because, the RCRA-required double liner was regarded as the state-of-the-art with respect to liner systems, it was not technically feasible to reduce the volume of leachate seeping into the aquifer under the conditions imposed. The worst-case conditions could be relaxed by assuming a lower leachate head in the landfill or by providing a functional leachate-collection system. Either condition would be reasonable and would reduce the mass loading rate. Another alternative is to reduce the mass available for leaching. In the previous simulations, the mass available to enter the aquifer was assumed to be infinite. Solute transport models can be used to estimate threshold values for the amounts of wastes initially landfilled.2 A threshold mass (Mt) can be derived so that the down-gradient, steady-state concentrations will be less than the MPC of the specific compound, viz., Mt = V(MPC x 1000) t
[17.2.2]
where: Mt V MPC t
the threshold mass in g/hectare the volume of leachate entering the aquifer in L/yr/hectare the maximum permissible concentration as g/L, and time in years; the amount of time between liner breakthrough and when the predicted concentration of the compound in the compliance well equals its MPC.
Using this estimation technique, Roy et al.1 estimated mass limitations for the compounds that exceeded their MPCs in the simulations. They found that benzene, carbon tetrachloride, dichloromethane, pyridine, tetrachloroethylene, 1,1,1-trichloroethylene, trichloroethylene and all chlorinated fluorocarbons would require strict mass limitations (<250 kg/ha). Other solvents could be safely landfilled at the site without mass restrictions: acetone, chlorobenzene, cresols, o-dichlorobenzene, diethyl ether, ethyl acetate, ethylbenzene, methanol, methyl ethyl ketone, methyl isobutyl ketone, nitrobenzene, toluene, and xylene. Some solvents (cyclohexanone, n-butyl alcohol, isobutyl alcohol, and carbon disulfide) would require some restrictions to keep the attenuation capacity of the site from being exceeded. These studies,1-3 demonstrated that the land disposal of wastes containing some organic solvents at sites using best-available liner technology may be environmentally acceptable. Wastes that contain chlorinated hydrocarbons, however, may require pretreatment such as incineration or stabilization before land disposal. If the mass-loading rate is controlled and the attenuation capacity of the site is carefully studied, the integrated and multidisciplinary approach outlined in this section can be applied to the management of solvent-containing wastes.
17.3 Environmental fate of glycol ethers
1169
REFERENCES 1 2 3 4 5
W. R. Roy, R. A. Griffin, J. K. Mitchell, and R. A. Mitchell, Environ. Geol. Water Sci., 13, 225 (1989). W. R. Roy and R. A. Griffin, J. Haz. Mat., 15, 365 (1987). R. A. Griffin and W. R. Roy. Feasibility of land disposal of organic solvents: Preliminary Assessment. Environmental Institute for Waste Management Studies, Report No. 10, University of Alabama, 1986. W. R. Roy and R. A. Griffin, Environ. Geol. Water Sci., 7, 241 (1985). W. J. George and P. D. Siegel. Assessment of recommended concentrations of selected organic solvents in drinking water. Environmental Institute for Waste Management Studies, Report No. 15, University of Alabama, 1988.
17.3 ENVIRONMENTAL FATE AND ECOTOXICOLOGICAL EFFECTS OF GLYCOL ETHERS James Devillers CTIS, Rillieux La Pape, France
Aurélie Chezeau, André Cicolella, and Eric Thybaud INERIS, Verneuil-en-Halatte, France
17.3.1 INTRODUCTION Glycol ethers and their acetates are widely used as solvents in the chemical, painting, printing, mining and furniture industries. They are employed in the production of paints, coatings, resins, inks, dyes, varnishes, lacquers, cleaning products, pesticides, deicing additives for gasoline and jet fuel, and so on.1 In 1997, the world production of glycol ethers was about 900,000 metric tons.2 There are two distinct series of glycol ethers namely the ethylene glycol ethers which are produced from ethylene oxide and the propylene glycol ethers derived from propylene oxide. The former series is more produced and used than the latter. Thus, inspection of the 42,000 chemical substances recorded by INRS (France) in the SEPIA data bank, between 1983 and 1998, reveals that 10% of them include ethylene glycol ethers and about 4% propylene glycol ethers.2 However, due to the reproductive toxicity of some ethylene glycol monoalkyl ethers,3-5 it is important to note that the worldwide tendency is to replace these chemicals by glycol ethers belonging to the propylenic series.2 Given the widespread use of glycol ethers, it is obvious that these chemicals enter the environment in substantial quantities. Thus, for example, the total releases to all environmental media in the United States for ethylene glycol monomethyl ether and ethylene glycol monoethyl ether in 1992 were 1688 and 496 metric tons, respectively.6 However, despite the potential hazard of these chemicals, the problems of the environmental contaminations with glycol ethers have not received much attention. There are two main reasons for this. First, these chemicals are not classified as priority pollutants, and hence, their occurrence in the different compartments of the environment is not systematically investigated. Thus, for example, there are no glycol ethers on the target list for the Superfund hazardous waste site cleanup program.6 Second, glycol ethers are moderately volatile colorless liquids with a high water solubility and a high solubility with numerous solvents. Consequently, the clas-
1170
J Devillers, A Chezeau, A Cicolella, E Thybaud
sical analytical methods routinely used for detecting the environmental pollutants do not provide reliable results with the glycol ethers, especially in the aquatic environments. Under these conditions, the aim of this chapter is to review the available literature on the occurrence, environmental fate, and ecotoxicity of glycol ethers. 17.3.2 OCCURRENCE Despite the poor applicability of the most widely used USEPA analytical methods, some ethylene, diethylene, and triethylene glycol ethers have been reported as present in Superfund hazardous waste sites in the US more often than some of the so-called priority pollutants.6 More specifically, Eckel and co-workers6 indicated that in Jacksonville (Florida), a landfill received a mixture of household waste and wastes from aircraft maintenance and paint stripping from 1968 to 1970. In 1984, sampling of residential wells in the vicinity revealed concentrations of 0.200, 0.050, and 0.010 mg/l of diethylene glycol diethyl ether, ethylene glycol monobutyl ether, and diethylene glycol monobutyl ether, respectively. One year later, concentrations of 0.050 to 0.100 mg/l of diethylene glycol diethyl ether were found in the most contaminated portion of the site. In 1989, some samples still indicated the presence of diethylene glycol diethyl ether and triethylene glycol dimethyl ether. This case study clearly illustrates that glycol ethers may persist in the environment for many years after a contamination. Concentrations of 0.012 to 0.500 mg/l of ethylene glycol monobutyl ether were also estimated in residential wells on properties near a factory (Union Chemical, Maine, USA) manufacturing furniture stripper containing N,N-dimethylformamide. In addition, in one soil sample located in that site, a concentration of 0.200 mg/kg of ethylene glycol monobutyl ether was also found. In another case study, Eckel and co-workers6 showed that ethylene glycol diethyl ether was detected with estimated concentrations in the range from 0.002 to 0.031 mg/l in eight residential wells adjacent to a landfill (Ohio) receiving a mixture of municipal waste and various industrial wastes, many of them from the rubber industry. Last, ethylene glycol monomethyl ether was detected in ground-water samples at concentrations of 30 to 42 mg/l (Winthrop landfill, Maine).6 In 1991, the high resolution capillary GC-MS analysis of a municipal wastewater collected from the influent of the Asnières-sur-Oise treatment plant located in northern suburban Paris (France) revealed the presence of ethylene glycol monobutyl ether (0.035 mg/l), diethylene glycol monobutyl ether (0.015 mg/l), propylene glycol monomethyl ether (0.070 mg/l), dipropylene glycol monomethyl ether (0.050 mg/l), and tripropylene glycol monoethyl ether (<0.001 mg/l).7 In the Hayashida River (Japan) mainly polluted by effluents from leather factories, among the pollutants separated by vacuum distillation and identified by GS-MS, ethylene glycol monobutyl ether, ethylene glycol monoethyl ether, and diethylene glycol monobutyl ether were found at concentrations of 5.68, 1.20, and 0.24 mg/l, respectively.8 In air samples collected in the pine forest area of Storkow (30 km south east of Berlin, Germany), the “Mediterranean Macchia” of Castel Porziano (Italy), and the Italian station located at the foot of Everest (Nepal), GC-MS analysis showed concentrations of ethylene glycol monobutyl ether of 1.25, 0.40, and 0.10 to 1.59 µg/m3, respectively.9
17.3 Environmental fate of glycol ethers
1171
17.3.3 ENVIRONMENTAL BEHAVIOR Due to their high water solubilities and low 1-octanol/water partition coefficients (log P), the glycol ethers, after release in the environment, will be preferentially found in the aquatic media and their accumulation in soils, sediments, and biota will be negligible. The available literature data on the biodegradation of glycol ethers reveal that most of these chemicals are biodegradable under aerobic conditions (Table 17.3.1), suggesting the compounds would not likely persist. Table 17.3.1. Aerobic biodegradation of glycol ethers and their acetates in aquatic environments and soils Name [CAS RN], Values
Comments
Ref.
5 d = 30% bio-oxidation, 10 d = 62%, 15 d = 74%, 20 d = 88%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water
10
5 d = 6% bio-oxidation, 10 d = 18%, 15 d = 23%, 20 d = 39%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water
10
ThOD = 1.68 g/g, BOD5 = 0.12 g/g, %ThOD = 7, COD = 1.69 g/g, %ThOD = 101
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
BOD5 = 0.50 g/g, %ThOD = 30
same conditions, adapted seed
11
Ethylene glycol monomethyl ether [109-86-4]
Ethylene glycol monomethyl ether acetate [110-49-6] ThOD = 1.63 g/g, BOD5 = 0.49 g/g, %ThOD = 30, COD = 1.60 g/g, %ThOD = 98
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
Ethylene glycol monoethyl ether [110-80-5] 5 d = 36% bio-oxidation, 10 d = 88%, 15 d = 92%, 20 d = 100%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water
10
5 d = 5% bio-oxidation, 10 d = 42%, 15 d = 50%, 20 d = 62%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water
10
ThOD = 1.96 g/g, BOD5 = 1.03 g/g, %ThOD = 53, COD = 1.92 g/g, %ThOD = 98
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
BOD5 = 1.27 g/g, %ThOD = 65
same conditions, adapted seed
11
COD removed = 91.7%
mixed culture, acclimation in a semi-continuous system
12
BOD5 = 0.353 g/g, BOD/ThOD = 18.1%
fresh water, standard dilution method, 10 mg/l
13
BOD5 = 0.021 g/g, BOD/ThOD = 1.1%
sea water, 10 mg/l
13
Ethylene glycol monoethyl ether acetate [111-15-9] 5 d = 36% bio-oxidation, 10 d = 79%, 15 d = 82%, 20 d = 80%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water
10
5 d = 10% bio-oxidation, 10 d = 44%, 15 d = 59%, 20 d = 69%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water
10
1172
J Devillers, A Chezeau, A Cicolella, E Thybaud
Name [CAS RN], Values
Comments
Ref.
ThOD = 1.82 g/g, BOD5 = 0.74 g/g, %ThOD = 41, COD = 1.76 g/g, %ThOD = 96
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
BOD5 = 0.442 g/g, BOD/ThOD = 24.3%
fresh water, standard dilution method, 10 mg/l
13
BOD5 = 0.448 g/g, BOD/ThOD = 24.7%
sea water, 10 mg/l
13
Ethylene glycol diethyl ether [629-14-1] ThOD = 2.31 g/g, BOD10 = 0.10 g/g
14
Ethylene glycol monoisopropyl ether [109-59-1] ThOD = 2.15 g/g, BOD5 = 0.18 g/g, %ThOD = 8, COD = 2.08 g/g, %ThOD = 97
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
Ethylene glycol monobutyl ether [111-76-2] 5 d = 26% bio-oxidation, 10 d = 74%, 15 d = 82%, 20 d = 88%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water
10
5 d = 29% bio-oxidation, 10 d = 64%, 15 d = 70%, 20 d = 75%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water
10
ThOD = 2.31 g/g, BOD5 = 0.71 g/g, %ThOD = 31, COD = 2.20 g/g, %ThOD = 95
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
BOD5 = 1.68 g/g, %ThOD = 73
same conditions, adapted seed
11
COD removed = 95.2%
mixed culture, acclimation in a semi-continuous system
12
BOD5 = 0.240 g/g, BOD/ThOD = 10.4%
fresh water, standard dilution method, 10 mg/l
13
BOD5 = 0.044 g/g, BOD/ThOD = 1.9%
sea water, 10 mg/l
13
5 d = 47% bio-oxidation, 15 d = 70%, 28 d = 75%
OECD Closed-Bottle test 301D, no prior acclimation/adaptation of seed, 20°C
15
28 d = 95% removal of DOC
OECD 301E, 10 mg/l, non-adapted domestic activated sludge
16
28 d = 100% removal
OECD 302B, 500 mg/l, non-adapted domestic activated sludge
16
1 d = 22% removal, 3 d = 63%, 5 d = 100%
OECD 302B, 450 mg/l, non-adapted domestic activated sludge
16
14 d = 96% BOD of ThOD
100 mg/l, activated sludge
16
Ethylene glycol monobutyl ether acetate [112-07-2] >90% DOC removal in 28 d
OECD 302B, Zahn-Wellens test
16
Diethylene glycol monomethyl ether [111-77-3] effluent from a biological sanitary waste treatment ThOD = 1.73 g/g, BOD5 = 0.12 g/g, plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, %ThOD = 7, COD = 1.71 g/g, %ThOD = 99 COD = ASTM D 1252-67
11
17.3 Environmental fate of glycol ethers
1173
Name [CAS RN], Values BOD5 = 0.95 mmol/mmol
Comments
Ref.
acclimated mixed culture, estimated by linear regression technique from a 20-day test
17
Diethylene glycol monomethyl ether acetate [629-38-9] ThOD = 1.68 g/g, BOD10 = 1.10 g/g
14
Diethylene glycol monoethyl ether [111-90-0] 5 d = 17% bio-oxidation, 10 d = 71%, 15 d = 75%, 20 d = 87%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water
10
5 d = 11% bio-oxidation, 10 d = 44%, 15 d = 57%, 20 d = 70%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water
10
ThOD = 1.91 g/g, BOD5 = 0.20 g/g, %ThOD = 11, COD = 1.85 g/g, %ThOD = 97
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
BOD5 = 0.58 g/g, %ThOD = 30
same conditions, adapted seed
11
COD removed = 95.4%
mixed culture, acclimation in a semi-continuous system
12
BOD5 = 5.50 mmol/mmol
acclimated mixed culture
17
Diethylene glycol diethyl ether [112-36-7] ThOD = 2.17 g/g, BOD10 = 0.10 g/g
14
Diethylene glycol monobutyl ether [112-34-5] ThOD = 2.17 g/g, BOD5 = 0.25 g/g, %ThOD = 11, COD = 2.08 g/g, %ThOD = 96
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
COD removed = 95.3%
mixed culture, acclimation in a semi-continuous system
12
BOD5 = 5.95 mmol/mmol
acclimated mixed culture
17
5 d = 27% bio-oxidation, 10 d = 60%, 15 d = 78%, 20 d = 81%
BOD APHA SM, no prior acclimation/adaptation of seed, 20°C
15
5 d = 3% bio-oxidation, 15 d = 70%, 28 d = 88%,
OECD Closed-Bottle test 301D, no prior acclimation/adaptation of seed, 20°C
15
28 d >60% removal
OECD 301C, modified MITI test, adapted activated sludge
16
28 d = 58% removal
OECD 301C, modified MITI test, adapted activated sludge
16
1 d = 14% removal, 3 d = 19%, 5 d = 60%, 6 d = 100%
OECD 302B, modified Zahn-Wellens test, industrial non-adapted activated sludge
16
9 d = 100% removal
OECD 302B, modified Zahn-Wellens test, non-adapted activated sludge
16
14 d = 94% removal
OECD 301A, domestic secondary effluent sewage, DOC measured
16
1174
J Devillers, A Chezeau, A Cicolella, E Thybaud
Name [CAS RN], Values
Comments
BOD5 = 0.05 g/g (5.2% ThOD), BOD10 = 0.39 g/g (57% ThOD), BOD20 = 1.08 g/g (72% ThOD) BOD = 0.25 g/g, COD = 2.08 g/g
Ref. 16
Dutch standard method, adapted sewage
16
Diethylene glycol monobutyl ether acetate [124-17-4] BOD5 = 13.3% ThOD, BOD10 = 18.4%, BOD15 = 24.6%, BOD20 = 67%
16
Triethylene glycol monoethyl ether [112-50-5] 5 d = 8% bio-oxidation, 10 d = 47%, 15 d = 63%, 20 d = 71%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water
10
5 d = 1% bio-oxidation, 10 d = 10%, 15 d = 12%, 20 d = 22%
filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water
10
ThOD = 1.89 g/g, BOD5 = 0.05 g/g, %ThOD = 3, COD = 1.84 g/g, %ThOD = 97
effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67
11
COD removed = 96.6%
mixed culture, acclimation in a semi-continuous system
12
BOD5 = 1.15 mmol/mmol
acclimated mixed culture, estimated by linear regression technique from a 20-day test
17
Triethylene glycol monobutyl ether [143-22-6] COD removed = 96.3%
mixed culture, acclimation in a semi-continuous system
12
Propylene glycol monomethyl ether [107-98-2] 50% removal: <1 d (0.2 ppm), <2 d (9.9 ppm), <5 d (100 ppm)
sand = 72%, silt = 16%, clay = 12%, OC = 2.5%, 9.9 x 106 bacteria/g of soil
18
50% removal: <1 d (0.4 ppm), <7 d (100 ppm)
sand = 74%, silt = 12%, clay = 14%, OC = 2.0%, 5.1 x 106 bacteria/g of soil
18
50% removal: <4 d (0.4 ppm), >56 d (100 ppm), <23 d (100 ppm with nutrients)
sand = 94%, silt = 4%, clay = 2%, OC = 0.4%, 9.3 x 105 bacteria/g of soil
18
BOD20 = 58% of ThOD
18
Propylene glycol monomethyl ether acetate [108-65-6] 50% removal: <1 d (2.5 ppm), <1 d (20 ppm),
sand = 72%, silt = 16%, clay = 12%, OC = 2.5%, 9.9 x 106 bacteria/g of soil
18
50% removal: <1 d (20 ppm)
sand = 74%, silt = 12%, clay = 14%, OC = 2.0%, 5.1 x 106 bacteria/g of soil
18
50% removal: <1 d (20 ppm)
sand = 94%, silt = 4%, clay = 2%, OC = 0.4%, 9.3 x 105 bacteria/g of soil
18
BOD20 = 62% of ThOD
18
17.3 Environmental fate of glycol ethers
Name [CAS RN], Values
1175
Comments
Ref.
Propylene glycol monophenyl ether [770-35-4] 50% removal: <1 d (1.4 ppm), <2 d (10 sand = 72%, silt = 16%, clay = 12%, OC = 2.5%, 9.9 x ppm), <5 d (107 ppm) 106 bacteria/g of soil
18
50% removal: <1 d (1.5 ppm), <7 d (104 sand = 74%, silt = 12%, clay = 14%, OC = 2.0%, 5.1 x ppm) 106 bacteria/g of soil
18
50% removal: <5 d (1.5 ppm), <23 d (108 sand = 94%, silt = 4%, clay = 2%, OC = 0.4%, 9.3 x 105 bacteria/g of soil ppm)
18
BOD20 = 52% of ThOD
18
Note: ThOD = theoretical oxygen demand or the weight ratio of oxygen required per mg of compound for complete conversion of the compound to dioxide and water; BOD = biochemical oxygen demand; COD = chemical oxygen demand; DOC = dissolved organic carbon.
Metabolism pathways involving oxidation of the alcohol functionality and cleavage of the ether bond have been proposed for a reduced number of glycol ethers.18-20 While glycol ethers can be considered as biodegradable under aerobic conditions, Eckel et al.6 have stressed that under anaerobic conditions, such as in the groundwater plume emanating from a landfill, these chemicals may persist for many years. In the same way, due to their physico-chemical properties, glycol ethers can act as cosolvents in mixtures with highly hydrophobic contaminants enhancing the solubility, mobility, and hence, the ecotoxicity of these chemicals. Abiotic degradation processes for organic chemicals include aqueous photolysis, hydrolysis, and atmospheric photooxidation. The primary abiotic degradation process affecting glycol ethers is atmospheric photooxidation mediated by hydroxyl (OH) radicals formed in the atmosphere.16 Photooxidation of glycol ethers is generally estimated from quantitative structure-property (QSPR) models due to the scarcity of experimental data.21-23 Thus, for example, Grosjean23 estimated atmospheric half-lives of 2 to 20 hours for ethylene glycol ethers (taking OH = 106 molecules cm-3). 17.3.4 ECOTOXICITY 17.3.4.1 Survival and growth The ecotoxicological effects of glycol ethers and their acetates have been measured on various organisms occupying different trophic levels in the environment. Most of the available data deal with lethality, immobilization of the organisms, inhibition of cell multiplication, or growth (Table 17.3.2).
1176
J Devillers, A Chezeau, A Cicolella, E Thybaud
Table 17.3.2. Effects of glycol ethers and their acetates on survival and growth of organisms Species
Results
Comments
Ref.
Ethylene glycol monomethyl ether [109-86-4]
Bacteria
Pseudomonas putida
16-h TGK >10000 mg/l
toxicity threshold, inhibition of cell multiplication
24
Pseudomonas aeruginosa
4-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria 3-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
8-d TGK = 100 mg/l
toxicity threshold, inhibition of cell multiplication
26
Blue-green Microcystis aeruginosa algae Algae
Scenedesmus quadricauda
8-d TGK >10000 mg/l
toxicity threshold, inhibition of cell multiplication
26
Yeasts
Candida sp.
4-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
4-m biocidal = 10-17% 42-d NG = 20%
tested in jet fuel and water mixtures NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
25 27
Gliomastix sp.
4-m biocidal = 17-25%
tested in jet fuel and water mixtures
25
Chilomonas paramaecium
48-h TGK = 2.2 mg/l
toxicity threshold, inhibition of cell multiplication
28
Uronema parduczi
20-h TGK > 10000 mg/l
toxicity threshold, inhibition of cell multiplication
29
Entosiphon sulcatum
72-h TGK = 1715 mg/l
toxicity threshold, inhibition of cell multiplication
30
Hydra vulgaris (syn. H. attenuata)
72-h LC50 = 29000 mg/l semi-static, adult polyps
31
Daphnia magna
24-h LC50 >10000 mg/l static, nominal concentrations 24-h EC50 >10000 mg/l static, nominal concentrations, immobilization
32 33
Artemia salina
24-h TLm >10000 mg/l
static, 24.5°C
10
Carassius auratus
24-h TLm >5000 mg/l
static, 20 ± 1°C, measured concentrations
34
Leuciscus idus melanotus
48-h LC50 >10000 mg/l 48-h LC0 >10000 mg/l static (Juhnke) 48-h LC100 >10000 mg/l
35
Oryzias latipes
24-h LC50 >1000 mg/l 48-h LC50 >1000 mg/l
36
Fungi
Protozoa
Coelenterates
Crustacea
Fish
Cladosporium resinae
static, same results at 10, 20, and 30°C
17.3 Environmental fate of glycol ethers
Species
Results
Comments
Ref.
Pimephales promelas
20% mortality = 50 mg/l, 0% mortality = 100 mg/l, 30% mortality = 200 mg/l, 70% mortality = 500 static with daily renewal, adults, mg/l, 7-d of exposure 70% mortality = 700 mg/l, 100% mortality = 1000 mg/l
Poecilia reticulata
7-d LC50 = 17434 mg/l
semi-static, 22 ± 1°C, rounded value
38
Lepomis macrochirus
96-h LC50 >10000 mg/l
static, nominal concentrations, L = 33-75 mm
39
Menidia beryllina
96-h LC50 >10000 mg/l
static, sea water, nominal concentrations, L = 40-100 mm
39
Xenopus laevis
0% mortality=2500 mg/l, 0% mortality = static with daily renewal, adults, 4-d 7500 mg/l, 0% mortality of exposure = 15000 mg/l
37
Fish
Amphibia
1177
37
Ethylene glycol monomethyl ether acetate [110-49-6]
Fish
Carassius auratus
24-h TLm = 190 mg/l
static, 20 ± 1°C, measured concentrations
34
Lepomis macrochirus
96-h LC50 = 45 mg/l
static, nominal concentrations, L = 33-75 mm
39
Menidia beryllina
96-h LC50 = 40 mg/l
static, sea water, nominal concentrations, L = 40-100 mm
39
Pseudomonas aeruginosa 4-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria 3-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
Ethylene glycol monoethyl ether [110-80-5]
Bacteria
Yeasts
Fungi
Vibrio fischeri
5-min EC50 = 376 mg/l, 15-min EC50 = 403 reduction in light output, 15°C mg/l, 30-min EC50 = 431 mg/l
40
Candida sp.
4-m biocidal = 2-5%
tested in jet fuel and water mixtures
25
4-m biocidal = 5-10% 42-d NG = 10%
tested in jet fuel and water mixtures NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
25 27
4-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Cladosporium resinae
Gliomastix sp.
1178
J Devillers, A Chezeau, A Cicolella, E Thybaud
Species Coelenterates
Crustacea
Results
Comments
Ref.
Hydra vulgaris (syn. H. attenuata)
72-h LC50 = 2300 mg/l
41
Daphnia magna
24-h EC50 = 54000 mg/l static, nominal concentrations, immobilization 48-h EC50 = 7671 mg/l static, 22 ± 1°C, nominal concentrations, immobilization
Artemia salina
24-h TLm >10000 mg/l
static, 24.5°C
10
Carassius auratus
24-h TLm >5000 mg/l
static, 20 ± 1°C, measured concentrations
34
Poecilia reticulata
7-d LC50 = 16399 mg/l
semi-static, 22 ± 1°C, rounded value
38
Lepomis macrochirus
96-h LC50 >10000 mg/l
static, nominal concentrations, L = 33-75 mm
39
Menidia beryllina
96-h LC50 >10000 mg/l
static, sea water, nominal concentrations, L = 40-100 mm
39
semi-static, adult polyps
31
42
Fish
Ethylene glycol monoethyl ether acetate [111-15-9] Crustacea
Artemia salina
24-h TLm = 4000 mg/l
static, 24.5°C
10
Fish
Carassius auratus
24-h TLm = 160 mg/l
static, 20 ± 1°C, measured concentrations
34
Pseudomonas aeruginosa 4-m biocidal = 0-2%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria 3-m biocidal = 0-2%
tested in jet fuel and water mixtures
25
Candida sp.
4-m biocidal = 2-5%
tested in jet fuel and water mixtures
25
Cladosporium resinae
4-m biocidal = 2-5%
tested in jet fuel and water mixtures
25
Gliomastix sp.
4-m biocidal = 2-5%
tested in jet fuel and water mixtures
25
Ethylene glycol monopropyl ether [2807-30-9] Bacteria Yeasts Fungi
Ethylene glycol monoisopropyl ether [109-59-1] Fungi
Cladosporium resinae
42-d NG = 10%
NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
27
Carassius auratus
24-h TLm >5000 mg/l
static, 20 ± 1°C, measured concentrations
34
Poecilia reticulata
7-d LC50 = 5466 mg/l
semi-static, 22 ± 1°C, rounded value
38
Fish
Ethylene glycol monobutyl ether [111-76-2]
Bacteria
Pseudomonas putida
16-h TGK = 700 mg/l
toxicity threshold, inhibition of cell multiplication
24
Pseudomonas aeruginosa
4-m biocidal = 1-2%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria
3-m biocidal = 1-2%
tested in jet fuel and water mixtures
25
17.3 Environmental fate of glycol ethers
Species
Comments
Ref.
8-d TGK = 35 mg/l
toxicity threshold, inhibition of cell multiplication
26
Scenedesmus quadricauda
8-d TGK = 900 mg/l
toxicity threshold, inhibition of cell multiplication
26
Selenastrum capricornutum
7-d EC50 >1000 mg/l, NOEC = 125 mg/l, LOEC = 250 mg/l
growth rate inhibition
16
Candida sp.
4-m biocidal = 2-3%
tested in jet fuel and water mixtures
25
4-m biocidal = 2-3% 42-d NG = 5%
tested in jet fuel and water mixtures NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
25 27
Gliomastix sp.
4-m biocidal = 2-3%
tested in jet fuel and water mixtures
25
Chilomonas paramaecium
48-h TGK = 911 mg/l
toxicity threshold, inhibition of cell multiplication
28
Uronema parduczi
20-h TGK = 463 mg/l
toxicity threshold, inhibition of cell multiplication
29
Entosiphon sulcatum
72-h TGK = 91 mg/l
toxicity threshold, inhibition of cell multiplication
30
Hydra vulgaris (syn. H. attenuata)
72-h LC50 = 690 mg/l
semi-static, adult polyps
31
Crassostrea virginica
24-h LC50 = 181 mg/l (143 - 228), 48-h LC50 = 160 mg/l (125 204), 72-h LC50 = 114 mg/l (93.9 - 138), 96-h LC50 = 89.4 mg/l (72 - 110)
static, 22 ± 1°C, 10 organisms/tank
43
static, nominal concentrations
32
Daphnia magna
24-h LC50 = 1720 mg/l, 24-h LC0 = 1140 mg/l, 24-h LC100 = 2500 mg/l 24-h EC50 = 1815 mg/l (1698-1940), 24-h EC0 = 1283 mg/l, 24-h EC100 = 2500 mg/l 48-h EC50 = 835 mg/l
static, nominal concentrations, immobilization
33
static, immobilization
16
24-h TLm = 1000 mg/l
static, 24.5°C
10
Blue-green Microcystis aeruginosa algae
Algae
Yeasts
Fungi
Protozoa
Coelenterates
Mollusca
Crustacea
1179
Cladosporium resinae
Artemia salina
Results
1180
J Devillers, A Chezeau, A Cicolella, E Thybaud
Species
Crustacea
Results
Comments
Ref.
Panaeus setiferus
24-h LC50 > 430 mg/l, 48-h LC50 = 173 mg/l (123 - 242), 72-h LC50 = 147 mg/l static, average size = 1.5 cm (116 - 186), 96-h LC50 = 130 mg/l (104 - 162)
Carassius auratus
24-h TLm = 1700 mg/l
Leuciscus idus melanotus
48-h LC50 = 1575 mg/l, static (Juhnke) 48-h LC0 = 1350 mg/l 48-h LC100 = 1620 mg/l 48-h LC50 = 1395 mg/l, static (Lüdemann) 48-h LC0 = 1170 mg/l, 48-h LC100 = 1490 mg/l
35
Pimephales promelas
96-h LC50 = 2137 mg/l
static
16
Poecilia reticulata
7-d LC50 = 983 mg/l
semi-static, 22 ± 1°C, rounded value
38
Notropis atherinoides
72-h LC50 >500 mg/l
static
16
96-h LC50 = 1490 mg/l
static, nominal concentrations, L = 33-75 mm
39
Fish Lepomis macrochirus
Cyprinodon variegatus
static, 20 ± 1°C, measured concentrations
24-h LC50 = 2950 mg/l, 96-h LC50 = 2950 mg/l 24-h LC50 (125 - 176), 48-h LC50 (107 - 147), 72-h LC50 (105 - 138), 96-h LC50 (100 - 133)
43
34
44
= 149 mg/l = 126 mg/l static, 22 ± 1°C
43
static, sea water, nominal concentrations, L = 40-100 mm
39
NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
27
4-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria 3-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Candida sp.
tested in jet fuel and water mixtures
25
Menidia beryllina
= 121 mg/l = 116 mg/l
96-h LC50 = 1250 mg/l
Ethylene glycol dibutyl ether [112-48-1] Fungi
Cladosporium resinae
42-d NG = 5%
Diethylene glycol monomethyl ether [111-77-3] Bacteria
Yeasts
Pseudomonas aeruginosa
4-m biocidal = 5-10%
17.3 Environmental fate of glycol ethers
Species
Algae
Fungi
Crustacea
Fish
1181
Results
Comments
Ref.
Scenedesmus subspicatus
72-h EC50 > 500 mg/l, 72-h EC20 > 500 mg/l, biomass 72-h EC90 > 500 mg/l
45
Selenastrum capricornutum
96-h EC50 > 1000 mg/l
biomass
45
4-m biocidal = 10-17% 42-d NG = 20%
tested in jet fuel and water mixtures NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
25 27
Gliomastix sp.
4-m biocidal > 25%
tested in jet fuel and water mixtures
25
Daphnia magna
24-h LC50 = 1495 mg/l (1300 - 1600), static, nominal concentrations 48-h LC50 = 1192 mg/l (1100 - 6500)
45
Oncorhynchus mykiss
96-h LC50 > 1000 mg/l
semi-static, nominal concentrations
45
Carassius auratus
24-h TLm >5000 mg/l
static, 20 ± 1°C, measured concentrations
34
Pimephales promelas
24-h LC50 = 6400 (6200 - 6600), 48-h LC50 = 6000 (6000 -6100), 72-h LC50 = 6000 (6000 -6100), 96-h LC50 = 5700 (5600 - 5900)
static
45
static, nominal concentrations, L = 33-75 mm
39
NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
27
Cladosporium resinae
Lepomis macrochirus
mg/l mg/l mg/l mg/l
96-h LC50 = 7500 mg/l
Diethylene glycol dimethyl ether [111-96-6] Fungi
Cladosporium resinae
42-d NG = 20%
Diethylene glycol monoethyl ether [111-90-0]
Bacteria
Vibrio fischeri
5-min EC50 = 1000 mg/l, reduction in light output, nominal 5-min EC50 = 1290 mg/l concentrations 15-min EC50 = 10954 mg/l (10592.8 - 11327.5) NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
Fungi
Cladosporium resinae
42-d NG = 20%
Coelenterates
Hydra vulgaris (syn. H. attenuata)
72-h LC50 = 17000 mg/l semi-static, adult polyps
46
47
27
31
1182
J Devillers, A Chezeau, A Cicolella, E Thybaud
Species
48
Orconectes immunis
96-h LC50 = 34700 mg/l flow-through, 15.5°C, 0.47 g, mea(29100 - 41400) sured concentrations
48
Artemia salina
24-h TLm >10000 mg/l
static, 24.5°C
10
Tanytarsus dissimilis
48-h LC50 = 18800 mg/l
static, 22.3°C, measured concentrations
48
Oncorhynchus mykiss
Amphibia
Ref.
Daphnia magna
EchinoderArbacia punctulata mata
Fish
Comments
48-h LC50 = 4670 mg/l static, 21.1°C, measured concentra(3620 - 6010), tions 48-h LC50 = 3340 mg/l static, 23.3°C, measured concentra(2120 - 5280) tions
Crustacea
Insecta
Results
4-h EC50 = 10661 mg/l static, marine, early embryo growth, (5576.5 - 20895.9) nominal concentrations 1-h EC50 = 3370 mg/l static, marine, sperm cell, nominal (3145.4 - 3610.8) concentrations 5-h EC50 = 4116 mg/l static, marine, increase in DNA, (3408 - 4907) nominal concentrations
47 47 49
96-h LC50 = 13400 mg/l flow-through, 14.9°C, 0.68 g, mea(11400 - 15700) sured concentrations
48
24-h TLm >5000 mg/l
34
Carassius auratus
static, 20 ± 1°C, measured concentrations 96-h LC50 = 20800 mg/l flow-through, 20.1°C, 0.92 g, mea(15700 - 27500) sured concentrations
Gambusia affinis
96-h LC50 = 15200 mg/l (12400 - 18700) 96-h LC50 = 12900 mg/l (11100 - 15000)
flow-through, 18.3°C, 0.23 g, measured concentrations flow-through, 19.9°C, 0.25 g, measured concentrations
48
Ictalurus punctatus
96-h LC50 = 6010 mg/l
flow-through, 17.3°C, 0.72 g, measured concentrations
48
Lepomis macrochirus
96-h LC50 >10000 mg/l static, nominal concentrations, L = 33-75 mm 96-h LC50 = 21400 mg/l flow-through, 18.3°C, 0.44 g, mea(19100 - 23900) sured concentrations
Pimephales promelas
96-h LC50 = 13900 mg/l (11600 - 16700) 96-h LC50 = 9650 mg/l (7910 - 11800)
flow-through, 24.8°C, 1.44 g, measured concentrations flow-through, 18.1°C, 0.35 g, measured concentrations
48
Menidia beryllina
96-h LC50 >10000 mg/l
static, sea water, nominal concentrations, L = 40-100 mm
39
Rana catesbeiana
96-h LC50 = 20900 mg/l flow-through, 17.2°C, 3.54 g, (19400 -22600) measured concentrations
48
39 48
48
Diethylene glycol diethyl ether [112-36-7] Fungi
Cladosporium resinae
42-d NG = 20%
NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
27
17.3 Environmental fate of glycol ethers
Species
1183
Results
Comments
Ref.
toxicity threshold, inhibition of cell multiplication growth inhibition, 25 ± 2°C
24
Diethylene glycol monobutyl ether [112-34-5] 16-h TGK = 255 mg/l Bacteria
Pseudomonas putida 16-h EC10 = 1170 mg/l
Blue-green Microcystis aeruginosa algae
Algae
Fungi
Protozoa
Crustacea
50
8-d TGK = 53 mg/l
toxicity threshold, inhibition of cell multiplication
26
Scenedesmus quadricauda
8-d TGK = 1000 mg/l
toxicity threshold, inhibition of cell multiplication
26
Scenedesmus subspicatus
96-h EC50 > 100 mg/l
limit test
50
Cladosporium resinae
42-d NG = 5%
NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
27
Chilomonas paramaecium
48-h TGK = 2774 mg/l
toxicity threshold, inhibition of cell multiplication
28
Uronema parduczi
20-h TGK = 420 mg/l
toxicity threshold, inhibition of cell multiplication
29
Entosiphon sulcatum
72-h TGK = 73 mg/l
toxicity threshold, inhibition of cell multiplication
30
static, nominal concentrations
32
static, nominal concentrations, immobilization
33
Daphnia magna
24-h LC50 = 2850 mg/l, 24-h LC0 = 1750 mg/l, 24-h LC100 = 3850 mg/l 24-h EC50 = 3200 mg/l (2990-3424), 24-h EC0 = 2333 mg/l, 24-h EC100 = 5000 mg/l 48-h EC50 >100 mg/l 48-h NOEC >100 mg/l
directive 84/449/ EEC, C2
16
static, 20 ± 1°C, measured concentrations
34
Carassius auratus
24-h TLm = 2700 mg/l
35
Leuciscus idus melanotus
48-h LC50 = 1805 mg/l, static (Juhnke) 48-h LC0 = 1140 mg/l, 48-h LC100 = 2185 mg/l 48-h LC50 = 2304 mg/l, static (Lüdemann) 48-h LC0 = 1820 mg/l, 48-h LC100 = 2400 mg/l 48-h LC50 = 2750 mg/l static
Poecilia reticulata
7-d LC50 = 1149 mg/l
semi-static, 22±1°C, rounded value
38
Lepomis macrochirus
96-h LC50 = 1300 mg/l
static, nominal concentrations, L = 33-75 mm
39
Menidia beryllina
96-h LC50 = 2000 mg/l
static, sea water, nominal concentrations, L = 40-100 mm
39
Fish
35
16
1184
J Devillers, A Chezeau, A Cicolella, E Thybaud
Species
Results
Comments
Ref.
Triethylene glycol monomethyl ether [112-35-6]
Bacteria
Yeasts Fungi
Pseudomonas aeruginosa
4-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria
3-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Candida sp.
4-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Cladosporium resinae
4-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Gliomastix sp.
4-m biocidal = 17-25%
tested in jet fuel and water mixtures
25
Triethylene glycol monoethyl ether [112-50-5] Pseudomonas aeruginosa
4-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
Sulfate-reducing bacteria
3-m biocidal = 2-5%
tested in jet fuel and water mixtures
25
Candida sp.
4-m biocidal = 5-10%
tested in jet fuel and water mixtures
25
Cladosporium resinae
4-m biocidal = 10-17%
tested in jet fuel and water mixtures
25
Gliomastix sp.
4-m biocidal = 17-25%
tested in jet fuel and water mixtures
25
Crustacea
Artemia salina
24-h TLm >10000 mg/l
static, 24.5°C
10
Fish
Carassius auratus
24-h TLm >5000 mg/l
static, 20 ± 1°C, measured concentrations
34
NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C
27
Bacteria
Yeasts Fungi
Propylene glycol monomethyl ether [107-98-2] Fungi
Cladosporium resinae
42-d NG = 20%
Propylene glycol monomethyl ether acetate [108-65-6] Fish
Lepomis macrochirus
24-h LC50 = 206 mg/l 96-h LC50 = 164 mg/l
44
It is difficult to draw definitive conclusions from the data listed in Table 17.3.2. Indeed, most of the data have been retrieved from rather old studies performed without GLP protocols. In addition, the toxicity values are generally based on nominal concentrations and the endpoints are different. However, from the data listed in Table 17.3.2, it appears that despite a difference of sensibility among species, glycol ethers do not present acute and subacute ecotoxicological effects to the majority of the tested organisms. However, it is interesting to note that the acetates seem to be more toxic that the corresponding parent compounds.10,34,39,51 In mammals, the acute toxicity of glycol ethers is also relatively low. The main target organs are the central nervous and haematopoitic systems. However, on the basis of the available data no significant difference exists between the acute toxicity of glycol ethers and their corresponding acetates.52
17.3 Environmental fate of glycol ethers
1185
17.3.4.2 Reproduction and development The reproductive and developmental toxicity of the ethylene glycol monomethyl and monoethyl ethers is well documented. Several longer-chain glycol ethers also have been investigated for their reproductive and developmental effects against rodents and rabbits.53-56 Conversely, there is a lack of information on the reproductive and developmental ecotoxicity of glycol ethers and their acetates. Bowden et al.31 have tested the teratogenic effects of four glycol ethers through their ability to inhibit the regeneration of isolated digestion regions of Hydra vulgaris (syn. H. attenuata). They have shown that the concentrations of ethylene glycol monomethyl, monoethyl, monobutyl, and diethylene glycol monoethyl ethers that were 50% inhibitory to regenerating digestive regions (IC50) after 72-h of exposure were 19,000, 1400, 540, and 19,000 mg/l, respectively. More specifically, at 10,000 mg/l of ethylene glycol monomethyl ether, the digestive regions regenerated the mouth and some tentacles. At 19,000 mg/l only tentacle buds were seen, while 38,000 mg/l produced disintegration of the coelenterates. Ethylene glycol monoethyl ether at 900 mg/l allowed the regeneration of the mouth, some tentacles and the basal disc. At 1900 mg/l four digestive regions showed wound healing while the remainder were dead. A concentration of 3700 mg/l was lethal to both polyps and digestive regions. At concentrations up to 370 mg/l of ethylene glycol monobutyl ether, digestive regions regenerated some tentacles and in some cases the basal disc. Normal wound healing only was observed at 740 mg/l while at 920 mg/l the wounds were healed but the region expanded. Last, the digestive regions at 10,000 mg/l of diethylene glycol monoethyl ether regenerated the mouth and some tentacles. At 20,000 mg/l only tentacle buds were seen while a concentration of 40,000 mg/l was lethal to both polyps and digestive regions. Using the LC50 (Table 17.3.2)/IC50 ratio as developmental hazard index, Bowden et al.31 ranked the four studied glycol ethers as follows: Ethylene glycol ethyl ether (1.7) > ethylene glycol monomethyl ether (1.5) > ethylene glycol monobutyl ether (1.3) > diethylene glycol monoethyl ether (0.9). Johnson et al.57,58 have also ranked glycol ethers according to the difference between their lowest concentrations overtly toxic to adults (A) and their lowest concentrations interfering with development (D) of the artificial embryos of reaggregated adult Hydra attenuata cells. The A/D ratios found by these authors were the following: Ethylene glycol monoethyl ether (5.0) > ethylene glycol monobutyl ether (4.4) > ethylene glycol monophenyl ether (3.3) > diethylene glycol dibutyl ether (2.3) > diethylene glycol monoethyl ether (2.2) > ethylene glycol monomethyl ether (1.3) > ethylene glycol monomethyl ether acetate (1.0) = ethylene glycol monoethyl ether acetate (1.0). Daston et al.37 have shown that A/D ratios were not constant across species and hence, there was no basis for using this parameter for developmental hazard assessment. Thus, for example, the A/D ratios calculated from the lowest observed effect levels (LOELs) of the ethylene glycol monomethyl ether were 8, >3, 0.5, and <0.3 for the mouse, Xenopus laevis, Pimephales promelas, and Drosophila melanogaster, respectively. If the A/D ratios were calculated from the NOELs (no observed effect levels), the values became >4, ≥6, 0.4, and ≤0.3 for the mammal, amphibian, fish, and insect, respectively. Teratogenicity of glycol ethers has been deeply investigated on the fruit fly, Drosophila melanogaster. Statistically significant increases in the incidence of wing notches and bent humeral bristles have been observed in Drosophila melanogaster exposed during development to ethylene glycol monomethyl ether (12.5, 15, 18, 22, and 25 mg/vial) and ethylene glycol monoethyl ether (54, 59, 65, 71, and 78 mg/vial).59 Wing notches, rare in control flies, were found in 13.8% of flies treated with ethylene glycol monomethyl ether
1186
J Devillers, A Chezeau, A Cicolella, E Thybaud
(7.5 µl/g).60 In general, male pupae are much more affected by ethylene glycol monomethyl ether than female pupae. However, teratogenicity appears strain dependent. Higher detoxification occurs with increased alcohol dehydrogenase (ADH) activity. Ethylene glycol monomethyl ether is much more toxic than its oxidation product, methoxyacetic acid, at the level of adult eclosion. Teratogenic effects were observed in an ADH-negative strain in spite of lacking ADH activity suggesting that apparently, ethylene glycol monomethyl ether is a teratogenic compound by itself against Drosophila melanogaster.61 Last, it is interesting to note that recently, Eisses has shown62 that administration of ethylene glycol monomethyl ether to larvae of fruit fly, containing the highly active alcohol dehydrogenase variant ADH-71k, exposed the mitotic germ cells and the mitotic somatic cells of the imaginal discs simultaneously to the mutagen methoxyacetaldehyde and the teratogen methoxyacetic acid, respectively. Consequently, the chances for specific gene mutations, though non-adaptive, were likely increased by a feedback mechanism. 17.3.5 CONCLUSION Despite their widespread use, glycol ethers and their acetates have received little attention as potential environmental contaminants. Based on their physico-chemical properties, they would tend to remain in the aquatic ecosystems where their bioconcentration, biomagnification and sorption onto sediments will appear negligible. Volatilization from water and hydrolysis or photolysis in the aquatic ecosystems are generally of minimal importance. Glycol ethers are also poorly sorbed to soil and their rapid removal in the atmosphere is expected. While glycol ethers are biodegradable under aerobic conditions, these chemicals may persist for many years under anaerobic conditions. Based on the available acute ecotoxicity data, glycol ethers and their acetates can be considered as practically non-toxic. However, there is a lack of information on their long-term effects on the biota. This is particularly annoying because the developmental toxicity of some of them has been clearly identified against mammals. Consequently, there is a need for studies dealing with the potential long-term effects of these chemicals against organisms occupying different trophic levels in the environment in order to see whether or not the classical methodological frameworks used for assessing the environmental risk of xenobiotics remain acceptable for this class of chemicals. 17.3.6 ACKNOWLEDGMENT This study was supported by the French Ministry of the Environment as part of the PNETOX program (1998). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13
R.J. Smialowicz, Occup. Hyg., 2, 269 (1996). Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 1-19. K. Nagano, E. Nakayama, M. Koyano, H. Oobayashi, H. Adachi, and T. Yamada, Jap. J. Ind. Health, 21, 29 (1979). Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 111-137. A. Cicolella, Cahiers de Notes Documentaires, 148, 359 (1992). W. Eckel, G. Foster, and B. Ross, Occup. Hyg., 2, 97 (1996). D.K. Nguyen, A. Bruchet, and P. Arpino, J. High Resol. Chrom., 17, 153 (1994). A. Yasuhara, H. Shiraishi, M. Tsuji, and T. Okuno, Environ. Sci. Technol., 15, 570 (1981). P. Ciccioli, E. Brancaleoni, A. Cecinato, R. Sparapani, and M. Frattoni, J. Chromatogr., 643, 55 (1993). K.S. Price, G.T. Waggy, and R.A. Conway, J.Water Pollut. Control Fed., 46, 63 (1974). A.L. Bridié, C.J.M. Wolff, and M. Winter, Water Res., 13, 627 (1979). T. Fuka, V. Sykora, and P. Pitter, Sci. Pap. Inst. Chem. Technol. Praze Technol. Water, F25, 203 (1983) (in Czech). S. Takemoto, Y. Kuge, and M. Nakamoto, Suishitsu Odaku Kenkyu, 4, 22 (1981) (in Japanese).
17.3 Environmental fate of glycol ethers
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62
1187
P. Pitter and J. Chudoba, Biodegradability of Organic Substances in the Aquatic Environment, CRC Press, Boca Raton, 1990. G.T. Waggy, R.A. Conway, J.L. Hansen, and R.L. Blessing, Environ. Toxicol. Chem., 13, 1277 (1994). C.A. Staples, R.J. Boatman, and M.L. Cano, Chemosphere, 36, 1585 (1998). L. Babeu and D.D. Vaishnav, J. Ind. Microbiol., 2, 107 (1987). S.J. Gonsior and R.J. West, Environ. Toxicol. Chem., 14, 1273 (1995). T. Harada and Y. Nagashima, J. Ferment. Technol., 53, 218 (1975). F. Kawai, Appl. Microbiol. Biotechnol., 44, 532 (1995). R. Atkinson, Int. J. Chem. Kinetics, 19, 799 (1987). P.H. Howard, R.S. Boethling, W.F. Jarvis, W.M. Meylan, and E.M. Michalenko, Handbook of Environmental Degradation Rates, CRC Press, Boca Raton, 1991. D. Grosjean, J. Air Waste Manage. Assoc., 40, 1397 (1990). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 10, 87 (1977). R.A. Neihof and C.A. Bailey, Appl. Environ. Microbiol., 35, 698 (1978). G. Bringmann and R. Kühn, Mitt. Internat. Verein. Limnol., 21, 275 (1978). K.H. Lee and H.A. Wong, Appl. Environ. Microbiol., 38, 24 (1979). G. Bringmann, R. Kühn, and A. Winter, Z. Wasser Abwasser Forsch., 13, 170 (1980). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 13, 26 (1980). G. Bringmann, Z. Wasser Abwasser Forsch., 11, 210 (1978). H.C. Bowden, O.K. Wilby, C.A. Botham, P.J. Adam, and F.W. Ross, Toxic. in Vitro, 9, 773 (1995). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 10, 161 (1977). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 15, 1 (1982). A.L. Bridié, C.J.M. Wolff, and M. Winter, Water Res., 13, 623 (1979). I. Juhnke and D. Lüdemann, Z. Wasser Abwasser Forsch., 11, 161 (1978). S. Tsuji, Y. Tonogai, Y. Ito, and S. Kanoh, Eisei Kagaku, 32, 46 (1986) (in Japanese). G.P. Daston, J.M. Rogers, D.J. Versteeg, T.D. Sabourin, D. Baines, and S.S. Marsh, Fund. Appl. Toxicol., 17, 696 (1991). H. Könemann, Toxicology, 19, 209 (1981). G.W. Dawson, A.L. Jennings, D. Drozdowski, and E. Rider, J. Hazard. Materials, 1, 303 (1975-1977). K.L.E. Kaiser and V.S. Palabrica, Water Poll. Res. J. Canada, 26, 361 (1991). IRCHA, Les Produits Chimiques dans l’Environnement. Registre des Données Normalisées de leurs Effets dans l’Environnement. Classeurs I et II et mises à jour, IRCHA, Vert-Le-Petit, 1981-1985. J. Hermens, H. Canton, P. Janssen, and R. de Jong, Aquat. Toxicol., 5, 143 (1984). Results from the MBA Laboratories, Houston, Texas (1984). W.B. Neely, Chemosphere, 13, 813 (1984). IUCLID Data Set, RIVM/ACT, Substance ID: 111-77-3, 19-Nov-98. C. Curtis, A. Lima, S.J. Lozano, and G.D. Veith, in Aquatic Toxicology and Hazard Assessment. Fifth Conference. ASTM STP 766, J.G. Pearson, R.B. Foster, and W.E. Bishop, Eds., American Society for Testing and Materials, Philadelphia, pp. 170-178. D. Nacci, E. Jackim, and R. Walsh, Environ. Toxicol. Chem., 5, 521 (1986). R.V. Thurston, T.A. Gilfoil, E.L. Meyn, R.K. Zajdel, T.I. Aoki, and G.D. Veith, Water Res., 19, 1145 (1985). E. Jackim and D. Nacci, Environ. Toxicol. Chem. 5, 561 (1986). IUCLID Data Set, RIVM/ACT, Substance ID: 112-34-5, 19-Nov-98. R.B. Sleet, Toxicologist, 11, 296 (1991). Anonymous in Ethers de Glycols. Quels Risques pour la Santé, INSERM, Paris, 1999, pp. 51-67. C.A. Kimmel, Occup. Hyg., 2, 131 (1996). A. Cicolella, Santé Publique, 2, 157 (1997). Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 111-137. Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 139-162. E.M. Johnson, B.E.G. Gabel, and J. Larson, Environ. Health Perspect., 57, 135 (1984). E.M. Johnson, L.M. Newman, B.E.G. Gabel, T.F. Boerner, and L.A. Dansky, J. Am. Coll. Toxicol., 7, 111 (1988). D. Lynch and M. Toraason, Occup. Hyg., 2, 171 (1996). R.L. Schuler, B.D. Hardin, and R.W. Niemeier, Teratogenesis Carcinog. Mutagen., 2, 293 (1982). K.T. Eisses, Teratogenesis Carcinog. Mutagen., 9, 315 (1989). K.T. Eisses, Teratogenesis Carcinog. Mutagen., 19, 183 (1999).
1188
Michelle Bergin, Armistead Russell
17.4 ORGANIC SOLVENT IMPACTS ON TROPOSPHERIC AIR POLLUTION Michelle Bergin and Armistead Russell Georgia Institute of Technology, Atlanta, Georgia, USA
17.4.1 SOURCES AND IMPACTS OF VOLATILE SOLVENTS Solvents, either by design or default, are often emitted in to the air, and the total mass of emissions of solvents is not small. In a typical city in the United States, solvents can rival automobile exhaust as the largest source category of volatilized organic compound (VOC) emissions into the atmosphere.1 In the United Kingdom, solvent usage accounted for 36% of the estimated total VOC mass emissions in 1995.2 Such widespread emissions leads to increased concentrations of many different compounds in the ambient environment, and their release has diverse impacts on air quality. A large variety of solvent-associated compounds are emitted, many of which are hydrocarbons, oxygenates. Those solvents may have multiple atmospheric impacts. For example, toluene is potentially toxic and can reach relatively high concentrations at small spatial scales, such as in a workplace. Toluene also contributes to the formation of tropospheric ozone at urban scales, while at regional scales toluene can lower the rate of tropospheric ozone formation. Other solvents likewise can have a range of impacts, ranging from local contamination to modification of the global climate system. This diversity of potential impacts is due, in part, to differences in the chemical properties and reactions that a compound may undergo in the atmosphere, differences in emissions patterns, and differences in the spatial and temporal scales of atmospheric phenomena. Transport and fate of chemical species is closely tied to the speed at which the compound degrades (from seconds to centuries, depending on the compound) as well as to the environmental conditions in which the compound is emitted. If a compound degrades very quickly, it may still have toxic effects near a source where concentrations can be high. In contrast, extremely stable compounds (such as chlorofluorocarbons; CFCs) are able to circumvent the globe, gradually accumulating to non-negligible concentrations.3 Of the myriad of solvents emitted into the air, the ones of primary concern are those with the greatest emissions rates, and/or those to which the environment has a high sensitivity. Compounds with very large emissions rates include tri- and tetrachloroethylene (e.g., from dry-cleaning), aromatics (benzene, toluene and xylenes, e.g., from coatings), alcohols, acetone and, historically, CFCs. While those compounds are often emitted from solvent use, other applications lead to their emission as well. For example, gasoline is rich in aromatics and alkanes, and in many cases fuel use dominates emissions of those compounds. CFCs have been used as refrigerants and as blowing agents. This diversity of originating sources makes identifying the relative contribution of solvents to air quality somewhat difficult since there are large uncertainties in our ability to quantify emissions rates from various source categories. Solvents with a high environmental sensitivity include benzene (a potent carcinogen), xylenes (which are very effective at producing ozone), formaldehyde (both toxic and a strong ozone precursor), and CFCs (ozone depleters and potential greenhouse gases). Most of the solvents of concern in terms of impacting ambient air are organic, either hydrocarbons, oxygenated organics (e.g., ethers, alcohols and ketones) or halogenated organics (e.g., dichlorobenzene). Some roles of these compounds in the atmosphere are discussed below.
17.4 Organic solvent impacts on tropospheric air pollution
1189
While the toxicity of some solvents is uncertain, the role of emissions on direct exposure is not in question. Indoors, vaporized solvents can accumulate to levels of concern for acute and/or chronic exposure. However, the toxicity of solvents outdoors is not typically of as great of concern as indoors except very near sources. Outdoors, solvents have adverse effects other than toxicity. The importance of CFC emissions on stratospheric ozone, for example, is significant, but the problem is well understood and measures are in place to alleviate the problem. Reactive compounds can also aid in the formation of other pollutants, referred to as secondary pollutants because they are not emitted, but formed from directly emitted primary precursors. Of particular concern is tropospheric ozone, a primary constituent of photochemical smog. In the remainder of this chapter, the impacts of solvents on air quality are discussed, with particular attention given to the formation of tropospheric ozone. This emphasis is motivated by current regulatory importance as well as by lingering scientific issues regarding the role of volatile organics in secondary pollution formation. 17.4.2 MODES AND SCALES OF IMPACT Many organic solvents are toxic, and direct exposure to the compound through the atmosphere (e.g., via inhalation) can be harmful. While toxic effects of solvents rely on direct exposure, many solvents also contribute to the formation of secondary pollutants such as tropospheric ozone or particulate matter (PM), which cause health problems and damage the environment on larger spatial scales such as over urban areas and multi-state/country regions. Very slowly reacting solvent compounds also impact the atmosphere on the global scale, which may cause imbalances in living systems and in the environment. While some mechanisms of environmental imbalance are understood, the risks associated with global atmospheric impacts are highly uncertain. Transport of solvents in the atmosphere is similar to most other gaseous pollutants, and is dominated by the wind and turbulent diffusion. There is little difference between the transport of different solvent compounds, and the fact that most solvents have much higher molecular weights than air does not lead to enhanced levels at the ground. Heavy solvents are, for the most part, as readily diffused as lighter solvents, although they may not vaporize as fast. The higher levels of many solvents measured near the ground are due to proximity to emissions sources, which are near the surface, and the fact that most solvents degrade chemically as they mix upwards. A major difference in the evolution of various solvents is how fast they react chemically. Some, such as formaldehyde, have very short lifetimes while others, such as CFCs, last decades. 17.4.2.1 Direct exposure Volatilization of solvents allows air to serve as a mode of direct exposure to many compounds known to be toxic. Generally, direct exposure is a risk near strong or contained sources, and can cause both acute and chronic responses. Most of the non-workplace exposure to solvents occurs indoors. This is not surprising since, on average, people spend a vast majority of their time indoors, and solvents are often used indoors. Outdoors, solvents rapidly disperse and can oxidize, leading to markedly lower levels than what is found indoors near a source. For example, indoor formaldehyde levels are often orders of magnitude greater than outdoors. There still are cases when outdoor exposure may be non-negligible, such as if one spends a significant amount of time near a major source. Toxic effects of solvents are fairly well understood, and many countries have developed regulatory structures to protect people from direct exposure. The toxic effects of solvent emissions on ecosystems are less well understood, but are of growing concern.
1190
Michelle Bergin, Armistead Russell
17.4.2.2 Formation of secondary compounds In addition to transport, organic compounds emitted into the air may also participate in complex sets of chemical reactions. While many of these reactions “cleanse” the atmosphere (most organic compounds ultimately react to form carbon dioxide), a number of undesirable side effects may also occur. Such adverse impacts include the formation of respiratory irritants and the destruction of protective components of the atmosphere. Ozone is a classic example of the complexity of secondary atmospheric impacts. Ozone is a highly reactive molecule consisting of three oxygen atoms (O3). In one part of the atmosphere ozone is beneficial, in another, it is a pollutant of major concern. Solvents and other organic emissions may either increase or decrease ozone concentrations, depending on the compound, location of reaction, and background chemistry. The mechanisms of some adverse secondary responses are discussed below. 17.4.2.3 Spatial scales of secondary effects Two layers of the Earth’s atmosphere are known to be adversely impacted by solvents - the troposphere and the stratosphere. These two layers are closest to Earth, and have distinct chemical and physical properties. The troposphere (our breathable atmosphere) is the closest layer, extending from the Earth to a height of between 10 to 15 km. The rate of chemical reaction generally determines the spatial scale over which emissions have an impact in the troposphere. Most non-halogenated solvents have lifetimes of a week or less, and elevated concentrations will only be found near the sources.4 Compounds that do not react rapidly in the troposphere (e.g., CFCs) are relatively uniformly distributed, and may eventually reach the stratosphere. The stratosphere is the next vertical layer of the atmosphere, extending from the tropopause (the top of the troposphere) to about 50 km in altitude. Little vertical mixing occurs in the stratosphere, and mixing between the troposphere and the stratosphere is slow. Impacts on the stratosphere can be considered global in scale, while impacts on the troposphere are generally urban or regional in scale. Distinct chemical systems of interest concerning solvents in the atmosphere are stratospheric ozone depletion, global climate change, and tropospheric photochemistry leading to enhanced production of ozone, particulate matter, and other secondary pollutants such as organonitrates. 17.4.2.3.1 Global impacts Because some solvent compounds are nearly inert, they can eventually reach the stratosphere where they participate in global scale atmospheric dynamics such as the destruction of stratospheric ozone and unnatural forcing of the climate system. Stratospheric ozone depletion by chlorofluorocarbons (CFCs) is a well-known example of global scale impacts. CFCs were initially viewed as environmentally superior to organic solvents. They are generally less toxic than other similarly acting compounds, less flammable and are virtually inert in the troposphere. Replacing solvents using volatile organic compounds (VOCs) with CFCs was hoped to reduce the formation of tropospheric ozone and other secondary pollutants. Because of their inert properties, there are no effective routes for the troposphere to remove CFCs, and, over the decades, emissions of CFCs have caused their accumulation, enabling them to slowly leak into the stratosphere. In the stratosphere, the strong ultraviolet (UV) light photodissociates CFCs, releasing chlorine, which then catalytically attacks ozone. CFC use has been largely eliminated for that reason. Partially halogenated organic solvents do not contribute as seriously to this problem since they react faster in the tropo-
17.4 Organic solvent impacts on tropospheric air pollution
1191
sphere than CFCs, so the associated chlorine does not reach the stratosphere as efficiently. CFCs and other solvent compounds also have a potential impact on global climate change. 17.4.2.3.2 Stratospheric ozone depletion Natural concentrations of stratospheric O3 are balanced by the production of ozone via photolysis of oxygen by strong UV light, and destruction by a number of pathways, including reactions with nitrogen oxides and oxidized hydrogen products that are present. Photolysis of an oxygen molecule leads to the production of two free oxygen atoms: O2 + hν —> O + O
[17.4.1]
Each oxygen atom can then combine with an oxygen molecule to form ozone: O + O2 —> O3
[17.4.2]
Ozone is then destroyed when it reacts with some other compound, e.g., with NO: O3 + NO —> NO2 + O2
[17.4.3]
Addition of either chlorine or bromine atoms leads to extra, and very efficient, pathways for ozone destruction. The free chlorine (or bromine) atom reacts with ozone, and the product of that reaction removes a free oxygen atom:
Net (reactions 4+5 together):
Cl + O3 —> ClO + O2
[17.4.4]
ClO + O —> Cl + O2
[17.4.5]
O3+ O —> 2 O2
[17.4.6]
Removing a free oxygen atoms also reduce ozone since one less ozone molecule will be formed via reaction 17.4.2. Thus, the chlorine atom reactions effectively remove two ozone molecules by destroying one and preventing the formation of another. Additionally, the original chlorine atom is regenerated to catalytically destroy more ozone. This reaction cycle can proceed thousands of times, destroying up to 100,000 molecules of O3 before the chlorine is removed from the system (e.g., by the formation of HCl). Reduction of ozone is greatly enhanced over the poles by a combination of extremely low temperatures, decreased transport and mixing, and the presence of polar stratospheric clouds that provide heterogeneous chemical pathways for the regeneration of atomic chlorine. The resulting rate of O3 destruction is much greater than the rate at which it can be naturally replenished. Current elevated levels of CFCs in the troposphere will provide a source of chlorine to the stratosphere for decades, such that the recent actions taken to reduce CFC emissions (through the Montreal Protocol) will have a delayed impact. 17.4.2.3 Global climate forcing Over the past decade, the potential for non-negligible changes in climate caused by human activity has been an issue of great concern. Very large uncertainties are associated with both estimations of possible effects on climate as well as estimations of the potential impacts of changes in climate. However, current consensus in the international scientific community is that observations suggest “a discernible human influence on global climate”.3
1192
Michelle Bergin, Armistead Russell
Solvent compounds, especially CFCs and their replacements, participate in climate change as “greenhouse gases”. Greenhouse gases allow short-wave solar radiation to pass through, much of which the earth absorbs and re-radiates as long-wave radiation. Greenhouse gases absorb the long-wave radiation, causing the atmosphere to heat up, thereby acting as a blanket to trap radiation that would normally vent back to space. Climate change is a controversial and complex issue, but it is likely that restrictions such as those from the Kyoto Protocol will be adopted for emissions of compounds strongly suspected of exacerbating climate change. Many countries have already adopted stringent policies to reduce greenhouse gas emissions. 17.4.2.4 Urban and regional scales Another area of concern regarding outdoor air is exposure to secondary pollutants that are due, in part, to chemical reactions involving solvent compounds. Examples include the formation of elevated levels of ozone, formaldehyde, organonitrates, and particulate matter. Formaldehyde, a suspected carcinogen, is an oxidation product of organic compounds. Tropospheric ozone and organonitrates, as discussed below, are formed from a series of reactions of organic gases and nitrogen oxides in the presence of sunlight. Particulate matter formation is linked to ozone, and some solvents may react to form particulate matter. The particulate matter of concern is small (generally less than 2.5 µm in diameter) usually formed by gas-to-aerosol condensation of compounds via atmospheric chemical reactions. Ozone and particulate matter are both regulated as “criteria” pollutants in the United States because they have been identified as risks to human health. Ozone is believed to cause respiratory problems and trigger asthma attacks, and PM has a variety of suspected adverse health outcomes (e.g., respiratory and coronary stress and failure). Many organonitrates, such as peroxyacetyl nitrate, are eye irritants and phytotoxins. Currently, the formation and effects of ozone are better understood than those of fine particulate matter and organonitrates. The following section of this chapter discusses the effects, formation, and control of tropospheric ozone. The role of solvents in forming particulate matter is currently viewed as less urgent. 17.4.3 TROPOSPHERIC OZONE Tropospheric ozone, a primary constituent of photochemical smog, is naturally present at concentrations on the order of 20-40 parts per billion (ppb).4 However, elevated levels of ground-level ozone are now found virtually worldwide, reaching in some cities concentrations of up to 10 times the natural background. 17.4.3.1 Effects Ozone is believed to be responsible for both acute (short-term) and chronic (long-term) impacts on human health, especially on lung functions. Major acute effects of ozone are decreased lung function and increased susceptibility to respiratory problems such as asthma attacks and pulmonary infection. Short-term exposure can also cause eye irritation, coughing, and breathing discomfort.5-7 Evidence of acute effects of ozone is believed to be “clear and compelling”.8 Chronic health effects may present a potentially far more serious problem; however, definitive evidence is difficult to obtain. Recent studies do suggest that ambient levels of ozone induce inflammation in human lungs, which is generally accepted as a precursor to irreversible lung damage,6 and chronic animal exposure studies at concentrations within current ambient peak levels indicate progressive and persistent lung function and structural abnormalities.5,8
17.4 Organic solvent impacts on tropospheric air pollution
1193
Crop damage caused by air pollution has also received much attention. It is estimated that 10% to 35% of the world’s grain production occurs in regions where ozone pollution likely reduces crop yields.9 Air pollution accounts for an estimated several billion dollar crop loss every year in the United States alone, and research and analysis suggests that about 90% of this crop loss can be directly or indirectly attributed to ozone.10 Evidence also indicates that ozone may cause short- and long-term damage to the growth of forest trees,11 as well as altering the biogenic hydrocarbon emissions of vegetation.12 17.4.3.2 Tropospheric photochemistry and ozone formation In the lowest part of the atmosphere, chemical interactions are very complex. A large number of chemical compounds are present, the levels of many of these compounds are greatly elevated, and emissions vary rapidly due to both natural and anthropogenic sources. Ozone formation in the troposphere results from non-linear interactions between NOx, VOCs, and sunlight.4,13 In remote regions, ozone formation is driven essentially by methane,14 however elsewhere most VOCs participate in ozone generation. For example, measurements of non-methane organic compounds in the South Coast Air Basin of California during the 1987 Southern California Air Quality Study, identified more than 280 ambient hydrocarbon and oxygenated organic species,15 many of which originated from solvents and contribute in differing degrees to ozone generation. The only significant process forming O3 in the lower atmosphere is the photolysis of NO2 (reaction with sunlight), followed by the rapid reactions of the oxygen atoms formed with O2. The only significant process forming O3 in the lower atmosphere is the photolysis of NO2 (reaction with sunlight), followed by the rapid reactions of the oxygen atoms formed with O2. NO2 + hν → O(3P) + NO
[17.4.7]
O(3P) + O2 + M → O3 + M This is reversed by the rapid reaction of O3 with NO, O3 + NO → NO + O2
[17.4.8]
This reaction cycle results in a photostationary state for O3, where concentrations only depend on the amount of sunlight available, dictated by the NO2 photolysis rate (k1) and the [NO2]/[NO] concentration ratio.
[O 3 ]steady −state
=
k1 [NO 2 ] k 2 [NO ]
[17.4.9]
Because of this photostationary state, ozone levels generally rise and fall with the sun, behavior that is referred to as “diurnal.” If the above NOx cycle were the only chemical process at work, the steady-state concentrations of ozone would be relatively low. However, when VOCs such as organic solvent compounds are present, they react to form radicals that may either (1) consume NO or (2) convert NO to NO2. This additional reaction cycle combined with the above photostationary state relationship causes O3 to increase.
1194
Michelle Bergin, Armistead Russell
Although many types of reactions are involved,4,13,16,17 the major processes for most VOCs can be summarized as follows: VOC + OH → RO2 + products
[17.4.10a]
RO2 + NO → NO2 + radicals
[17.4.10b]
radicals → ...→ OH + products
[17.4.10c]
products → ...→...+ CO2
[17.4.10d]
The last two pseudo-reactions given comprise many steps, and the products often include formaldehyde, carbon monoxide and organonitrates. The rate of ozone increase caused by these processes depends on the amount of VOCs present, the type of VOCs present, and the level of OH radicals and other species with which the VOCs can react. One of the major determinants of a compound’s impact on ozone is the rate of the reaction of the particular VOC with the hydroxyl radical via reaction [17.4.10a], above. The total amount of ozone formed is largely determined by the amount of VOC and NOx available. The dependence of O3 production on the initial amounts of VOC and NOx is frequently represented by means of an ozone isopleth diagram. An example of such a diagram is shown in Figure 17.4.1. The diagram is a contour plot of ozone maxima obtained from a large number of air quality model simulations using an atmospheric chemical mechanism. Initial concentrations of VOC and NOx are varied; all other variables are held constant. Notice that there is a “ridge” along a certain VOC-to-NOx ratio where the highest ozone concentrations occur at given VOC levels. This is referred to as the “optimum” VOC-to-NOx ratio. While the atmosphere is more complicated than this idealized system, important features are very similar. VOC-to-NOx ratios sufficiently low to retard ozone formation from an optimum ratio (represented in the upper left quadrant of Figure 17.4.1) can occur in central cities and in plumes immediately downwind of strong NOx sources. Rural environments tend to be characterized by fairly high VOC-to-NOx ratios because of the relatively rapid removal of NOx
Figure 17.4.1. Ozone isopleth diagram showing the dependencies of ozone on varying levels of initial VOCs and NOx. Concentrations are given in ppb. [Adapted from M.S. Bergin et al., Enc. of Env. Analysis and Remediation, 29, 3029, (1998)]
17.4 Organic solvent impacts on tropospheric air pollution
1195
from non-local sources as compared to that of VOCs, coupled with the usual absence of strong local NOx sources and the presence of natural VOC sources. In such rural environments, the formation of ozone is limited more by the absence emissions of NOx, and most ozone present was directly transported from upwind. Indeed, in most of the troposphere, except in areas of strong NOx sources, the availability of NOx governs ozone production. 17.4.3.3 Assessing solvent impacts on ozone and VOC reactivity As mentioned previously, the contribution of solvents to the VOC levels, and hence ozone formation, is significant. For example, in Los Angeles, about 25% of the VOC mass is from solvent use.1 This fraction is down from earlier years due to various controls such as using water-based paints and enclosing/controlling paint spraying operations. On the other hand, reduction in the use of CFCs as propellants has led to an increase in organic emissions from substituted compounds.2 However, the impact on ozone formation by a specific source is not directly proportional to the amount of VOC emitted by that source. A major determinant of the ozone forming potential is the reactivity of the compound or compound mixture emitted. Reactivity can be viewed as the propensity for a compound to form ozone, and this propensity varies dramatically between compounds and between environments.18,19 As seen in Table 17.4.1, ‘box’ model (single cell) simulations designed to represent summertime conditions in Los Angeles, California indicate that the amount of carbon associated with each class of compound only roughly corresponds to the amount of ozone formed from those compounds. Methane, which reacts very slowly but comprises most of the carbon, contributes little to ozone formation. Alkenes and aromatics are only a small part of the total carbon, but lead to much of the ozone formation. Table 17.4.1. Percentage of ozone production attributable to each organic. The percentages shown should be viewed as only approximate, and will depend upon local emissions characteristics. (*While not considered organic carbon, carbon monoxide acts to facilitate ozone formation similar to organic compounds.) [Adapted from F.M. Bowman and J.H. Seinfeld, J. Geophys. Res., 99, 5309, (1994) and M.S. Bergin et al., Env. Sci. Technol., 29, 3029 (1998)] Percent of carbon in each specified class
Percent of ozone due to specified organic class
carbon monoxide*
35
6
methane
40
1
aldehydes and ketones
1
3
non-methane alkanes, ~4C
8
17
non-methane alkanes, ~8C
5
16
aromatics, including toluene
3
5
aromatics, including xylenes and others
3
13
ethene
2
12
biogenic alkenes ans isoprene
1
10
other alkenes
2
17
Compound Class
1196
Michelle Bergin, Armistead Russell
17.4.3.3.1 Quantification of solvent emissions on ozone formation Two methods are generally employed to quantify the role pollutants play in forming ozone: experimental and computational. Both types of estimation approaches have their limitations. In the case of physical experiments, it is difficult to fully simulate ambient conditions, so the results do not have general applicability. In the case of computational approaches, uncertainties and approximations in the model for airshed conditions, in its formulation, and in the chemical mechanism cause uncertainties in the predicted ozone impacts. For these reasons, modeling predictions and experimental measurements are used together. 17.4.3.3.1.1 Experimental analysis Experimental analysis is performed using environmental ‘smog’ chambers, either with a series of single hydrocarbons irradiated in the presence of NOx or using complex mixtures to simulate, for example, automobile exhaust emitted into characteristic urban ambient conditions. Such chambers are large reaction vessels (some with internal volumes of cubic meters), in which air and small amounts of hydrocarbons and NOx are injected, and then irradiated with real or artificial light. Both indoor and outdoor chambers are used so behaviors can be evaluated under natural radiative conditions and under controlled conditions. While these experiments18-23 clearly indicate differences in ozone formation from individual hydrocarbons, they do not represent some important physical systems of urban pollution such as the mixing processes and continuing emissions cycles. Such experiments have focused both on groups of compounds as well as specific VOCs, including solvents. A particular limitation has been studying very low vapor pressure solvents because it is difficult to get enough of the compound into the vapor phase in the chamber to appreciably change the ozone levels. Another limitation is the expense of using smog chambers to simulate a large range of conditions that might occur in the atmosphere. On the other hand, smog chambers are very powerful, if not fundamental, for developing chemical mechanisms that describe the reaction pathways that can be used in computational approaches. 17.4.3.3.1.2 Computational analysis (air quality models) Given the limitations of physical experiments to simulate atmospheric conditions, computer models have been developed to assess the impact of emissions on ozone. These models, called airshed models, are computerized representations of the atmospheric processes responsible for air pollution, and are core to air quality management.23 They have been applied in two fashions to assess how solvents affect ozone. One approach is to conduct a number of simulations with varying levels of solvent emissions.2 The second approach is to evaluate individual compounds and then calculate the incremental reactivity of solvent mixtures.19,21,24-28 Derwent and Pearson2 examined the impact of solvent emissions on ozone by simulating air parcel trajectories ending in the United Kingdom and perturbing the emissions to account for an anticipated 30% mass reduction in VOCs from solvents between 1995 and 2007. They found a small decrease in ozone-from 78 to 77 ppb in the mean peak ozone in the UK, and a 9 ppb reduction from 129 ppb outside of London. A more substantial decrease of 33 ppb from the 129 ppb peak outside of London was found from reducing non-solvent mass VOC emissions by 30% outside of the UK and 40% within the UK. This suggests that the VOC emissions from sources other than solvents have a higher average reactivity, as is discussed by McBride et al.29 While the types of simulations conducted by Derwent and Pearson2 are important to understanding the net effect of solvent emissions on ozone, there is an unanswered associated and important question, that being which specific solvents have the greatest impacts.
17.4 Organic solvent impacts on tropospheric air pollution
1197
This question is critical to assessing if one solvent leads to significantly more ozone formation than a viable substitute (or vice versa). To evaluate the contribution of individual organic compounds to ozone formation, the use of incremental reactivities (IR) was proposed,18-21 defined as the change in ozone caused by a change in the emissions of a VOC in an air pollution episode. To remove the dependence on the amount of VOC added, incremental reactivity is defined by equation [17.4.11] as the limit as the amount of VOC added approaches zero, i.e., as the derivative of ozone with respect to VOC: IRi =
∂[O 3 ]
∂[VOCi ]
[17.4.11]
Here, IRi is the incremental reactivity and the subscript i denotes the VOC being examined. This definition takes into account the effects of all aspects of the organic’s reaction mechanism and the effects of the environment where the VOC is emitted. A similar quantity is the relative reactivity,23 RRi: RRi =
Ri
[17.4.12]
N
∑F i =1
Bi
Ri
where: FB i IRi
mass fraction of compound i in the reference mixture incremental reactivity of species i (grams ozone formed per gram compound i emitted)
In this case, the incremental reactivity is normalized by the reactivities of a suite of organics, thus removing much of the environmental dependencies found when using IRs defined by [17.4.11]. This metric provides a means for directly comparing individual compounds to each other in terms of their likely impact on ozone. A number of investigators have performed calculations to quantify incremental and/or relative reactivities for various solvents and other organics23-28,30 and references therein. Those studies found very similar results for the relative reactivities of most compounds found in solvents. Figure 17.4.2 (based on references 23, 24 and 26) shows the relative reactivities for some of the more common compounds, as well as possible solvent substitutes and isoprene, a naturally emitted organic. (For a more extensive list of relative reactivities, see 19, 27 and 30.) As can be seen, even normalized compound reactivities can vary by orders of magnitude. Some compounds even exhibit “negative” reactivities, that is that their emission can lead to ozone decreases under specific conditions. In particular, negative reactivities are found most commonly when the levels of NOx are low, e.g., in non-urban locations. For example, Kahn et al.,26 found that a solvent can promote ozone formation in one area (e.g., near downtown Los Angeles), but retard ozone formation further downwind. Kahn also found that the relative reactivities of the eight different solvents studied were similar in very different locations, e.g., Los Angeles, Switzerland and Mexico City. Looking at Figure 17.4.2, it is apparent that alkenes and aromatic hydrocarbons with multiple alkyl substitutions (e.g., xylenes and tri-methyl benzene) have relatively high reactivities. Alcohols, ethers and alkanes have lower reactivities. Halogenated organics have some of the lowest reactivities, so low that they are often considered unreactive. This suggests that there are two ways to mitigate how solvents contribute to air quality problems. The more traditional method is to reduce the mass of organic solvent emissions (e.g., by using water-based paints). A second approach is to reduce the overall reactivity of the solvent
1198
Michelle Bergin, Armistead Russell
Figure 17.4.2. Solvent relative reactivities based on mass of ozone formed per gram of solvent emitted into the gas phase. PCBTF is para-chlorobenzo-trifluoride, BTF is benzo-trifluoride and TBA is tertiary butyl acetate. [Adapted from M.S. Bergin et al., Env. Sci. Technol., 29, 3029 (1998) and M. Khan et al., Atmos. Env., 33, 1085 (1999)].
used, e.g., by switching to ethers, alcohols, alkanes or halogenated compounds. Solvent substitution, however, is complicated by the need to maintain product quality. 17.4.4 REGULATORY APPROACHES TO OZONE CONTROL AND SOLVENTS Historically, regulatory approaches to reducing ozone concentrations have relied reducing the mass emissions of VOCs,2,4 and this has led to stringent controls on solvents. Two factors are important in determining if an organic solvent is considered a VOC: its reactivity (discussed above) and its vapor pressure. In the U.S., traditionally, if a compound was less reactive than ethane, it was considered unreactive. Such compounds include many halogenated species and some acetates and ethers. Recently, acetone was also added as an unreactive compound. A vapor pressure threshold is also used in many areas (e.g., Europe) since it is viewed that compounds with very low vapor pressures will not be emitted rapidly into the atmosphere. It has been argued that a vapor pressure limit may not be appropriate since, given time, even lower vapor pressure compounds will have ample time to evaporate. Just recently, California is considering regulations that more fully account for the full range of reactivities that solvents possess. This is due, in part, to make it easier for manufacturers to meet stringent regulations being adopted in that state to help them meet their air quality goals. It is likely that other areas will also have to employ increasingly more stringent regulations, to both lower ozone and alleviate other environmental damage. In many countries, greater focus is now being placed on reducing NOx emissions to mitigate ozone formation. This has important ramifications for solvent use, indicating the
17.4 Organic solvent impacts on tropospheric air pollution
1199
regulatory focus is now turning from VOCs towards NOx, the other main precursor to ozone. Another imminent regulatory issue is the control of ambient fine particulate matter. While the role of solvent emissions in forming particulate matter is not well understood, studies to date do not suggest they are a major contributor. 17.4.5 SUMMARY Solvents are, and will continue to be, one of the major classes of organic compounds emitted into the atmosphere. These compounds have a wide range of air quality impacts. Accumulation of toxic compounds indoors is of concern, although outdoors the concern of toxicity is significantly less substantial due to rapid dilution. In the stratosphere, some of the halogenated solvents lead to depletion of the protective layer of ozone, while in the troposphere solvents generally lead to increased ozone levels, where it adversely affects health and the environment. The former has led to regulations of CFCs, and the latter to regulations of organic solvents. Some solvents are also considered to be precursors to the formation of secondary tropospheric pollutants other than ozone, such as particulate matter, however these relationships are currently less certain. In the aggregate, total VOC emissions from solvents in the U.S. are the second largest single source category in polluted urban areas, falling just behind motor vehicle VOC emissions both in terms of mass and urban ozone production. For now, regulations are designed to reduce the loss of ozone in the stratosphere and the formation of excess ozone in the troposphere. However, while some solvents are very reactive, others are substantially less reactive, suggesting that there is considerable opportunity to reduce urban ozone formation from solvents by utilizing substitutes with low ozone forming potentials. Currently, most regulations are targeted at reducing the mass of VOC emissions, not their relative impacts on ozone. REFERENCES 1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16
SCAQMD (South Coast Air Quality Management District). (1996). 1997 Air Quality Management Plan. November. R.G. Derwent and J.K. Pearson, Environ. Technol., 18, 1029 (1997). Climate Change 1995: The Science of Climate Change. Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Ed. J.T. Houghton, et al. Cambridge University Press, Cambridge, Great Britain. 1996. NRC (National Research Council). Rethinking the Ozone Problem in Urban and Regional Air Pollution. National Academy Press, Washington, DC. (1991). OTA (U.S. Congress, Office of Technology Assessment). Catching Our Breath: Next Steps for Reducing Urban Ozone, OTA-O-412, U.S. Government Printing Office, Washington, DC. July,1989. S.M. Horvath and D.J.McKee (1994). In D.J. McKee, ed., Tropospheric Ozone: Human Health and Agricultural Impacts, CRC Press/Lewis Publishers, Boca Raton, FL, Chapter 3. M. Lippmann, Environ. Sci. & Technol., 25(12), 1954, (1991). M. Lippmann, J. of Exposure Analysis and Environmental Epidemiology, 3(1), 103, (1993). W.L.Chameides, P.S. Kasibhatla, J Yienger, and H. Levy II., Science, 264(5155), 74, (1994). D.T. Tingey, D.M. Olszyk, A.A.Herstrom, and E.H. Lee, (1994). “Effects of Ozone on Crops” in D.J. McKee, ed., Tropospheric Ozone: Human Health and Agricultural Impacts, CRC Press/Lewis Publishers, Boca Raton, FL, Chapter 6. S.B.McLaughlin and L.J. Downing, Nature, 374(6519), 252, (1995). W. Mehlhorn, Nature, 327, 417, (1989). J.H. Seinfeld and S.N. Pandis, Atmospheric Chemistry and Physic: From Air Pollution to Climate Change, New York: J. Wiley & Sons, (1998). J.A. Logan, M.J. Pather, S.C. Wofsy, and M.B. McElroy, J. Geophys. Res., 86, 7210, (1981). F.W.Lurmann and H.H. Main, Analysis of the Ambient VOC Data Collected in the Southern California Air Quality Study, final report, Contract A832-130, California Air Resources Board, Sacramento, CA., 1992. R. Atkinson, Atmos. Environ., 24A, 1, (1990).
1200
17 18 19 20 21 22
23
24 25 26 27 28 29 30
Michelle Bergin, Armistead Russell
R. Atkinson, J. Phys. Chem. Ref. Data, Monograph No. 2, (1994). W.P.L. Carter, Atmos. Environ., 24A, 481, (1990). W.P.L.Carter, (1991). Development of Ozone Reactivity Scales for Volatile Organic Compounds, EPA 600/3-91-050. U.S. Environmental Protection Agency, Research Triangle Park, NC (August). W.P.L. Carter and R. Atkinson, Environ. Sci. and Technol., 23, 864, (1989). W.P.L. Carter, Atmos. Environ., 29, 2513, (1995). N.A. Kelly and P. Wang, (1996) Part I: Indoor Smog Chamber Study of Reactivity in Kelly, N.A.; Wang, P.; Japar, S.M.; Hurley, M.D.; and Wallington, T.J. (1996). Measurement of the Atmosphere Reactivity of Emissions from Gasoline and Alternative-Fueled Vehicles: Assessment of Available Methodologies, Second-Year Final Report, CRC Contract No. AQ-6-1-92 and NREL Contract No. AF-2-112961. Environmental Research Consortium, (September). M.S. Bergin, A.G. Russell, W.P.L. Carter, B.E. Croes, and J.H. Seinfeld, Ozone Control and VOC Reactivity, in the Encyclopedia of Environmental Analysis and Remediation, Ed. R.A. Meyers, J. Wiley & Sons, Inc, New York, NY. 1998. M.S. Bergin et al., Env. Sci Technol., 29, 3029 (1998) F.M. Bowman and J.H. Seinfeld, J. Geophys. Res., 99, 5309, (1994a). M. Khan et al., Atmos. Env., 33, 1085 (1999). W.P.L. Carter, J. Air and Waste Mgmt. Assoc., 44, 881, (1994a). R.G. Derwent, and M.E. Jenkin, Atmos. Environ., 25(A):1661-1673 (1991). S. B. McBride et al., Env. Sci. Technol., 31, 238a, (1997) http://www.cert.ucr.edu/~carter/bycarter.htm
18
Concentration of Solvents in Various Industrial Environments 18.1 MEASUREMENT AND ESTIMATION OF SOLVENTS EMISSION AND ODOR Margot Scheithauer Institut für Holztechnologie Dresden, Germany
18.1.1 DEFINITION “SOLVENT” AND “VOLATILE ORGANIC COMPOUNDS” (VOC) Solvents are generally understood to be substances that can physically dissolve other substances, more narrowly they are inorganic and organic liquids able to dissolve other gaseous, liquid, or solid substances. A qualifier for the suitability as a solvent is that, during the solution, neither the solvent nor the dissolved substance undergoes chemical change, i.e., the components of the solution may be recovered in their original form by physical separation processes, such as distillation, crystallization, sublimation, evaporation, adsorption. From a chemical point of view, solvents or volatile organic compounds, VOCs, vary widely. They are often classified using their boiling point, and vapor pressure. These properties do not define their suitability as solvents. In the context of VOCs, only organic solvents are of relevance. Therefore, in the following discussion, only organic solvents will be discussed. Definition of solvent according to Council Directive 1999/13/EC of 11 March 19991 Organic solvent shall mean any VOC which is used alone or in combination with other agents, and without undergoing a chemical change, to dissolve raw materials, products or waste materials, or is used as a cleaning agent to dissolve contaminants, or as a dissolver, or as a dispersion medium, or as a viscosity adjuster, or as a surface tension adjuster, or a plasticizer, or as a preservative. Definition of solvent according to ISO/DIS 4618-4: 19992 Paints and varnishes - terms and definition for coating materials: Solvent: A single liquid or blends of liquids, volatile under specified drying conditions, and in which the binder is completely soluble. For certain application purposes, solvents are defined more specifically: Technische Regeln für Gefahrstoffe TRGS 610 (Technical Regulations for Hazardous Substances) as applied to high solvent-containing primers and flooring glues the definition is:
1202
Margot Scheithauer
Solvents are volatile organic compounds as well as mixtures thereof at a boiling point < 200°C, which under normal conditions (20°C and 1013 hPa) are liquid and are applied for dissolving and diluting other substances without undergoing chemical change.3 As can be seen, there are clear differences in the way VOCs are defined.4 The most general, and hence least disputable, one is: Definition of volatile organic compounds according to DIN ISO 11890/1,2,5,6 Solvent is generally any organic liquid and/or any organic solid substance, which evaporates by itself under prevailing conditions (temperature and pressure). Definition of VOC according to Council Directive 1999/13/EC1 Volatile organic compound, VOC, shall mean any organic compound having at 293.15 K a vapor pressure of 0.01 kPa or more, or having a corresponding volatility under the particular conditions of use. For the purpose of this Directive, the fraction of creosote, which exceeds this value of vapor pressure at 293.15 K shall be considered as a VOC. In the Lösemittelverordnung Österreichs (Austrian Regulations on Solvents) of 1995,7 which on this issue corresponds to the österreichische LackieranlagenVerordnung (Austrian Regulation on Varnishing Plants) of 1995,8 VOCs have a maximum boiling point of 200°C. Hence, the volatile compounds at a boiling point > 200°C are not included. These comprise, e.g., some film forming media, such as butyldiglycol, butyldiglycolacetate, and texanol. Similarly, reactive solvents are excluded from this regulation. Definitions of VOCs in the USA according to ASTM D 3960-19 Volatile Organic Compound (VOC), means any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates, and ammonium carbonate, which participates in atmospheric photochemical reactions. This includes any such organic compound other than the following, which have been determined to have negligible photochemical reactivity: Methane; ethane; methylene chloride (dichloromethane); 1.1.1-trichloroethane (methyl chloroform); 1.1.1-trichloro-2.2.2-trifluorethane (CFC-113), furthermore cyclic, branched or linear completely methylated siloxanes: acetone: and perfluorocarbon compounds. The regulatory definition under the control of the U.S. EPA may change. Certain organic compounds that may be released under the specified bake conditions are not classified as VOC as they do not participate in atmospheric photochemical reactions. Such non-photochemically active compounds are referred to as exempt volatile compounds in the practice. An example of an exempt compound in the USA according to the EPA is tertiary-butyl-acetate (TBA, B.p. of 98 °C). It is described as a substance not harmful in air.10 Substances contained in air are differentiated in their mixtures according to their volatility:11 Abbreviation
Compound class
Range of boiling point in °C
VVOC
very volatile organic compounds
< 0 up to 50 ÷ 100
VOC
volatile organic compounds
50 ÷ 100 up to 250 ÷ 260
SVOC
semi-volatile organic compounds
250 ÷ 260 up to 380 ÷ 500
TVOC means “total volatile organic compounds”.
18.1 Measurement and estimation of solvents emission and odor
1203
18.1.2 REVIEW OF SOURCES OF SOLVENT EMISSIONS 18.1.2.1 Causes for emissions The basic human needs include: eating, drinking and breathing. In a 60-year lifetime, human being takes up about 30 t of food, 60 t of drink and 300 t of air.12 Thus air pollution may have a large influence on human health. In comparison with outside air and the ambient air within public transportation vehicles and terminals, room air quality is of the greatest interest, since the population in North American cities stays indoors more than 93% time according to findings by Szalai13 and more than 91% according to Chapin,14 of this time > 67 % is spent in living space and about 4-5 % in the workplace. Sources for volatile substances in indoor air are diverse and originate, apart from that brought in with outside air and produced by the living occupants (people, pets) and their activities (e.g., smoking), from the materials themselves which have been applied for building and furnishing the room. These include: • building materials • glues • floor covering • wallpaper • internal decorative textiles • paints • furniture, upholstered furniture, etc. These materials may contain volatile organic, but also inorganic, compounds, which are evaporated during use. The emission of these volatile compounds is influenced by a large number of factors, such as: • chemical/physical structure of the material (thickness, surface structure) • volatility, polarity of the volatile compounds • room temperature and humidity • room load, i.e., the ratio of the area of emitting surface and the volume of room air • ventilation, rate of air flow • load of the external air entering (dust, substances) The residual monomers from plastics, reactive products (e.g., formaldehyde), degradation products, flame protection media, softeners etc. may be emitted from building materials. 18.1.2.2 Emissions of VOCs from varnishes and paints A substantial part of emitted materials are solvents. They preferably originate from varnishes, paints and glues. Table 18.1.1 shows a survey of the main types of varnishes referring to their average solvent content. Table 18.1.1. Solvent share in various types of varnishes Material
Solvent content, wt%
Stain
~95
Cellulose nitrate varnish
~75
PUR varnish Polyester varnish UV roller varnish based on acrylate
35-70 35 2-10
1204
Margot Scheithauer
Despite the introduction of water based and powder varnishes, cellulose nitrate varnishes with their high solvent content, are still widely used due to their easy application and their low prices. The current varnish application still requires the use of solvents. The chosen coating procedure determines the necessary processing viscosity, which may be adjusted in different ways: • Thinning the binding agent/varnish with a solvent consisting of one or more organic compounds without altering the binding agent chemically. Nowadays, the solvent can partly or almost totally be replaced by water. • Thinning the binding agent with a reactive thinner, i.e., one or more monomers/oligomers included in the system, which react with the binding agent during the hardening process and is built into a molecule. According to definitions, they do not have characteristics of solvents. They partly take over the solvent function while they remain in the liquid state. • Temperature increase of the coating system, which limits the required amount of solvents. • Liquid application of a melted solid system Profiled construction components form parts of complicated shapes, which must rely on the spraying of coatings for decorative and protective purposes. The spraying requires low-viscosity material. Viscosity reduction is usually achieved by adding solvents but sometimes by increasing the material temperature. Parts with large flat surfaces however, may be coated by roller-coating, which tolerates higher viscosity materials. After fulfilling their function as solvents, they are expected to completely evaporate from the varnish system. Emission of solvents from a varnish system occurs in the course of the “life-cycle” of a varnish in several different locations as the following survey shows (Figure 18.1.1). As early as in the production stage of the varnish system, solvent emissions contaminate the ambient air of the factory. About 90 % of the solvents contained in the varnish system evaporates during its application which affects the air quality in the workplaces. These emissions are in the milligram per cubic meter of air range. The thinning effect decreases the solvent concentration in the ambient air at the varnisher’s workplace into the microgram to nanogram per cubic meter.
Figure 18.1.1. Emission of VOCs in the life-cycle of a varnish.
18.1 Measurement and estimation of solvents emission and odor
1205
About 10 % of the solvent remains in the varnish and it is emitted, depending on the type of solvent and varnish, with variable rate on their way to the consumers and in their living spaces. According to the European Commission, the furniture industry alone used more than 185,000 tons of solvents/year in 1994.15 Not all the VOCs enter the air, however, since remedial efforts, such as absorption and solvent reclaim, after-burning of solvents, the use of bio-filters, result in emission reduction. Nevertheless, it is still evident that limitation of solvents, VOCs, reduction in varnishes and coatings may help in reducing emissions. 18.1.2.3 VOC emissions from emulsion paints Although emulsion paints and plasters contain less than 2% VOCs these are produced in such large quantities (3.5 Million t in 1995)16 and applied over such an extensive surface that 100,000 t of VOCs are emitted from these materials per year. These enter the environment in an uncontrolled way. Prior to an interior application of the coating, there is no VOC in room atmosphere, unless sources already exist in the room. Then the load - from the beginning of the application - increases very rapidly. After the application, the VOC concentration in the ambient air decreases exponentially. If these emissions are to be monitored it is desirable to have methods available to record VOC concentrations typical of emitting stages.17 18.1.3 MEASURING OF VOC-CONTENT IN PAINTS AND VARNISHES 18.1.3.1 Definition of low-emissive coating materials In order to limit the sources of solvent emission in ambient air in the most effective manner, primary goal must be to reduce the VOC content in coating materials. In Germany, there is a draft for the Varnishing Plant Regulation18 - regulating the application of low-emissive varnishes, which will require that coating materials must initially not contain more than 420 g/l VOC. Four years after the introduction of the regulation this is reduced to 250 g/l VOC. The “Jury Umweltzeichen” (Environmental Trade Mark), which, since 1999, rewards certain products with the environmental trade mark goes even further: For flat, plain materials (e.g., interior doors, panels, floors with varnished surfaces, pre-fabricated parquet) the coating systems applied must not exceed the VOC content = max. 250 g/l.19 Varnishing plants are exempt if they employ waste gas purification facilities corresponding to the EU VOC Guideline.20 The draft of the Varnishing Plant Regulation follows the US method according to ASTM D 3960-121 in determining the VOC content. 18.1.3.2 Determination of the VOC content according to ASTM D 3960-1 ASTM D 3960-1 is the method used in the US to determine the VOC content of coating materials. The VOC value is calculated as follows: VOC value [g / l ] =
mass of volatile components [g ] − mass of water [g ] volume of varnish [l ] − volume of water [l ]
[18.1.1]
For varnishes exclusively containing solvents, this formula results in high values. This calculation is, however, of dubious value for water-containing varnishes. The unit g/l does
1206
Margot Scheithauer
not refer to 1 liter of an actual varnish, but to a hypothetical product, which only expresses the binding agent in liters. This is demonstrated in Example 1, which will then also be calculated according to other methods. Example 1: Water spray varnish: 30% solid matter content (binding agent and pigments) 10% solvent content 60% water content varnish density: 1.0 g/cm3 VOC (g/l) = (700 g - 600 g)/(1 l - 0.6 l) = 250 g/l 18.1.3.3 Determination of the VOC content according to ISO/DIS 11 890/15 and 26 18.1.3.3.1 VOC content > 15% The calculation according to ISO/DIS 11 890/1 gives a more logical result. VOCs are given here as a mass share in % of the ready-to-use varnish. At a VOC content > 15 % (of the mass share), the determination is made by the differential approach: VOC [% of the mass share] = 100 - nfA - mw
[18.1.2]
where: nfA mw
non-volatile parts determined by ISO 325122 (1 h/110°C) mass as water in % of the mass parts, determined according to ISO 76023 (after Karl-Fischer)
Many more recently available techniques for water determination are not sufficiently common and thus are not considered. Example 2: Water solvent stain: 5% solid matter content = non-volatile parts 25% solvent 70% water share VOC (mass %) = 100 - 5 - 70 = 25 % For density ≠ 1 VOC [g/l] = 10(100 - nfA - mw) ρs where: ρs
varnish density in g/ml at 23°C
In this case, the VOC content refers to 1 l of actual varnish. Example 3: Solvent stain: 5% solid matter content 95% solvent share VOC (g/l) = (100 - 5 - 0) × 0.9 = 855 g/l Example 4: Solvent/water stain: 5% solid matter content 70% solvent 25% water share
[18.1.3]
18.1 Measurement and estimation of solvents emission and odor
1207
Density: 0.9 g/cm3 VOC (g/l) = (100 - 5 - 25) × 0.9 × 10 = 630 g/l DIN ISO 11 890 also allows the calculation to be made using the method prescribed by ASTM D 3960: VOClw [g / l ] = 1000
100 − nfAmw ρ mw s 100 − ρ s ρw
[18.1.4]
where: ρs nfA mw ρw
varnish density in g/ml at 23°C non-volatile parts determined by ISO 325122 (1 h/110°C) mass as water in % mass parts, determined acc. to ISO 76023 (after Karl-Fischer) density, in g/ml, of water at 23°C (ρw = 0.997537 g/ml at 23°C)
Example 1 Water spray varnish: 30% solid matter content 10% solvent 60% water content varnish density 1.0 g/cm3 water density 1.0 g/cm3 VOClw = [(100 - 30 - 60)/(100 - 1.0 × 60)] × 1.0 × 1000 = 250 g/l Example 1: Water spray varnish but with a density of 0.9 g/cm3 VOClw = [(100 - 3 - 60)/(100 - 0.9 × 60)] × 0.9 × 1000 = 196 g/l Or another example: Example 5: Water varnish primer: 15% solid matter content 10% solvent 75% water varnish density 1.0 g/cm3 VOClw = [(100 - 15 -75)/(100 - 1.0 × 75)] × 1.0 × 1000 = 400 g/l The calculation permits the deduction of exempt compounds, which are defined as organic compounds which do not participate in photochemical reactions in the atmosphere. i =n
VOClwe [g / l ] = 1000
100 − nfA − mw ∑ m eci i =1
i =n m m 100 − ρ s w − ρ s ∑ eci ρw i =1 ρ eci
ρs
[18.1.5]
where: VOClwe nfA mw meci ρs
VOC content, in g/l, contained in the product in its ready-to-use condition minus water and minus exempt compounds content of non-solvent parts, as a mass share in % water content, as mass share in % content, as mass share in %, of the exempt compound i density, in g/ml, of sample at 23°C
1208
Margot Scheithauer ρw ρeci 1000
density, in g/ml, of water at 23°C (ρw = 0.997537 g/ml at 23°C); density, in g/ml, of the exempt compound i conversion factor × 1000 in g/l
Example 4: Exempt compounds are propanol and ethanol Solvent water stain 5% solid matter content 20% propanol (density 0.8 g/cm3) (density 0.9 cm3) 10% ethanol (density 0.79 g/cm3) 40% other solvent 25% water VOClwe = [(100-5-25-(10))/{100-0.9(25/1.0)-0.9[(20/0.8)+(10/0.79)]}]× 0.9×1000=825 g/l Calculated according to DIN ISO 11 890-1, for the same water stain applies (example 4) after approach 1: 70% after approach 2: 630 g/l (propanol and ethanol are treated as VOC) 18.1.3.3.2 VOC content > 0.1 and < 15 % For VOC content > 0,1 (by mass) and < 15 % the determination is made using gas chromatography (ISO/DIS 11 890-2).6 A hot or the cold injection is used depending on the sample properties. After assessing the peak areas, the quantitative assessment and evaluation of the VOC content in the product in its ready-to-use condition, the calculation is made in the simplest case according to: i =n
VOC [g / l ] = ∑ m i ρ s 1000
[18.1.6]
i =1
where: VOC mi ρs 1000
VOC content, in g/l, of the ready-to-use product; mass, in g, of compound i in 1 g of the sample density, in g/ml, of the sample at 23°C conversion factor × 1000 in g/l.
Table 18.1.2 shows how the application of the various formulae affects the result of calculation. The VOC contents, calculated after ISO/DIS 11 890, are credible for Examples 2, 3, 4. Values calculated according to ASTM, using ratio solvent - water in Examples 2 and 4, are almost the same as VOC contents calculated according to ISO/DIS. Only by increasing the solid matter content, the VOC content can be reduced in examples calculated according to ASTM (cf. Examples 5 and 1), while an increase in the water content, in spite of the same solvent share, leads to an increase of the ASTM value, because a higher water content increases the sum of the denominator and consequently increases the VOC content. 18.1.3.4 Determination of VOC-content in water-thinnable emulsion paints (in-can VOC) DIN 55 64924 describes an approach, which, via the so-called total evaporation method, assesses the content of volatile organic compounds in water-thinnable emulsion paints (in-can VOC).
18.1 Measurement and estimation of solvents emission and odor
1209
Table 18.1.2. Examples for the VOC content calculated by various methods VOC content according to Example
Coating material wt%
Non-vola- Solvent Water Density tile parts fraction fraction 3 g/cm wt% wt% wt%
ASTM D 3960 g/l
DIN ISO 11 890/1 %
g/l
1
Water spray varnish
30
10
60
1.000 0.900
250 196
10 10
100 90
2
Water/solvent stain
5
25
70
1.000
833
25
250
3
Solvent stain
5
95
0
0.900
855
95
855
4
Solvent/water stain
5
70
25
0.900
839
70
630
5
Water varnish primer
15
10
75
1.000
400
10
100
Measuring principle: The VOCs are totally evaporated from a very small amount (a few milliliters) of the thinned original sample by means of a head-space injector and subsequent gas chromatographic analysis. All components, whose retention time is lower than the retention time of tetradecane (boiling point 252.6°C) are included as VOCs. VOC content calculation: VOC [mg / kg ] = 1000
mVOC Eρ
[18.1.7]
where: mVOC Eρ
mass in mg VOC, related to originally weighted-in quantity of the original sample Eρ originally weighted-in quantity in g of the original sample
VOC[g / l ] =
VOC[ mg / kg ] ρ df 100
[18.1.8]
where: ρdf
density in g/ml of the original sample (emulsion paint)
18.1.4 MEASUREMENT OF SOLVENT EMISSIONS IN INDUSTRIAL PLANTS 18.1.4.1 Plant requirements In spite of considerable reductions in the use of organic solvents, the worldwide-adopted ozone values in the troposphere are not being attained. It is crucial to apply consistent measures worldwide to reduce solvent emissions. In Europe, for example, the Council Directive 1999/13/EG1 “On limiting emissions of volatile organic compounds, due to the use of organic solvents in certain activities and installations” [EU-VOC-Richtlinie] came into force on March 11, 1999. It has to be transferred into national laws within two years. The application area refers to activities in applying solvents, as enlisted in Appendix IIA of the EU-VOC-Richtlinie (Table 18.1.3).
1210
Margot Scheithauer
Table 18.1.3 Threshold and emission controls d
c a
Activity
Web offset printing (>15)
b
Threshold
15-25
100
30(1)
>25
20
30(1)
Publication rotogravure (>25) Other rotogravure, flexography, rotary screen, printing, laminating or varnishing units (>15) rotary screen printing on textile/cardboard (>30) Surface (1) cleaning (>1)
e
Fugitive Total emission emission values, % Exist. New Exist. New
Emission limit, mgC/Nm³
75
15-25 >25 >30(1)
10
100 100 100
Special provisions (1)
Solvent residue in finished product is not to be considered as part of fugitive emissions
15
(1)
25 20 20
Threshold for rotary screen printing on textile and on cardboard
(1)
1-5 >5
(2)
20 20(2)
Using compounds specified (2) in Article 5(6) and (8); Limit refers to mass of compounds in mg/Nm³, and not to total carbon
15 10
(1)
Other surface cleaning (>2)
2-10 >10
75(1) 75(1)
20(1) 15(1)
Installations which demonstrate to the competent authority that the average organic solvent content of all cleaning materials used does not exceed 30 wt% are exempt from application of these values (1)
Vehicle coating (<15) and vehicle refinishing
>0,5
(1)
50
Compliance in accordance with Article 9(3) should be demonstrated based on 15 minute average measurements
25
(1)
Coil coating (>25)
50(1)
5
10
For installations which use techniques which allow reuse of recovered solvents, the emission limit shall be 150
18.1 Measurement and estimation of solvents emission and odor
d
c a
Activity
b
Threshold
Emission limit, mgC/Nm³
1211
e
Fugitive Total emission emission values, % Exist. New Exist. New
Special provisions (1)
Other coating, including metal, plastic, textile(5), fabric, film and paper coating
5-15 >15
100(1)(4) 50/75(2)(3)(4)
Emission limit value applies to coating application and drying processes operated under (2) contained conditions. The first emission limit value applies to drying processes, the second to coating application (3) processes. For textile coating installations which use techniques which allow reuse of recovered solvents, the emission limit applied to coating application and drying processes taken (4) together shall be 150. Coating activities which cannot be applied under contained conditions (such as shipbuilding, aircraft painting) may be exempted from these values, in accordance with Article (5) 5(3)(b). Rotary screen printing on textile is covered by activity No 3
20(4) 20(4)
(>5)
Winding wire coating (>5)
10 g/kg(1) 5 g/kg(2)
(1)Applies to installations where average diameter of wire (2) ≤0.1mm. Applies to all other installations (1)
Coating of wooden surfaces (>15)
15-25 >25
100(1) 50/70(2)
25 20
Emission limit applies to coating application and drying processes operated under con(2) tained conditions. The first value applies to drying processes, the second to coating application processes
1212
Margot Scheithauer
d
c a
Activity
b
Threshold
Emission limit, mgC/Nm³
e
Fugitive Total emission emission values, % Exist. New Exist. New
Special provisions (1)
20 g/kg(1)(2)(3)
Expressed in mass of solvent emitted per kilogram of product (2) cleaned and dried. The emission limit in Article 5(8) does (3) not apply to this sector. The following exemption refers only to Greece: the total emission limit value does not apply, for a period of 12 years after the date on which this Directive is brought into effect, to existing installations located in remote areas and/or islands, with a population of no more than 2000 permanent inhabitants where the use of advanced technology equipment is not economically feasible
11 kg/m³
Does not apply to impregnation with creosote
85 g/m² 75 g/m² 150 g/m²
Emission limits are expressed in grams of solvent emitted per (1) m² of product produced. For leather coating activities in furnishing and particular leather goods used as small consumer goods like bags, belts, wallets, etc.
Footwear manufacture (>5)
25 g per pair
Total emission limit values are expressed in grams of solvent emitted per pair of complete footwear produced
Wood and plastic lamination (>5)
30 g/m²
Dry cleaning
Wood impregnations (>25)
Coating of leather (>10)
(1)
100(1)
45
10-25 >25 (>10)(1)
(1)
Adhesive coating (>5)
5-15 >15
(1)
50 50(1)
25 20
If techniques are used which allow reuse of recovered solvent, the emission limit value in waste gases shall be 150
18.1 Measurement and estimation of solvents emission and odor
d
c a
Activity
Manufacture of coating preparations, varnishes, inks and adhesives (>100)
Threshold
Emission limit, mgC/Nm³
100-1000 >1000
150 150
b
1213
e
Fugitive Total emission emission values, % Exist. New Exist. New
5% of solvent input
5 3
Special provisions
The fugitive emission value does not include solvent sold as part of a coatings preparation in a sealed container (1)
Rubber conversion (>15)
20(1)
25(2)
25% of solvent input
If techniques are used which allow reuse of recovered solvent, the emission limit value in (2) waste gases shall be 150. The fugitive emission value does not include solvent sold as part of products or preparations in a sealed container
f see below the table
Total emission limit values for installations processing individual batches of seeds and other vegetable matter should be set by the competent authority on a case-by-case basis, applying the best available (2) Applies to all techniques. fractionation processes excluding de-gumming (the removal (3) of gums from the oil). Applies to de-gumming
(1)
Vegetable oil and animal fat extraction and vegetable oil refining activities (>10)
(1)
Manufacturing of pharmaceutical products (>50)
20(1)
5(2)
15(2)
5% 15% of solvent input
If techniques are used which allow reuse of recovered solvent, the emission limit value in (2) waste gases shall be 150. The fugitive emission limit value does not include solvent sold as part of products or preparations in a sealed container
a (solvent consumption threshold in tonnes/year); b(solvent consumption threshold in tonnes/year); cvalues in waste gases; d(percentage of solvent input); elimit values; fAnimal fat: 1.5 kg/tonne, Castor: 3 kg/tonne, Rape seed: 1 kg/tonne, Sunflower seed: 1 kg/tonne, Soya beans (normal crush): 0.8 kg/tonne, Soya beans (white flakes): 1.2 kg/tonne, Other seeds and other vegetable matter: 3 kg/tonne(1) 1.5 kg/tonne(2) 4 kg/tonne(3)
The quoted threshold levels for the solvent consumption in tonnes/year is of decisive importance as to whether a plant falls under this directive. This value changes depending on the technical feasibility within the industry. For example, the threshold level for solvent consumption is for illustration-gravure-printing > 25 tonnes/year, for wood coating > 15 tonnes/year.
1214
Margot Scheithauer
An analysis of the activity requires data on solvent consumption as calculated from the amount of solvent purchased both as pure solvent and included in solvent containing materials less the amount of solvent retained and/or contained in waste. The calculated difference includes all emissions including diffuse emissions such as the solvent loss from drying racks or solvent initially retained by pained, varnished, printed or dry cleaned articles. These measurements are essential if the statutes are to be enforced and emissions are to be effectively reduced. Emissive limit values are stipulated for plants covered under this directive for exhaust-gases in mgC/Nm³ as well as limits for diffuse emissions in % solvents input. If these data are not available, total emission limit values are used. Definition of “Emission Limit Value”: The “Emission Limit Value” is understood as the mass of volatile organic compounds, the concentration, the percentage and/or the amount of emission - ascertained under normal conditions - expressed in certain specific parameters, which in one or several time periods must not be exceeded [EU-VOC-Richtlinie].1 Plant measurement must be made to provide data for analyzing the actual situation as well as for being able to prove the emission reduction which business must attain. 18.1.4.2 The determination of the total carbon content in mg C/Nm³ 18.1.4.2.1 Flame ionization detector (FID) The total C/Nm³ is assessed according to the Guideline VDI 3481/page 325 by means of a flame ionization detector (FID). This device is the component of a mobile device for random sample tests or a continuously measuring device for total carbon concentration measurement in an exhaust-gas flow. This approach measures the total organic substance in an exhaust gas. Should the composition of the solvents contained in the exhaust air not be known, their concentration may be quoted in carbon equivalents as mgC/m³. Measuring principle: An FID detects ionized organically compounded C atoms in a hydrogen flame. The ion flow developed in the induced electrical field is electrically amplified and measured. The ion flow arising when burning carbon compounds is proportional to the mass of carbon atoms exposed to the flame per time unit. The detection of organic compounds with heteroatoms, e.g., N, O, S, Cl, is generally less sensitive. The calibration of the device is done in most cases with propane as the test gas. Procedural data (measuring ranges, proof limits, etc.) by FIDs of various types are compiled in a table in the Guidelines VDI. Figure 18.1.2 shows a measuring arrangement for assessing emissions from a drying plant by means of an FID. 18.1.4.2.2 Silica gel approach The determination of the total carbon concentration in an exhaust gas by means of the silica gel approach is effected according to the Guideline VDI 3481, p. 2.26 A partial flow of the exhaust air to be tested is guided through a sorption pipe filled with silica Figure 18.1.2. Emission measurement by FID. gel. The organic compounds are absorbed
18.1 Measurement and estimation of solvents emission and odor
1215
by the silica gel. Subsequently, the organic compounds are desorbed in an oxygen flow at an increased temperature and burnt to carbon dioxide (CO2), which is determined quantitatively. This is a discontinuous method. The data reflect concentration in the measured time intervals and the causes of deviations are difficult to ascertain. 18.1.4.3 Qualitative and quantitative assessment of individual components in the exhaust-gas While the methods mentioned under 18.1.4.2 reflect the total carbon concentration as a summary parameter, it is still necessary to know concentrations of the individual solvents to assess the plant emission. 18.1.4.3.1 Indicator tubes For many solvents and other substances directly indicating detector tubes are available.27 They are used for rapid assessment of emission. Since they often do not react specifically with a certain compound alone, errors have to be considered due to cross-sensitivity. Detector tubes contain sorption-active agents. A gas flow of a defined rate is fed through and the substance in the gas flow produces a color reaction with the agent in the tube. The color intensity is checked against the scale on the test tube and the ppm value can be registered. Figure 18.1.3 shows such a tube for determining ethyl acetate.27 The color change in this case is from yellow to light green. Reaction principle: CH3-COOC2H5 + CrVI → CrIII + various oxidation products This reaction is not substance-specific. Other alcohols, at varying cross-sensitivity as well as methyl ethyl ketone, are reactive as well. 18.1.4.3.2 Quantitative solvent determination in exhaust gas of Figure 18.1.3. Indicaplants by means of gas-chromatography tion tube for ethyl aceA defined amount of exhaust gas is fed into an adsorption medium (actate 500/a-D. tivated carbon, thenax, etc.) and its concentration is enriched. There is a subsequent extraction by means of a solvent. The mixture is then injected into a gas chromatograph. The individual components are determined according to Guideline VDI 34 82, page 1.28 Measuring principle: A heatable capillary coated on the inside with various materials (polyester or silicon resins, silica gel) is used as a separating unit. The gaseous substance mixture is transported by means of an inert gas (nitrogen, argon, etc.). The individual solvents remain in the column for different time periods depending on their polarity. After leaving the column solvent is detected by sensitive physical methods. By combining the highly resolving capillary gas chromatography with a mass spectrometer even unknown substances may be identified in the mixtures. The quantitative evaluation of the gas chromatograms is automatically done with appropriate software. 1st Example: Emission measurement in an industrial varnishing plant The measurement was done in a furniture company, which applies different varnish systems on several varnishing lines.29 For preliminary investigations, four chimneys were selected for the measurement as typical sources of emission.
1216
Margot Scheithauer
Based on the analysis of the varnish materials and measurements performed over extended period of time, it is certain that the materials and concentrations processed during the measuring gave a representative cross-section of the production activity. Table 18.1.4 shows a summary of the varnish and solvent amounts processed per day, which were emitted from chimneys 1 to 4. Table 18.1.4. Summary of the varnish and solvent amounts processed in the 1st example Measuring point chimney
Varnish
Solvent share, % 34 76 80 80 80
10 - 15
1
1 2 3 3 4
2
5 6
56 34
50 - 60 10
7
38
30
Varnish processed, l/day
Total solvent, l/day
30 - 35 5 5
~40 - 45
37
3
12.5
4
6
34
3
1 8 6
34 52.88 34
~240 120-130 50
~170
At the chimneys, the following exhaust air parameters were established: Chimney diameter mm Exhaust air flow rate m/s Exhaust air temperature °C Humidity % Volume flow (operation) m³/h Volume flow Nm³/h Heat flow MW Table 18.1.5 contains a summary of the established emission values expressed in mg C total/Nm³ as well as the mass flow g C/h.
1
10.35 - 11.20 11.45 - 12.00 13.45 - 14.35
35 200
112.6 96.5 128.6
3 964 3 397 4 527
93 87
Mass flow, TGE/m³
Odorant concentration(1), GE/m³
Mass flow, gC/h
Carbon concentration, mgC/Nm³
Volume flow, Nm³/h
Time
Measuring point chimney
Table 18.1.5. Measuring results for total carbon concentration at different measuring points
3 519 3 292
(1)
Mass flow, TGE/m³
Odorant concentration(1), GE/m³
1217
Mass flow, gC/h
Carbon concentration, mgC/Nm³
Volume flow, Nm³/h
Time
Measuring point chimney
18.1 Measurement and estimation of solvents emission and odor
2
11.25 - 12.00 12.45 - 13.10 13.10 - 14.30
39 240
72.4 56.3 56.3
2 841 2 209 2 209
91 87 93
4 151 3 969 4 242
3
13.50 - 14.40 09.25 - 09.55 09.55 - 10.25
33 900
48.2 56.3 64.3
1 634 1 908 2 180
100 115
3 747 4 309
4
09.15 - 09.45 09.45 - 10.15 10.15 - 10.45
108 650
112.5 152.8 160.8
12 223 16 602 17 471
98 213
12 837 27 900
The odorant concentration does not refer to the times stated.
The EU VOC Directive for plants for wood coating with a solvent consumption of 15 25 t/year requires an emission limit of 100 mg C/Nm³. In the example, this is only met at chimneys 2 and 3. As a parallel to these measurements, using activated carbon tubes and subsequent gas chromatographic evaluation it was determined that the main components of VOCs were acetone, ethyl acetate, toluene, butyl acetate, xylene, ethylbenzene (Table 18.1.6). Butyl acetate and ethyl acetate were the main components in the respective solvent mixtures. Table 18.1.6. Measured results for individual components at different measuring points Acetone MPC*
a
Ethyl acetate
Conc
MF
b
1
6.5 4.5
2
a
Conc
MF
b
229 158
15.5 11.0
11.0 4.5
432 176
3
-
4
9.0 5.5
Toluene a
Butyl acetate
Conc
MF
b
546 387
9.0 5.5
24.5 9.0
961 353
-
4.5 15.5
978 597
20.0 27.0
a
Ethyl benzene a
b
Xylene
Conc
MF
b
Conc
MF
Conca
MFb
317 194
15.5 11.0
545 387
1.0 1.0
352 352
4.2 2.5
158 88
14.5 6.5
569 255
24.5 14.5
961 569
1.0 -
39 -
6.5 -
255 -
152 525
2.0 5.5
68 186
3.5 8.0
118 271
0.5 1.0
17 34
2.0 5.5
68 186
2 173 2 933
9.0 18.0
978 1 956
17.0 37.0
1 847 4 020
2.0 3.5
217 380
8.0 17.0
869 1 847
*MPC = Measuring point chimney; aconcentration in mg/Nm³; bMF = mass flow in g/h
2nd Example: Emissions measured in a spray-room The usual technique of measuring VOC emissions in plants by means of the FID according to VDI 3481/page 3 has the shortcoming that individual components in the gas mixture cannot be measured separately, but only determined as total carbon. Also the device needs to be placed in the immediate vicinity of the measuring point and requires frequent calibration.
1218
Figure 18.1.4. Photoacoustic Multigas Monitor 1312.
Margot Scheithauer
A portable multi-gas monitor 1312 [Bruel & Kjaer]30 weighing 9 kg (Figure 18.1.4) enables sampling in six places up to 50 m away from the monitor. It can determine at each measuring point the concentration of up to a maximum 5 components as well as the humidity. This is done by a multi-point sampler with a dosing apparatus controlled by system software. The device needs calibration only four times a year. Beyond that, the device may also be used for determining the total C concentration.
Measuring principle (Figure 18.1.5) 1. The “new” air sample is hermetically sealed in the analysis cell by closing the inlet and outlet valves. 2. Light from an infrared light source is reflected by a mirror, passed through a mechanical chopper, which pulsates it, and then passes through one of the optical filters in the filter carousel. 3. The light transmitted by the optical filter is selectively absorbed by the gas monitored, causing the temperature of the gas to increase. Because the light is pulsating, the gas temperature increases and decreases, causing an equivalent increase and decrease in the pressure of the gas (an acoustic signal) in the closed cell. 4. Two microphones mounted in the cell wall measure this acoustic signal, which is directly proportional to the concentration of the monitored gas present in the cell. 5. The filter carousel turns so that light is transmitted through the next optical filter, and the new signal is measured. The number of times this step is repeated depends on the number of gases being measured.
Figure 18.1.5. Measurement principle of Photoacoustic Multigas Monitor Type 1312.
18.1 Measurement and estimation of solvents emission and odor
1219
Figure 18.1.6. Maximum concentration of n-butyl acetate in the spray-room of a joiner’s with windows and door closed. A - Spray wall, B - Rack track for shelving the varnished parts.
The response time is only 13 s for a single gas or water-vapor, or approx. 40 s if 5 gases and water-vapor are measured. The measurements are quasi-continuous. When measuring a single gas, the measuring value is updated every 30 s, with five gases and humidity the update time is 105 s. The measured values are stored, statistically evaluated and numerical or graphical output can be obtained either immediately or on demand. Such devices make the emission measurement effort much easier than with the older, less automated equipment. Figure 18.1.6 shows the results of emission investigations in the spray-room of a joiner’s with and without ventilation.31 Windows and doors were closed. The varnisher was standing in front of a spray-wall, opposite to him was placed a platform rack truck for shelving the varnished parts. Those areas in the joiner’s, which showed roughly the same emission values of n-butyl acetate, are marked on each drawing. The emission load in the room was decreased by the technical ventilation to about a tenth of the values. 18.1.5 “ODOR” DEFINITION One of the senses of the living organism is the sense of smell. Smell is caused by a substance whose effect is largely dependent on its chemical structure. The odorous substances perceived by human are suspended in the air as volatile substances. The degree of response to a substance depends on its vapor pressure, thus, in turn on its concentration in the air and its molecular weight. An odorant also has to be sufficiently water and fat-soluble, which enables it to interact with the olfactory receptors.
1220
Margot Scheithauer
Whether an odorous impression is considered pleasant or unpleasant is largely determined by the functional groups of the chemical compounds (Table 18.1.7). Table 18.1.7. Scent qualities of various chemical compounds in relation to their functional groups33,34 Odorous impression
Functional group
Substance group
Pleasant
-OH -OR -CHO -COR -COOR -CN -NO2
Alcohols Ethers Aldehydes Ketones Esters Cyanogen compounds Nitro compounds
Unpleasant
-SH -SR -CSR -NC -NH2
Merkaptanes Thioethers Thioketones Nitriles Amines
This division should be, however, only regarded as a rough guideline, since the so-called “pleasant” odor, at higher concentrations, can easily be perceived as very “unpleasant”. Almost all solvents, such as, ketones, esters, glycols, alcohols, aromatic and aliphatic hydrocarbons, contribute to a more or less intensive smell. “Odor” is not a parameter of substance, but a summarized parameter of effects. Its determination is based on the fact that the sense of a smell can be used subjectively to evaluate certain substances (odorants). The concentration of those substances suspended in the air can be determined, which is called odorant concentration. Odor threshold or perception threshold35-38 The odor threshold is a concentration of the odorant in the air, given in mg/m³. The odor threshold corresponds to an odor unit (GE). It is the amount of odorous substance, which - distributed in 1 m³ of scent-neutral synthetic air - initiates in just 50% of the evaluators a perception of smell and in the other 50%, no response. The odor threshold is very specific to substance. It is determined in several measuring series and the results form a Gaussian distribution curve. Since this is ultimately a subjective evaluation, one should not be surprised to find more as well as, less, reliable data in literature. Examples are shown in Table 18.1.8. Odorant concentration The odorant concentration of a sample is the multiple of the odor threshold and it is determined in odor units (GE) per 1 m³ neutral air. Odor intensity39 Since the odor threshold alone is an insufficient evaluation criterion for an odorant, the increase in response with increasing odorant concentration may additionally be taken into account as a scale of reference. The increase in response is mainly material and/or mixture dependent for a given odorant concentration and is called odor intensity. [Schön, p. 68]32
18.1 Measurement and estimation of solvents emission and odor
Table 18.1.8. Odor thresholds [Geruchs-Immissions-Richtlinie]34 Compound
of
Odor threshold, mg/m³
Butyl acetate
0.03
Ethyl acetate
selected
1221
compounds
Compound
from
literature
Odor threshold, mg/m³
Ethanol
19.1
22
1-butanol
0.4
Benzene
16.2
Acetone
48
Toluene
7.6
Ethyl acrylate
Xylene
0.35
Dichloromethane
0.002 706
Figure 18.1.7 shows the interaction between odorant concentration and odor intensity in an artificial example.41 Odor intensity is assessed by means of a scale from 1 to 6 [VDI 3882/1]39 (Table 18.1.9). Table 18.1.9. Correlation of odor with an intensity level Odor
Figure 18.1.7. Interdependence of odorant concentration and odor intensity in an artificial example.
Intensity level
Extremely high
6
Very high
5
High
4
Moderate
3
Low
2
Very low
1
No scent
0
Hedonic odor tone An odor may be quantified by determining the odor concentration, the character of the odor (pleasant or unpleasant) is not considered. Unpleasant odors may result in deteriorating health and should therefore be avoided. By determining the hedonic odor effect, the emotional reaction initiated by an irritation to the sense of smell may be included. It should not be confused with the kind of smell (it smells like ...) or with the odor intensity (it smells “strong” to Figure 18.1.8. Evaluation of the hedonic effect of an odor “weak”). It may be determined not only for 40 sample [VDI 3882/2].
1222
Margot Scheithauer
a single odorant, but also for an odorant mixture. Odor samples of an odorant concentration above the odor threshold are ranked by evaluators according to the odorants’ position in the following response range between “extremely unpleasant” to “extremely pleasant”. In many cases of odor evaluation, there are big differences between the results among the individual evaluators, since they have a widely varied background of experience. 18.1.6 MEASUREMENT OF ODOR IN MATERIALS AND INDUSTRIAL PLANTS 18.1.6.1 Introduction Various methods are known for determining odors. One of particular interest is pupil dilatation. This is based on the fact, that pupils dilatate when a strong odor is sensed, as may be caused by concentrated ammonia. A simple sampling of odors is based on the Öko-Tex Standard 200.42 The samples are conditioned in a desiccator and their odor is subsequently judged by 6 evaluators according to a five-point scale: 1 odorless 2 low odor 3 bearable odor 4 annoying odor 5 unbearable odor Such a scale also serves the “Deutsche Gütegemeinschaft Möbel” (German Quality Association of Furniture Manufacturers) as a methodical approach for evaluation of odors of furniture, cover fabric, leather, etc. Level 3, however, is defined as “commodity typical”. It needs to be achieved in order for the respective product to have the attribute “tested for noxious substances”. In principle, odors may be determined by means of sample recognition with the help of arrays of gas sensors, so-called electronic noses. Unknown samples are compared with known samples. Hence, olfactometric investigations need to precede. Such a measuring device is applicable only in a specific case and has to be trained prior to use. Odor can become a controllable quality feature of a product. Samples of good quality can be made distinguishable from samples of bad quality. In practice, odor determination by means of the olfactometer has been widely applied and generally adopted. 18.1.6.2 Odor determination by means of the “electronic nose” The principle of odor determination in different fields of application has been discussed in detail by Moy and Collins43 and Schulz.44 Measuring principle: The substance mixture in question, which causes the odor, undergoes an overall investigation by means of a sensor array. In doing so, the device makes use of the semi-conductive properties of various metal oxides, which are on the surface of 12 sensors which react to the gaseous substances which come in contact with them. The response also depends on the temperature, humidity and flow rate of the gas. Each sensor issues its own reply signal as soon as the sample of air touches its surface. The measurement is done within about two minutes, readjustment of the sensors takes about 4 minutes. The measuring device is linked to a computer with relevant software. Each odor is, according to the twelve sensors, represented by 12 graphs showing a characteristic “profile” (“fingerprint”). These sensor-specific, time-dependent, series of
18.1 Measurement and estimation of solvents emission and odor
1223
electric conductivity data, which are produced by the presence of the odorous substances provide the data for the evaluation. They are processed by means of neuronal networks. The available odor samples can be shown as two or three-dimensional. Comparative or referential samples are treated in the same way. By means of sample series, they may serve the purposes of the identification of new samples. It remains an issue, that odor-relevant compounds exist in much lower concentration in most cases than the less odorous compounds, thus they are also less influential on the sensor signal. Much research effort has yet to be undertaken to apply the electronic nose in the future. 18.1.6.3 Odor determination by means of the olfactometer The method of odor determination by means of the olfactometer is based on the guidelines Richtlinien VDI 3881, pages 1 to 435-38 and VDI 3882, pages 1 and 2.39,40 Measuring principle: An odorant sample is diluted in a defined way with neutral air in an olfactometer and offered to test persons as an odor sample. The test persons are exposed to several dilution levels. Should an odor be perceived, it shall be confirmed by Figure 18.1.9. Principle of an olfactometer. 1 diluent air pressing a button. An olfactometric detera) outer air, b) synthetic air (bottle); 2 acitvated charcoal filter air from 2d to rinse by-pass, 3 glass bulb pump with mination requires at least 5, and preferably electronically controlled drive (a,b); 4 mix chamber 8 evaluators. These evaluators need to pass (with installation a); 5 evaluator’s mask; 6 sample air (if a suitability test in accordance with Guidenecessary pre-mixed); 7 glass bulb pump with electroni- line VDI 3882, p. 2.40 The principle of an cally controlled drive (a,b); 8 by-pass valve; 9 mix 45 chamber valve; 10 control terminal, a) display olfactometer is shown in Figure 18.1.9. The results are presented in odor (programme step), b) display, c) terminal keyboard; 11 printer: I - IV control line; 12 additional pump (24 l x units/m3 of neutral air and they are automatmin-1). ically displayed as the averaged data of the panel’s evaluation. 18.1.6.4 Example for odor determination for selected materials: Determination of odorant concentration in varnished furniture surfaces The odor potential of furniture is determined primarily by the applied varnishes although adhesives also play a minor role. Furniture varnishes may contain up to 80% solvents. The residual solvents remaining in the varnish thus determine the VOC and odorant concentration of furniture surfaces. A specific test method has been developed for determining the VOC and odorant concentration of coated surfaces,46-48 as shown schematically in Figure 18.1.10. Similar to formaldehyde determination,49 sample testing is done using a test chamber approach. After some atmospheric conditioning, of the varnished furniture, the samples are stored in a test chamber (typically 1 m3) under the following conditions: 23°C±0.5; 45±3% relative humidity; charged with 1 = 1 m2 of emitting sample surface per 1 m3 test chamber volume air flow rate at sample surface: between 0.1 and 0.3 m/s. The samples remain in the test chamber for
1224
Margot Scheithauer
an initial 48 hours without an air exchange in the test chamber, in order to simulate a “worst case scenario”, i.e., that, for instance, the new furniture is sealed off air-tight during transport, so that residual solvents accumulate. After 48 h of storage time without an air exchange, air samples are taken in parallel for the determination of VOC emission by means of gas chromatograph and for determination of odor by the olfactometer. Subsequently, there is an air exchange of one air change per hour, and the Figure 18.1.10. Test approach for the determination of VOC and slow-down curve of VOC and odorant concentration in coated surfaces. odorant concentration is determined. Figure 18.1.11 shows the decreasing VOC concentration depending on the storage time in the test chamber with an air exchange 1,50,46 and Figure 18.1.12 shows the decreasing odorant under the same test conditions.50,46 The curves for VOC and the odorant concentration follow a similar pattern, they are, however, not identical. While the VOC concentration in water varnish is the lowest after 11 days, the odorant concentration after this time is equally high with both DD-varnish and water varnish. This, however, does not generally apply to these types of varnish. Also DD-varnishes of different origins may differ greatly in emitting residual solvents. In single-solvent systems, it is easy to see the interdependence between VOC and odorant concentration. This does not necessarily apply to solvent mixtures. This is due to the large variations in odor thresholds and the different evaporating behavior of the various solvents. Acrylate varnishes, for example, need contain only a few µg of ethyl acrylate in order to produce odor, since odor threshold level is at 0.002 mg/m3. Thus the determination of the
Figure 18.1.11. VOC concentration depending on the storage time in the test chamber with air exchange 1.
Figure 18.1.12. Decreasing odorant concentration depending on the storage time in the test chamber with air exchange 1.
18.1 Measurement and estimation of solvents emission and odor
1225
VOC concentration alone is not sufficient for determining the quality of coated surfaces must also be evaluated. 18.1.6.5 Example of odor determination in industrial plants: Odor measurement in an industrial varnishing plant In a study parallel to the measurements of the VOC emission at the varnishing lines of a furniture company, as described in Section 18.1.4.3.2, 1st Example, odorant concentrations were sampled in the exhaust air from the individual chimneys by means of olfactometry.29 All conditions of measurement described in Section 18.1.4.3.2 also apply to the determination of the odorant concentration. The results are contained in Table 18.1.5 as odorant concentration in GE (odorant units)/m3 and in the mass flow in TGE/h (one thousand odorant units per hour). The evaluation of the analytical determination of individual components resulted in the main components being butyl acetate and ethyl acetate. Due to the very low odorant threshold of butyl acetate (0.03 mg/m3), as compared to other available substances, (ethyl acetate 22 mg/m3), butyl acetate may be assumed to be responsible for for the bulk of detectable odor. An immediate comparison of samples taken at the same point in time resulted in the following data: concentration of butyl acetate: 8 mg/m3 odorant concentration: 115 GE(odorant units)/m3 Adopting the value quoted in the literature for the odorous threshold of butyl acetate at 0.03 mg/m3, which equals one odorant unit, the odorant concentration estimates a butyl acetate concentration of about 4 mg/m3. This value is in the same range as the value established for butyl acetate by gas chromatography. In these comparisons one has to consider that odorant thresholds may deviate by one to two exponents to even ten, according to literature sources.51 Measurements taken in the housing area about 300 m away from the chimneys of the varnishing plants could not establish any solvent-typical components in the gas analysis. The maximum concentration value for TVOC was at 0.004 mg/m3. Also the spreading calculation, which was carried out on the basis of the determined solvent concentration, showed that there should be no significant odor annoyance in the vicinity of the emission source. The results of the spreading calculation on the basis of measured solvent concentrations are considerably more precise, since solvent concentrations may be determined more exactly than odorant concentrations. The spreading calculation was done using the software package “IMMPROG-P” by AIRINFO AG, Switzerland, and carried out according to a method of the Odor Emission Guideline of the German State of Northrhine-Westfalia/Geruchsimmissions-Richtlinie des deutschen Bundeslandes Nordrhein-Westfalen.34 REFERENCES 1 2 3
4
Council Directive 1999/13/EC of March 11, 1999 of the limitation of emissions of volatile organic compounds due to the use of organic solvents in certain activities and installations. ISO/DIS 4618-4: 1999-12-03, Paints and varnishes - Terms and definitions for coating materials - Part 4: Terms relating to raw materials. Technische Regeln für Gefahrstoffe TRGS 610 “Ersatzstoffe, Ersatzverfahren und Verwendungsbeschränkungen für stark lösemittelhaltige Vorstriche und Bodenbelagsklebstoffe”; Absatz 2.7, 04.06.1992. M. Scheithauer, I-Lack 66, 325 - 331 (1998).
1226
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
27 28 29 30 31 32 33 34
35 36 37 38 39 40 41
Margot Scheithauer
ISO/DIS 11890-1, Draft 03/1998, Paints and varnishes - Determination of volatile organic compound (VOC) content. Part 1: Difference method (DIN ISO 11890-1). ISO/DIS 11890-2, draft 03/1998, Paints and varnishes - Determination of volatile organic compound (VOC) content - Part 2: Gas chromatographic method (DIN ISO 11890-2). Lösemittelverordnung 1995 in Österreich 872.VO im Bundesgesetzblatt für die Republik Österreich. Lackieranlagen-Verordnung 1995 in Österreich 873.VO im Bundesgesetzblatt für die Republik Österreich. ASTM D 3960-98 (1998-11-10). Standard Practice for Determining Volatile Organic Compound (VOC) Content of Paints and Related Coatings. Farbe und Lack, 105, 12/99, 111 (1999). B. Seifert, Richtwerte für die Innenraumluft; Bundesgesundheitsblatt - Gesundheitsforschung Gesundheitsschutz 3; Springer-Verlag, Berlin, 1999, pp. 270 - 278. J. Witthauer, 4. Freiberger Polymertag, Freiberg, May 27-28, 1999, conference paper A/19. A. Szalai, The use of Time: Daily activities of urban and suburban populations in twelve countries, Den Haag, Paris: Mouton, 1972. F. S. Chapin, Human activity patterns in the city, Wiley - Interscience, New York, 1974. EUWID Möbel 24, 5 (1994). F. Busato, Mod. Paint Coat., March 97, 30 - 33 (1997). H. Zeh, 4. Freiberger Polymertag, Freiberg, May 27-28, 1999, Conference paper P/2. Lackieranlagen-Verordnung in Deutschland (Varnishing Plant Regulation): Umweltbundesamt Deutschland III 3.2 - 52337/7); draft April 1, 1996. Deutsches Institut für Gütesicherung und Kennzeichnung e.V.: “RAL-UZ 38: Emissionsarme Produkte aus Holz und Holzwerkstoffen” Eigenverlag, St. Augustin, 1999, p.4. M. Große Ophoff, Holz- und Kunststoffverarbeitung, 12/96, 52-55 (1996). ASTM D 3960 - 98: Standard Practice for Determining Volatile Organic Compound (VOC) Content of Paints and Related Coatings, 1998. ISO 3251: 1993. Paints and varnishes - Determination of non-volatile matter of paints, varnishes and binders for paints and varnishes. ISO 760: 1978. Determination of water - Karl-Fischer method (general method). E DIN 55 649: 1998 - 10. Paints and varnishes - Determination of volatile organic compounds content in waterthinnable emulsion paints (In-can VOC); draft: October 1998. Richtlinie VDI 3481 / Blatt 3: Gaseous emission measurement - Determination of volatile organic compounds, especially solvent, flame ionisation detector (FID), 10.95. Richtlinie VDI 3481 / Blatt 2 E: Messen gasförmiger Emissionen (Gaseous emission measurement) Bestimmung des durch Adsorption an Kieselgel erfassbaren organisch gebundenen Kohlenstoffs in Abgasen; 11.96. Dräger, Dräger-Röhrchen Handbuch, Lübeck, 1991, 249. Richtlinie VDI 3482/Blatt 1. Gaseous air pollution measurement; gas-chromatographic determination of organic compounds, fundamentals; 02.86. M. Broege, Gutachten zu von Lackieranlagen verursachten Geruchsimmissionen, Institut für Holztechnologie Dresden, 1993. 1312 Photoacoustic Multi-gas Monitor, Product Data from INNOVA, Air Tech Instruments A/S, Denmark, 04/97. R. Kusian, M. Henkel, Forschungsbericht des Institutes für Holztechnologie Dresden, Germany (1997): Untersuchung der Emissionsverhältnisse bei der Oberflächenbehandlung von Holz und Holzwerkstoffen. M. Schön, R. Hübner, Geruch-Messung und Beseitigung, Vogel-Buchverlag, Würzburg, 1996. A. L. Lehninger, Biochemie,Weinheim-New York, Verlag Chemie, 1979, p. 19. Feststellung und Beurteilung von Geruchsimmissionen (Geruchsimmissionsrichtlinie, GJR); Der Minister für Umwelt, Raumordnung und Landwirtschaft des Landes Nordrhein-Westfalen, Stand 15.02.1993, Anhang B.Richtlinie VDI 3881. Blatt 1: Olfactometry; odor threshold determination; fundamentals; 05 / 1986. Blatt 2: Olfactometry, odor threshold determination; sampling; 01/1987. Blatt 3: Olfactometry; odor threshold determination; olfactometers with gas jet dilution; 11/1986. Blatt 4: Olfaktometrie; Geruchsschwellenbestimmung; Anwendungsvorschriften und Verfahrenskenngrößen; 12/86 Richtlinie VDI 3882. Blatt 1: Olfactometry; determination of odor intensity; October 1992. Blatt 2: Olfactometry - Determination of hedonic odor tone, September 1994. M. Paduch, VDI-Berichte 1059, “Aktuelle Aufgaben der Messtechnik in der Luftreinhaltung” zum Kolloquium Heidelberg, Düsseldorf, Juni 2-4, 1993, pp. 593-607.
18.2 Prediction of organic solvents emission
42
43 44 45 46 47 48 49 50 51
1227
Öko-Tex Standard 200: Prüfverfahren für die Vergabe der Berechtigung zur Kennzeichnung von Teppichböden mit “Schadstoff geprüft nach Öko-Tex Standard 100" (Österreichisches Forschungsinstitut) 1992. L. Moy, M. Collins, LaborPraxis - Journal für Labor und Analytik, 20/5, 14-18 (1996). H. Schulz, 4. Freiberger Polymertag, Freiberg, May 27-28, 1999, conference paper C 4/5. Ströhlein, Labor-, Mess- und Umwelttechnik, Kaarst, Germany, Product Data p. 25, principle of an olfactometer. M. Scheithauer, K. Aehlig, M. Broege, Holz- und Kunststoffverarbeitung, 1/96, 58-61 (1996). M. Scheithauer, K. Aehlig, Konferenz im ITD, Poznan (Poland), 1995. K. Aehlig, M. Scheithauer, M. Broege, Holz, 5, 26-32; (1996). prENV 717-1: 1998: Holzwerkstoffe Bestimmung der Formaldehydabgabe Teil 1: Formaldehydabgabe nach der Prüfkammer-Methode (1998). M. Broege, K. Aehlig, 4. Freiberger Polymertag, Freiberg, Mai 27-28, 1999, conference paper R 4/5. G. Scharfenberger, Chemie in Labor und Biotechnik, 42, 498-502 (1991).
18.2 PREDICTION OF ORGANIC SOLVENTS EMISSION DURING TECHNOLOGICAL PROCESSES Krzysztof M. Benczek, Joanna Kurpiewska Central Institute for Labor Protection, Warsaw, Poland
18.2.1 INTRODUCTION The concentration of toxic substances in air during technological process is very important factor for occupational safety. Typical examples of processes, which have the potential to harm workers, are metal degreasing, painting, and wood impregnation. If metal processing involves several steps some of which may be done in more than one manufacturing facility, the semi-processed metal parts must be protected during transportation and storage. Such protective coatings of grease and rust preventatives must be removed in degreasing operation. For many processes (e.g., painting, galvanic metal deposition), clean surface is an important requirement. The cleaning process may be done in automated and enclosed equipment or it may be done manually in the open. The degreasing agent may be an organic solvent, a solvent blend, or a water solution, usually alkaline in nature. Depending on the process used the operation may pose no risk to the worker or be a serious occupational hazard. Similarly environmental emissions may be negligible or of serious concern. We present a method of evaluating the quantities of emissions from such processes which involve solvents. The method may be applied to such diverse operations as painting, wood preservation, impregnation of porous materials, gluing, cleaning, filing open tanks, general solvent handling operations, and many others. We have selected metal degreasing as a representative example to demonstrate how the method may be applied. 18.2.2 METHODS OF DEGREASING Six methods can be identified which differ in the degreasing agent used:
1228
Krzysztof M. Benczek, Joanna Kurpiewska
• degreasing in liquid organic solvents, such as naphtha, petroleum, chlorinated aliphatic hydrocarbons, etc.
• degreasing in hot vapors of halogenated solvents such as, trichloroethylene, perchloroethylene, fluorochloroethane and so on
• degreasing in alkaline water solutions of hydroxides, phosphates, surfactants, emulsifiers, common inhibitors, etc.
• degreasing in an emulsion of organic solvents in water • degreasing in water steam • supersonic degreasing in stabilized chlorinated hydrocarbons. Degreasing process can be conducted automatically in different ways by: dipping spraying using high pressure supersonic pulsating washing Process can be conducted in open or enclosed equipment. Manual degreasing is still very popular but it is very time and labor consuming, expensive, and large amounts of solvents are lost. 18.2.3 SOLVENTS In metal degreasing these solvents are most frequently used: • naphtha solvent, • naphtha anti-corrosive • 1-butanol, • 1,1,2-trichloroethylene, • 1,1,1- trichloroethane, • extraction naphtha, • petroleum(mineral) spirits. 18.2.4 IDENTIFICATION OF THE EMITTED COMPOUNDS In many cases, solvent mixtures are used. Their composition must be identified. Rodofos is one example of such solvent used in Poland. Its composition was determined by gas chromatography. Analyses were performed using a Hewlett-Packard gas chromatograph model 5890 coupled with computerized mass spectrometer instrument, model 5970. Capillary column 50 m x 0.32 mm i.d., d.f.= 0.52 µm FFAP and helium as a carrier gas was used at temperature of 40oC. Ions from 20 to 400 amu were counted, delay time was 3.5 min. Samples were collected by drawing a known volume of air through a bubbler containing 1 ml of carbon disulfide. Volume of the injected sample was 1-5 µl. Chromatograms are presented in Figure 18.2.1. 18.2.5 EMISSION OF ORGANIC SOLVENTS DURING TECHNOLOGICAL PROCESSES The concentration of substances emitted to the air during the degreasing processes reached the steady-state constant value:
• • • • •
C = E/q
[18.2.1]
18.2 Prediction of organic solvents emission
1229
Figure 18.2.1. Chromatogram of substances emitted during degreasing. Oven temperature 40oC. 1 - benzene, 2 2,2-dichloromethylpropane, 3 - butyl ester of formic acid, 4 - methylbenzene, 5 - water, 6 - 1-butanol, 7, 8, 10, 11, 12, 13, 14 - derivatives of methyl ethyl benzene, 9 - propyl benzene. where: C q E
concentration, mg/m3 sum of ventilation flow rates, m3/min emission, mg/min.
Thus, E = C×q
[18.2.2]
The value C can be measured, but a value for q is difficult to estimate, because it includes mechanical and gravitational ventilation (central air conditioning, influence of open doors, windows, fans and natural ventilation). When the process of degreasing ends, the decay of concentration can be described by: Ct = Ce × exp(-t × q/V)
[18.2.3]
where: Ce t V
concentration of the emission at the termination of the process, mg/m3, duration of process, min, room volume, m3.
After a transformation: q = (lnCe - lnCt) × V/t
[18.2.4]
The quantity emitted to a room of known volume depends only on the changes of concentration at time (t).
1230
Krzysztof M. Benczek, Joanna Kurpiewska
18.2.6 VERIFICATION OF THE METHOD Emission measuring setup is given in Figure 18.2.2. Cyclohexane was pumped (with E=6.123 mg/min) into glass bottle (V=44.5 l) equipped with a fan. The air in the bottle was analyzed by gas chromatography using a flame ionization detector - OVA 118. After 20 min., when the concentration reached 2.250 mg/l, the pump was stopped and the decay of concentration with time was measured. Results are given in Tables 18.2.1 and 18.2.2 (value E from the equation [18.2.2], q from the equation [18.2.4]). The standard deviation of the sample was 0.0106, of the method - 0.0102 and bias was 0.08 mg/min (1.3%). Tests were then performed in the macro-scale - emission of cyclohexane in a room, when the emission ended when the process stopped. The estimate is precise when the room volFigure 18.2.2. Emission measur- ume and ventilation flow rate are known. ing setup. The method was verified in the industrial environment in automotive plant. Here, the hazardous substances continue to be emitted to the air after the process had stopped. The concentration measured near the outTable 18.2.1. Emission of cyclohexane
Table 18.2.2. Ventilation flow rate
t, min
Ct, mg/l
E, mg/min
t, min
Ct, mg/l
Q, l/min
1
0.138
6.210
1
2.200
1.000
2
0.273
6.212
2
2.215
1.011
3
0.405
6.213
3
2.100
1.024
4
0.533
6.201
4
2.055
1.008
5
0.659
6.201
5
2.010
1.004
6
0.780
6.184
6
1.965
0.993
8
1.020
6.200
8
1.880
0.999
10
1.250
6.212
10
1.800
0.992
12
1.465
6.200
12
1.720
0.996
15
1.770
6.188
15
1.605
1.002
20
2.250
6.218
20
1.435
E ave. - 6.203, σ n= 0.0102, σn-1= 0.0106
1.000
q ave. = 1.0026, σn = 8.7103×10 , σn-1 = 9.1354×10-3 -3
let of exhaust was 620 mg/m3, and because the exhaust flow rate was known to be 50 m3/min, the emission was precisely estimated according equation [18.2.2] as 31 g/min or 312.48 kg/week. This value was comparable with the average solvent loss - 326.22 kg/week.
18.2 Prediction of organic solvents emission
1231
18.2.7 RELATIONSHIPS BETWEEN EMISSION AND TECHNOLOGICAL PARAMETERS Emission of solvents depends on the evaporation rate of the solvent in the process. The evaporation rate from the surface depends on the concentration in the layer on the surface and the coefficient of mass transfer on the air-side. This relation is approximately true for degreasing operations using both liquid and vapors of organic solvents. The concentration of solvent in a saturated vapor layer depends on temperature and vapor pressure. The coefficient of mass transfer on the air-side depends on the air velocity in the layer on the surface and Schmidt's number (includes dynamic vapor viscosity, vapor density, and diffusion coefficient). Emissions are measured in mass unit per unit of time and the amount depends on surface area and the rate of evaporation, which, in turn, depends on temperature, air velocity over the surface of solvent and the mass of solvent carried out on the wetted parts which have been degreased. 18.2.7.1 Laboratory test stand A thin-metal, flat dish 6 cm in diameter was filled with solvent up to 2-3 mm from the upper edge and placed on a laboratory balance (Figure 18.2.3). The amount of evaporated solvent was measured with (precision 0.1 mg) as the difference between the mass of the dish and solvent at the start of the test and the mass of the dish with solvent after pre-determined period of time. The test was repeated under different conditions of temperature and air velocity near the surface of the solvent. The results were reported as the evaporated mass per 1 minute. 18.2.7.2 The influence of temperature on emission Temperature was measured with a mercury thermometer with a range from 0 to 30oC. Air velocity was 0.3 to 0.4 m/s in this temperature region. The relationship is linear and the equations expressing emission [g/m3/h] relative to temperature [oC] for different solvents are: trichloroethylene E= 63.6 × t + 699.6 naphtha solvent E= 9.3 × t + 4.9 naphtha anti-corrosive E= 0.55 × t + 5.65 1-butanol E= 6.29 × t + 16.9 1,1,1- trichloroethane E= 94.7 × t + 805.6 extraction naphtha E= 49.5 × t + 1147 petroleum(mineral) spirits E= 11.9 × t + 76.7 wood preservatives E= 2.0 × t + 42
Figure 18.2.3. Test laboratory stand.
1232
Krzysztof M. Benczek, Joanna Kurpiewska
18.2.7.3 The influence of air velocity on emission Air velocity was measured with a TSI air velocity meter in range 0.2- 1.5 m/s. Temperature was 20oC. The relationship has linear character and the equations expressing emissions [g/m3/h] relative to air velocity [m/s] for different solvents are: trichloroethylene E= 657 × v + 1766 naphtha solvent E= 109 × v + 159 naphtha anti-corrosive E= 9.1 × v + 14.53 1-butanol E= 72 × v + 121 1,1,1- trichloroethane E= 1403 × v + 2120 extraction naphtha E= 632 × v + 830 petroleum(mineral) spirits E= 160 × v + 242 wood preservatives E = 516 × v 18.2.7.4 The relationship between the mass of solvent on wet parts and emissions The amount of solvent carried out on a degreased part depends on its surface, shape and roughness and on the viscosity and density of the solvent. The amount of solvent carried out on a degreased surface was measured by gravimetric method. Tests show that the amount of solvent retained on 1 m2 of flat surface is: trichloroethylene 34.7 g naphtha solvent 26.0 g naphtha anti-corrosive 34.5 g 1-butanol 31.3 g 1,1,1- trichloroethane 38.5 g extraction naphtha 19.0 g petroleum(mineral) spirits 19.2 g 18.2.8 EMISSION OF SOLVENTS Combining these factors, the final equations for the evaluated solvents are: trichloroethylene naphtha solvent naphtha anti-corrosive 1-butanol 1,1,1- trichloroethane extraction naphtha petroleum(mineral) spirits wood preservatives
E= (657v + 63.6t + 699.6) × Pr + 34.7P E= (109v + 9.3t + 4.9) × Pr + 26P E= (9.1v + 0.55t + 5.65) × Pr + 34.5P E= (72v + 6.3t + 17) × Pr + 31.3P E= (1403v + 94.7t + 805.6) × Pr + 38.5P E= (632v + 49.5t + 1147) × Pr + 19P E= (160v + 12t + 76.7) × Pr + 19.2P E= (516v + 2t + 42) × Pr
[18.2.5] [18.2.6] [18.2.7] [18.2.8] [18.2.9] [18.2.10] [18.2.11] [18.2.12]
where: E v t Pr P
emission of the solvent, g/h air velocity near the surface of solvent (in still air it is 0.3 m/s) temperature, oC the surface of solvent, m2 the surface of elements degreased in one hour, m2/h.
18.2.9 VERIFICATION IN INDUSTRIAL CONDITIONS Research was conducted in an industrial hall where an ALDI machine was used for degreasing with trichloroethylene vapor. The volume of the solvent was 1000 dm3, the surface area was 0.825 m2 (110 cm by 75 cm). A metal basket with degreased parts was intro-
18.2 Prediction of organic solvents emission
1233
duced into the degreasing compartment and vapors of trichloroethylene condensed on surface of the parts. After 30 s, the basket was removed from the machine. The average temperature at the surface was 24oC. There was an exhaust in the hall with a flow rate of 50 m3/min. The solvent loss was 30 l/day and it was independent of the number of degreased elements. The solvent loss of 30 l/day equals 1.25 l/h or 1.825 kg/h with density of trichloroethylene equal 1.46 g/cm3. The average concentration of trichloroethylene measured near the exhaust by portable IR spectrophotometer-Miran was 680 mg/m3. Calculation (equation [18.2.2]) with q=50 m3/min gives emission of 34 g/min or 2040 g/h. According to equation [18.2.5] for trichloroethylene: E= (657v + 63.6t + 699.6) × Pr + 34.7P v=0.3 m/s (still air near the surface) t=24oC Pr=0.825 m2 In vapor degreasing, the coefficient of mass carried out on wet details ready to ship P=0. Then E= (657 × 0.3 + 63.6 × 24 + 699.6) × 0,825 = 1999 g/h The measured value of 2040 g/h and the calculated value of 1999 g/h are in good agreement, meaning that the equations can be successfully applied to predict the organic solvents emission during process of automatic degreasing. Process of manual degreasing was verified for washing motor parts in a metal dish 0.72 m2 filled with 500 l of extraction naphtha. The surface of parts was 0.227 m2. This process took place in an open garage at temperature 14oC. Air velocity in vicinity of parts was 0.48 m/s. The residual solvent volume after degreasing was 4.58 l. The process lasted 7 minutes. Emission estimated according the loss of the solvent during the process was 420 ml. Taking into account the weight density of naphtha - 0.72 g/cm3, emission was equal 302.4 g/7 min or 2592 g/h. Applying the equation [18.2.10] for extraction naphtha E= (632v + 49.5t + 1147) × Pr + 19P and: v = 0.48 m/s t = 14oC Pr = 0.72 + 0.227 = 0.947 m2, surface of the dish and details P = 0.947 m2/7min [8.12 m2/h], solvent carried out on the details and on the surface of the dish after pouring out extraction naphtha Thus: E= (632 × 0.48 + 49.5 × 14 + 1147) × 0.947 + 19 × 8.12 = 2184 g/h Again measured value of 2592 g/h and the calculated value of 2184 g/h were in good agreement thus the equation can be applied for predicting the organic solvents emission during the processes of manual degreasing.
1234
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
18.3 INDOOR AIR POLLUTION BY SOLVENTS CONTAINED IN PAINTS AND VARNISHES Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Thbingen, Thbingen, Germany
Gerhard Volland Otto-Graf-Institut, Universit¬t Stuttgart, Stuttgart, Germany
18.3.1 COMPOSITION - SOLVENTS IN PAINTS AND VARNISHES. THEORETICAL ASPECTS Solvents are normally defined as fluids having a maximum b.p. of 250oC, which are able to dissolve other components of paints, especially binders. They evaporate under drying conditions when paint films are formed.1,2,3,4 Solvents must not react with the painted or varnished product. The composition of paints and varnishes is determined by application requirements, drying temperature, and drying time.2,5 Depending on the properties of paints and varnishes, different mixtures of solvents are added. Functions and properties of solvents in paints and varnishes:2,5 • Dissolve several components, especially binders • Influence and control paint viscosity • Wet pigments (influence on solubility, hydrogen bonding by solvents, prevent the separation of pigment) • Influence and control flow properties (e.g., butyl acetate, butanol, glycol ethers) • Influence skin formation.6 The aim is to produce a homogeneous cure when the paint or varnish film hardens without the formation of a stable surface film during the drying period. The correct composition of the solvent will avoid trapping solvents under the surface film. • Influence drying process. Acceleration by low boilers, production of a flawless surface by medium and high boilers (chemical and physical drying processes) • Influence surface tension (e.g., increase by rapid evaporation of solvents) • Influence mechanical properties of paints (e.g., adhesion properties) • Influence blushing or blooming of paints by preventing the absorption of condensed water by various solvents, e.g., ethanol and glycol ether • Influence gloss (e.g., improvement with high boilers) • Prevent defects particularly in varnish coating (e.g., background wetting) • Influence electrostatic properties (spray painting) • Influence defined surface properties (structural change) • Influence durability of paints and varnishes • Influence product suitability, e.g., spraying and dipping lacquers which need to dry at room temperature In addition to their effect on performance and properties solvents interact with other components in paints and varnishes in significant ways. Interactions between binders and solvents in paints and varnishes are very important.2,5 With the aid of solubility parameters solvents or mixtures of solvents which produce
18.3 Indoor air pollution by solvents
1235
the required properties may be selected. The influence extends to dissolving of binders, reduction of paint viscosity, pigment wetting, etc. Optimized dissolving of binders can be achieved by selecting the appropriate solvent mixtures, in which the density approaches that of the binder solubility range. Nevertheless the selection of an optimized solvent mixtures is difficult because there are conflicting requirements and outcomes.2,5 On one hand, the chemical nature of the solvents should be similar to those of the binder to improve the flow but on the other hand, the solubility and hydrogen-bonding abilities of the solvents should be at the edge of the binder solubility range, because this results in rapid drying with low retention of solvents. 18.3.2 OCCURRENCE OF SOLVENTS IN PAINTS AND VARNISHES 18.3.2.1 Solvents in products Classification Authorized commissions in Germany describe various groups of solvents in paints and varnishes: The Association of Varnish Industry VdL (Verband der Lackindustrie) classifies solvents generally used in paints and varnishes in the following groups:4 Aliphatics, alcohols, aromatic hydrocarbons, esters, glycols, glycol ethers, ketones, terpenes, gasoline, water. The commission of Hygiene, Health and Environmental Protection (Hygiene, Gesundheit und Umweltschutz) which belongs to the German Standards Commission DIN (Deutsches Institut für Normung e.V.) also describes solvents which may be present in paints as follows:10 • Aliphatics (white spirit, turpentine substitutes, cyclohexane) • Aromatic hydrocarbons (toluene, xylenes, naphtha, styrene) • Alcohols (methyl alcohol, ethyl alcohol, propyl alcohol, butyl alcohol, glycols) • Ketones (acetone, methyl ethyl ketone, methyl isobutyl ketone, cyclohexanone) • Esters (methyl acetate, ethyl acetate, butyl acetate) • Others: methylpyrrolidone, oils of turpentine Chlorinated hydrocarbons are not contained in the above list. Benzene is not included as it is obsolete, although it is sometimes found in some petroleum solvent-based paints. Shortcomings The description of paints and varnishes is usually neither complete nor reliable: • Instructions and comments from manufacturers and suppliers differ in quantity and quality. Usually manufacturers or suppliers are not committed to indicate the exact details of their product’s composition. They often omit information (e.g., information concerning product components, additives or by-products). • Lack of standards. In Europe there are some mandatory standards concerning products used in construction (e.g., asbestos, formaldehyde in particle board, organic solvents in certain activities and installations are prohibited). • Solvents or other organic compounds may be present, in low concentrations, in some products (even in water-based paints), e.g., as additives or by-products although there is no reference to them. • Additives (e.g., low molecular compounds such as esters or glycol ethers with high boiling points) serve as aids for the formation of films, in repelling water, in assisting plasticization or for other functions. In solvent-based paints, additives or diluting agents are often intentionally mixed in.
1236
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
Table 18.3.1. Systematology of paint and varnish products (after reference 10) 1. Paint coatings 1.1 Coloring agents 1.1.1 soluble pigments 1.1.1.1 natural pigments 1.1.1.2 synthetic pigments 1.1.2 insoluble pigments 1.1.2.1 inorganic pigments 1.1.2.2 organic pigments 1.1.2.2.1 animal and vegetable pigments 1.1.2.2.2 synthetic pigments 1.2 Binders 1.2.1 water dilutable binders 1.2.1.1 slaked lime (lime colors) 1.2.1.2 standard cement (cement colors) 1.2.1.3 sodium silicate (colors of 1 or 2 components) 1.2.1.4 vegetable glues (limewash) 1.2.1.5 casein (lime-casein products, alkali-lime products) 1.2.1.6 dispersions 1.2.1.6.1 natural resin emulsion paints 1.2.1.6.2 plastomer emulsion paints (PVAC (homopolymers, copolymers), PVP, polyacrylates (PMMA, styrol-acetate)) 1.2.1.6.3 water emulsifiable varnish systems (aqueous acrylate systems, aqueous polyurethane systems) 1.2.2 solvent dilutable binders 1.2.2.1 oil paints 1.2.2.2 varnishes 1.2.2.2.1 products drying by air oxidation (nitrocellulose varnishes, aldehyde resin varnishes, oil varnishes) 1.2.2.2.2 physical drying products (polymer resin varnishes, polyvinyl chloride varnishes, polyvinyl acetate varnishes, polyacrylate varnishes, chlorinated rubber varnishes) 1.2.2.2.3 chemical curing products (phenolic varnishes, aminoplast varnishes, melamine resin varnishes) 1.2.2.2.4 acid-curing varnishes 1.2.2.2.5 epoxy resin varnishes 1.2.2.2.6 polyurethane varnishes 1.2.2.2.7 unsaturated polyester varnishes
• Residuals or by-products can result from various processes (e.g., residual monomers in a polymerization processes).
• Substances which can be classified as solvents are not always exactly defined, e.g., • •
substances in paints with maximum boiling point above 250oC. Plasticizers and solvents cannot often be strictly separated (e.g., various SVOCs). Analytical problems and difficulties in assessment (e.g., mixtures of substances, very low concentrations). Even new products are often not well documented (e.g., concerning composition, see above).
18.3 Indoor air pollution by solvents
1237
18.3.2.2 Paints and varnishes Definitions and systematology of paints and varnishes are given by various authorized commissions, e.g., European Committee for Standardization (CEN)1 and DIN-Commissions9 (see also Table 18.3.1). Various databases list the paint and varnish products and their composition generally applied in Europe, see Table 18.3.2). Table 18.3.2. Types of paints and varnishes mainly used in Europe - product groups, important compounds, applications (after references 2,5,11,23,25) Product groups
Compounds
Application
Silicate paints/prod- potash-waterglass (binder), inorganic/organic pigwall paints, outside house ucts, emulsion paints ments, mineral fillers, synthetic resins (dispersions, paints DIN 18363 (M-SK01) stabilizers), water (dispersing agent) water- and weatherproof Silicate paints/prod- potassium silicate (binder, fixing solution), inorganic painting, interior coating ucts, (M-SK02) pigments, adjuvants (resistant to chemicals) Emulsion paints (M-DF02)
synthetic resins (dispersions), mineral fillers, inorganic/organic pigments, water (dispersing agent), ad- outside house paints, interior ditives (film-forming agents < 3 %: e.g., glycol-ethers, coating (wall paints) esters, glycols, hydrocarbons), formaldehyde < 0,1 %
synthetic resins (dispersions), mineral fillers, inorEmulsion paints, ganic/organic pigments, water (dispersing agent), adinterior coating (wall paint) solvent-free (M-DF01) ditives (film-forming agents: no solvents), formaldehyde < 0,1 % Emulsion paints, outside house paint, water dilutable
emulsions of plastic materials (acrylate, vinyl), pigments, water (dispersing agent), organic solvents: < 0,1-4 %, (glycols, glycol-ethers, mineral varnish)
Emulsion paints, varnishes (M-LW01)
synthetic resins, alkyd resins, copolymerizates, polyurethane resins, inorganic/organic pigments, mineral fillers, additives, water (dispersing agents), organic various applications solvents: 5-10 %, glycols, glycol-ethers, esters, mineral varnish
Aldehyde resin varnishes, aromatic compounds (M-LL03)
aldehyde resins (binders), inorganic/organic pigments, fillers, 30-55 % solvents: mineral varnish (mixture of covering varnishes, primers hydrocarbons), other solvents (< 10 % esters, ethers, alcohols)
Aldehyde resin varnishes, low levels of aromatic compounds (M-LL02)
aldehyde resins (binders), inorganic/organic pigments, fillers, 30-55 % solvents: mineral varnish (mixture of covering varnishes, primers hydrocarbons), other solvents (< 10 % esters, ethers, alcohols)
Aldehyde resin no aromatic substances (M-LL01)
aldehyde resins (binders), inorganic/organic pigments, fillers, 30-55 % solvents: mineral varnish (mixture of covering varnishes, primers hydrocarbons), other solvents (< 10 % esters, ethers, alcohols)
exterior coating
1238
Product groups
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
Compounds
Application
Polymer resin paints, copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral high levels of aromatic % solvents: mineral varnish (mixtures of hydrocarbackground) compounds (M-PL03) bons), other solvents (< 10 %, esters, ethers, alcohols) Polymer resin paints, copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral backlow levels of aromatic % solvents: mineral varnish (mixtures of hydrocarground) compounds (M-PL02) bons), other solvents (< 10 %, esters, ethers, alcohols) Polymer resin paints, no aromatic compounds (M-PL01)
copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral back% solvents: mineral varnish (mixtures of hydrocarground) bons), other solvents (< 10 %, esters, ethers, alcohols)
Polymer resin paints, copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral backdilutable with solvents % solvents: mineral varnish (mixtures of hydrocarground) (M-PL04) bons), other solvents (< 10 %, esters, ethers, alcohols)
Natural resin paints
natural resins (e.g., shellac) or chemical modified natural resins (e.g., colophonium derivates), additions (e.g., methyl cellulose, natural latex, casein), inorganic, organic pigments (mainly natural origin), min- various applications eral fillers, additives (organic solvents: alcohols, terpenes, oil of turpentine, limonenes), essential oils (eucalyptus oil, oil of rosemary, oil of bergamot)
natural resins (e.g., shellac) or chemical modified natural resins (e.g., colophonium derivates), additions Natural resin paints, (e.g., methyl cellulose, natural latex, casein), inor- various applications solvent-free ganic, organic pigments (mainly natural origin), mineral fillers, additives (see above): < 1 % Oil paints, terpene products (M-LL04)
oils (linseed oil, wood oil, soya oil), natural resins and modified natural resins, mineral pigments, wetting covering varnishes, primers agent, flow improver, solvents: oils of turpentine, isoaliphatics, terpenes (citrus, orange)
oils (linseed oil, wood oil, soya oil), natural resins and Oil paints, terpene-free modified natural resins, mineral pigments, wetting covering varnishes, primers (M-LL05) agents, flow improvers, solvents: isoaliphatics (dearomatized) Oil paints solvent-free standard oils, calcium carbonate, pigments, siccatives, exterior coating (paneling) No. 665 lemon oil water-soluble Clear lacquers/glazing composition (wood), low levels of aromatic compounds (M-KH03)
alkyd resins, nitro resins, polymer resins, pigments, fillers, 40-50 % solvents: mineral varnish (mixtures of hydrocarbons), other solvents (< 10 %, esters, ethers, alcohols)
Lead chromate enamels, aromatic compounds
synthetic resins (e.g., aldehyde resins, PVCpolymerizates) inorganic/ organic pigments (lead priming coat (steel, alumichromate up to 20 %), fillers, 30-55 % solvents (mixnum, zinc) tures of mineral varnish, glycol-ethers, aromatic compounds)
interior coating (clear lacquers), exterior and interior coatings (glazing compositions, wood)
18.3 Indoor air pollution by solvents
Product groups Silicone resin products, water dilutable (M-SF01)
1239
Compounds
Application
emulsions of silicone resins mineral fillers, inorganic/organic pigments, water (dispersing agent), additives (film forming agents < 3%)
Solvent composition is an important aspect in classifying paints and varnishes.5 Main groups are: • Solvent products. These products contain solvents of various mixtures, types and concentrations depending on the properties desired (e.g., application method, surface film or skin formation, see above 18.3.1, 18.3.2.1, Table 18.3.1 and 18.3.2). Solvents are normally the main components of these products (for example up to 80 % in nitrocellulose varnishes, low solids). • Solvent reduced products. These contain solvents in lower concentrations compared to conventional products and hence have a higher content of solids.1,2 The main groups of solvent reduced paints are medium solid contents (solids 55-65 %) and high solid contents (solids 60-80 %). • Waterborne coatings. This group contains deionized water as a dispersing agent. Normally cosolvents are added (up to 25 %). The term “waterborne coatings” is mainly applied to industrial coating materials, which differ from silicate colors, wood preservative varnishes and emulsion paints. • Solvent-free products. The products of this group are produced and applied without organic solvents: powder coatings, radiation curing systems, solvent-free water coating (without cosolvents). The use of paints and varnishes containing high solvent concentrations is becoming less common, solvent-reduced products, waterborne coatings and solvent-free products are applied more often. Whereas environmental and health-related concerns call for the reduction of solvents in paints and varnish products (see 18.3.3.2 and 18.3.4), qualitative aspects still demand the use of solvents in some fields of application.12 Solvent-reduced products should achieve the same qualitative properties as solvent-containing products (e.g., application properties, periods of guarantee, limited costs, loading capacities, surface properties). The development of solvent-reduced or solvent-free varnish products with high quality (e.g., durable, good application properties) and limited costs must be encouraged if not mandated. In some fields of application (e.g., waterborne coatings, high solids in varnish coatings of vehicles) a lot of developmental work has already been done. Additionally, other components in varnishes apart from solvents or modifications of application techniques can improve the properties of solvent-reduced paint and varnish products. Nevertheless, a wide rage of quality exists in paints and varnish products which are offered commercially and, in some cases (e.g., concerning film forming processes, processibility, corrosion protection, purification, special applications or wood preservation), solvent-based products are still preferred. In the field of wood preservation especially, solvent-based products (alkyd resins) are used because of some technical advantages (e.g., more solid surfaces) but alternative high solid systems are available.12
1240
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
18.3.3 EMISSION OF SOLVENTS 18.3.3.1 Emission Solvents are usually the most significant emission products coming from building materials and interior furnishings.2,5,6,13 All painted products are potential sources of emission. Even the so-called “bio” paints or natural paints emit various substances,5 e.g., mineral varnishes, natural oils and even synthetic terpene-like compounds. Depending on the products and the components which make them up, the various parameters listed below can determine the emission and behavior of solvents in ambient air from paints and varnishes:2,5,6 • Film formation. During the film formation stage solvents are emitted. The emission rate is directly proportional to the VOC concentrations in the product and inversely proportional to the film thickness (first order of kinetics). When the film has completely formed, the emission is controlled by diffusion processes, the emission rate is now inversely proportional to the square of the film thickness. • Application of the paints and varnishes, methods of application of the paint or varnish, e.g. speed of application of the paint. • Characteristics of solvents in paints and varnishes, e.g., volatility (boiling point), dynamic characteristics of evaporation and concentrations. Substances having a low boiling point evaporate fast, mostly during application and cause a rapid skin formation. Thus the risk of exposure is mainly with the painters. Medium boilers allow the surface to remain open for a while (evaporation of volatile products). The evaporation of substances with a high boiling point is slow, taking several weeks or months after application, resulting in exposure to the building occupants. • Characteristics of other compounds in paints and varnishes (e.g., relationship of binders to solvents, possible reactions between solvents and other compounds). • Characteristics of surfaces which have been painted (e.g., area, structure of surface). • Characteristics of emission processes, e.g., type of emissions (e.g., diffusion), dynamics of emissions (constant of evaporation), interrelations (e.g., diffusion and back diffusion). The quantitative assessment of emission processes can be described with various models. The usefulness of these models differs. Some models describe these processes very well, as proven by various experiments or measurements (e.g., test chambers). Basic equations which describe emission processes are shown in Table 18.3.3. The emission processes of solvents from paints and varnishes can be divided into two phases:2,5,6 1. Emissions during application of paints. This deals with complex interrelations dependent on various parameters. 2. Emissions after application process. Here the course is governed by complex emission processes dependent on various parameters (e.g., film formation, surface area). Most solvent products, especially organic solvents and some additives, emitted from paints and varnishes are VOCs. The largest components of VOCs are solvents, e.g., aliphatic and aromatic hydrocarbons, alcohols, amines, acids, aldehydes, esters, ketones, terpenes. The definition of the term VOC varies, a standard definition is published by CEN (European Committee for Standardization):1 VOCs are any organic liquids and/or solids that evaporate spontaneously at the prevailing temperature and pressure of the atmosphere.
18.3 Indoor air pollution by solvents
1241
VOCC (volatile organic compound content) is defined as follows:1 Mass of the volatile organic compounds in a coating material, as determined under specified conditions. Table 18.3.3. Example for basic calculations of VOC-emissions during application of emulsion paints (after references 2,5,6,13) 1. At the beginning of the application process (t=0) the mass of VOC changes positively (vSTdscw) , on ht e other hand VOC evaporates (first order of kinetics). dmw/dt = vSTdscw - klmw/ds [18.3.1] 2. In the ambient air the mass of VOC increases because of the evaporation out of the wall and decreases according to the ventilation rates. dmL/dt = k1mw/ds - k2mL [18.3.2] 3. If the connected differential equations are solved and integrated (from t=0 until the end of application t=A/vST), the following equations are received: mw(t) = vSTds2cw/k1(1-exp(-klt/ds)) [18.3.3] and [18.3.4] mL(t) = ((1-exp(Bt)/B - (1-exp(k2t)/k2)vSTdscwexp(-k2t)) with B = k2 - k1/d2 [18.3.5] 4. After finishing application only evaporation is relevant (equation [18.3.1] is simplified): dmw/dt = - k1mW/ds [18.3.6] 5. The course of VOC in the ambient air does not change (equation [18.3.2] corresponds to equation [18.3.7]): dmL/dt = k1mw/ds - k2mL [18.3.7] 6. The solution of these differential equations describes the quantities of VOC in the wall (equation [18.3.8] and the course of VOC in the ambient air [18.3.9]: mw(t) = mw,AEexp(-k1(t-tAE)/ds) [18.3.8] mL(t) = ((k1mw,AE/ds) (exp(B(t- tAE))-1)/B + mL, AE) exp(-k2(t-tAE)) [18.3.9] with B = k2 - k1/ds [18.3.5] where: A area of the wall a coating thickness B fraction of binder cW VOC-concentration in the wall cL VOC-concentration in the ambient air D density ds thickness of the layer of the paint application (=a/D) k1 constant of evaporation k2 ventilation rate of the indoor air mL mass of VOC in the indoor air mL,AE mL at the end of the application mw mass of VOC in the wall mW,AE mW at the end of the application RM VOC-content in the dispersion t time V volume of the indoor air vST spreading velocity
In VOCs, especially these emitted from coating materials, the evaporation temperature is specified by European regulations: b.p. max. 250oC (according to DIN ISO 11890-1,-2 or 96/13/EC)14 or b.p. max. 260oC (according to VDI Guidelines 4300-6).15 In
1242
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
some US governmental legislation VOCs are defined solely as those compounds that are photochemically active in the atmosphere (ASTM D 3960).1 Regulations and assessment of VOCs are under discussion in Germany, e.g., discussion of various threshold limit values of TVOC (total volatile organic compounds): < 200 µg/m3 according studies of Molhave16 and 300 µg/m3 according to BGA.17 The percentage of VOC emissions caused by solvents has increased recently mainly because of the reduction of VOC coming from vehicular traffic, e.g., in Germany the estimated anthropogenic VOC emissions (without methane) caused by solvents in 1994 was about 51% compared to 37% in 1988.12 The solvent VOC emissions are mainly connected with coating materials which are responsible for more than 50% of solvent-based VOC emissions in Germany in 1995 (about 38% caused by varnishes, 4 % by building materials, 19% by other processes such as metal cutting).12 The main source of emissions of solvents during varnish processing are from equipment without licence requirements (about 60%) and from trade and paint work (about 15%); equipment which has licence requirements (manufacturing plants) contributes only 10% of VOC emissions.12 Therefore, solvents released during the application processes of paints and varnishes must be reduced. New EC regulations concerning reduction of solvents in special industrial plants or working processes (e.g., varnish coating of vehicles) have already been formulated.8 Similar regulations concerning other working fields (e.g., the use of solvent-based paints in home workshops) have been prepared throughout Europe and are already in force in some countries. 18.3.3.2 Immission In Europe, various regulations and schemes for the assessment of VOCs are in use according to special conditions, e.g.: • Indoor air: Schemes and values proposed by central environmental institutions commissions of UBA (Sachverständigenrat für Umweltfragen),18,19 20 (Umweltbundesamt). • Place of work: Special values (e.g., TLV-, MAK-, TRK-values) proposed by special institutions, e.g., commissions of ACIGH (American Conference of Governmental Industrial Hygienists)21 or DFG (Deutsche Forschungsgemeinschaft).22 TLV (Threshold Limit Values) refer to airborne concentrations of substances. They represent conditions to which workers may be repeatedly exposed during an 8-hour workday in a 40-hour week. MAK-values (Maximale Arbeitsplatzkonzentration) describe maximum concentrations of individual substances allowed in the work place (8 hours per day or 40 hours per week with some exceptions allowed). MAK-values are obligatory limits in Germany but in cases where MAK values cannot be evaluated (e.g., carcinogenic compounds) TRK values (Technische Richtkonzentrationen) are used. These recommendations are based on current technical knowledge.22 • Atmosphere, especially the significance of photooxidation. Various measures and regulations (e.g., reduction of VOCs)8 intend to limit substances which play a part in photooxidizing processes (e.g., ozone formation). When defining threshold limit values, it is important that prevailing conditions and methods are specified, e.g., ventilation rates, methods of sampling, determination and calculation.1,17,23
18.3 Indoor air pollution by solvents
1243
Immission processes of solvents from paints and varnishes are determined by surrounding conditions: e.g., parameters of the ambient air, indoor air parameters, e.g., ventilation (ventilation rates), air distribution, air movements, composition of ambient air, concentrations of air compounds, humidity, temperature, volume of the indoor air, extreme conditions (e.g., worst case). 18.3.4 EFFECTS ON HEALTH OF SOLVENTS FROM PAINTS AND VARNISHES Only the effects resulting directly from the exposure to paints and varnish solvent are discussed in this chapter. For the effects of other solvents, see Chapter 20.1. 18.3.4.1 Exposure Usually the effects of solvents in paints and varnishes on health are dependent on concentration and exposure time. Adverse health effects may follow exposure to paints, varnishes and their solvents at the workplace. The conditions at paint and lacquer manufacturing sites (e.g., manufacturing methods, use of exhaust hoods, etc.) are responsible for the levels of evaporated solvents measured in the air.24 Adverse health effects depend on how the paint is applied, paint properties and working conditions (e.g., increased risk with spray painting). There is sufficient evidence to substantiate the fact that solvents to which painters have been exposed, are responsible for incidences of cancer. However, within paint manufacturing plants, this evidence is inadequate.25 Nevertheless, person working at home, occupants of painted rooms and children of parents which have been exposed are also at risk. Those involved with the abuse of solvents as a psychoactive substance (e.g., aromatic hydrocarbons in spray paints, mainly semi-volatile or nonvolatile components) are at a much higher risk. The main path of entry of solvents from paints and varnishes to the body is by inhalation. Volatile paint compounds present a particularly high risk as do some forms of paint application (e.g., spray painting with the risk of inhalation of even less volatile and nonvolatile paint components). Other pathways should, however, also be considered as dermal contact. In addition to solvents, other compounds from paints and varnishes can cause various diseases, often similar to the effects caused by solvents (e.g., asbestos as paint filler or in the construction and shipyard industry, silica, dusts, thermal decomposition products, contaminations of solvents, chromium, iron and lead compounds in paint pigments). It is often difficult to associate a particular components of paints and varnishes with adverse health effects. In most cases, the paint and varnish products were found to be a relevant cause of illness, but their individual compounds were not. 18.3.4.2 Health effects 18.3.4.2.1 Toxic responses of skin and mucose membranes The following symptoms involving the skin and mucose membranes may occur as a result of using paints and varnishes: • Irritations of skin and mucose membranes • Allergic diseases of skin and mucose membranes • Removal of grease from skin (removal of sebaceous matter, with subsequent adverse skin conditions such as infection) • Changes in the lens of the eye and corneal changes • Absorption of solvents through the skin (e.g, benzene, toluene, xylene, methyl alcohol, methyl ethyl ketone, glycol ethers)
1244
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
Workers in the paint manufacturing industry and painters have experienced occupational diseases, especially dermatosis affecting the hands and arms.26-28 Whereas several paint components have been shown to cause non-allergic and allergic contact eczema, organic solvents were shown to provoke mainly non-allergic contact eczema and some solvents cause only irritation (e.g., some ketones and esters). In another study by Mancuso et al.28 it was shown that the occupational contact dermatitis of workers in shoe factories was probably a result of solvents in varnishes and adhesives. This study was based on interviews, medical examinations and patch test series. Eye diseases stemmed from both non-allergic and allergic reactions, and in some cases corneal and lens changes were noted.29 In a further study,30 with water based paints there was a significant reduction in eye and skin disease and worker discomfort on the job. 18.3.4.2.2 Neurological disorders Indoor air immissions of organic solvents from paints and varnishes can cause neurological disorders: • Neuro-psychological and neuro-behavioral symptoms (e.g., subjective symptoms, multiple chemical sensitivity - MCS) • Neurophysiological symptoms • Neurological diseases (e.g., polyneuropathy) • Neuropsychiatric diseases Other compounds in paints and varnishes apart from solvents can cause neurological disorders (e.g., lead). However, it is difficult to prove that solvents specifically cause neurological changes. A study of production plants producing dyes and varnishes,31 showed that mixtures of organic solvents are responsible for several neurological and neurophysiological symptoms: headache, dizziness, increased emotional excitability, memory and concentration disorders, mood instability, fatigue. Neurological examinations, however, showed no significant changes in the central and peripheral nervous system but EEG and VEP anomalies were seen. In contrast to the studies mentioned above, a study of shipyard spray painters32 exposed to xylene and mixed organic solvents described neurophysiological changes, e.g., decreased nerve function and, in addition, neuropsychological symptoms, e.g., mood changes and fatigue. Similar results including reduced nerve conduction were shown by workers exposed to styrene. Other studies found several dose-response relationships between solvent mixtures and neuro-behavioral effects among paint manufacturing employees and painters.33,34 Significant relationships concerning the total amounts of hydrocarbons, lifetime exposure and lifetime-weighted average were described. In earlier reviews and cross-sectional studies, various symptoms and neuro-behavioral effects were described for workers in the paint manufacturing industry, house painters, car and industry painters, and shipyard painters.35 Subjective symptoms (fatigue, loss of concentration, emotional instability, short-term memory disorders, headache) or effects on psychomotoric performance are examples of these symptoms. However, similar former studies did not find symptoms in house painters using mainly water-based paints.36 Different results were found concerning neuro-physiological changes and neurological diseases. Electroencephalographic changes and a slight decrease in cerebral blood flow of paint industry workers was noted by Oerbaek et al.37 and there were occasional cases of
18.3 Indoor air pollution by solvents
1245
clinical polyneuropathy in spray painters exposed mainly to methyl n-butyl ketone as described by Mallov.38 Furthermore, slight neurological impairments of car and industry painters was noted by Elofsson et al.29 and Maizlish et al.39 and in house painters by Askergren et al.36 Nevertheless, no similar effects were described in other studies.42,43 In particular, no effects on the peripheral nervous system were reported in workers using water-based paints.36 Inconsistent results were reported for severe diseases of the central nervous system (neuropsychiatric diseases, encephalopathy) for painters and other persons exposed to solvents.42-44,47 Environmental exposure to organic solvents is supposed to be one cause of multiple chemical sensitivity which characterizes neuropsychological disorders. Organic solvents from paints seem to play a role.45,46 The relevant exposure characteristics (dose, time, possible dose-response relationship) are also discussed.45 18.3.4.2.3 Carcinogenic effects Various solvents which occur in paints and varnishes exhibit a carcinogenic potential. Benzene and all isomers of dinitrobenzenes are particular examples (see also Chapter 20.1). This was confirmed by several extensive studies of occupational exposures in paint manufacturing plants and in painters.25 These data show the risk of contracting cancer to be about 20% above the national average. In many studies, increased risks were described mainly for lung cancer (about 40% above the national average), leukemia, bladder cancer, liver cancer, and childhood cancers where there has been parental exposure.25 In a meta-analysis using standardized mortality ratios the relationship between painting exposure48 and cancer mortality showed the highest risks for leukemia and liver cancer. Elevated risks were also predicted for lung cancer, oesophageal cancer, stomach cancer and bladder cancer. The development of leukemia, especially in the case of benzene mixed with other organic solvents, being most prominent and the development of lung cancer (main risks being lead chromate and asbestos) to a lesser degree. It should be noted, however, that more critical risk factors such as smoking and alcohol can obscure such correlations. Although high risks for cancer resulting from occupational exposures were mentioned, no significant information concerning the occupation of painted rooms could be noted. This area has been inadequately studied. As mentioned above, lung cancer is a major concern. Painters, as opposed to those involved in paint manufacture were shown to be at greatest risk for contracting lung cancer.25,48 In fact, there was no evidence of increased risk in persons involved in the manufacture of paints.25,49 Other cancers of the respiratory tract are documented also, e.g., cancer of the nasal cavity,50 pleural mesothelioma with high incidences in painters and paper-hangers51 and cancers of the larynx.52 The evidence of several types of leukemia is convincing with regard to occupational exposures to paints and solvents.25,48 Increased risks for contracting other haematopoietic neoplasmas25 such as Hodgkin´s disease,53 non-Hodgkin´s lymphoma,54 multiple myeloma,55 reticulum-cell sarcoma and lymphosarcoma56 were also reported. Probably all neoplasmas were caused mainly by organic solvents. Some former studies, however, included solvents which are rarely used now, especially benzene.25
1246
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
Davico et al.57 and Crane58 showed clonal aberrations in chromosome 8 of workers exposed to paints which is correlated with higher risks of acute myelocytic leukemia. Occupational exposure to dyestuff manufacture, paints, solvents and inks are an important risk factor for cancer development of the urinary tract (especially aromatic amines in bladder cancer).25,59 A major risk factor responsible for about 50% of bladder cancer cases in western countries is smoking. A second risk factor is occupational exposure (including exposure to paints and solvents).59 Some studies showed greater risks according to certain exposures: e.g., spray painting,60,61 lacquering and painting of furniture and cars or sign-post painting.62 An increased risk of cancer of the prostate was found for workers in paint manufacturing plants, but no increase in risks could be evaluated for exposure to paints.63 In one study an increased risk for testicular cancer was shown.25 Inconsistent results were noted for cancers of the gastrointestinal tract.25 Some studies showed higher rates of stomach and intestinal cancers, but other risk factors apart from paints cannot be excluded. Higher risks for liver cancer was found by Chen and Seaton.48 Norell et al.64 noted an excess of pancreatic cancers, especially as a result of exposure to paint thinners. No significant risks were described concerning cancer of the biliary tract.65 A distinct relation between parental occupational exposure and childhood cancer was shown for solvents and paints. High parental exposure resulted in higher incidences of childhood cancers.66 In the same study, however, generally more cancers were found as a result of parental use of alcohol and tobacco smoke. Childhood leukemia and nervous system cancers, in particular, are the types suspected to be caused by parental exposure to paints and solvents.67 Kishi et al.68 described an elevated risk for acute lymphatic leukemia in children of mothers with prenatal exposures to benzene and to paints. In former studies with small numbers of children these tendencies could also be shown, mainly in male painters whose children showed a higher incidence of childhood leukemia and brain tumors.25 18.3.4.2.4 Respiratory effects The following respiratory symptoms are provoked by solvents:
• • • •
Irritations Allergic reactions Changes in lung function parameters (mainly obstructive ventilation) Pathohistological changes Irritations of the air passages were described in people exposed to fumes in paint factories.27,69 VOCs in paints can provoke respiratory symptoms (wheezing, breathlessness) in asthmatics. Conventional water-based paints with only small amounts of VOCs have also been shown to cause such symptoms, but there were no effects using VOC-free paints.70 No differences were found in the same study looking at lung function and airway responsiveness. Toxic pyrolysis products in paints and polymer films probably evoke asthma-like symptoms similar to PVC pyrolysis.71 A decrease in forced vital capacity, expiratory volume in one second and of peak exspiratory flow was observed after exposure to water-based paints.72 Bronchial obstruction in painters was confirmed by White and Baker,69 but other studies could not detect changes in lung function parameters in house painters who were exposed to solvent-based and water-based paints.36 Beving et al.73 did not find obstructive effects in car painters. An increase
18.3 Indoor air pollution by solvents
1247
in mortality from chronic obstructive bronchitis of painters was recorded by Engholm and Englund,74 but not confirmed in other studies.25 Histological changes in the nasal mucosa of spray painters were also noted in a study of Hellquist et al.75 18.3.4.2.5 Toxic responses of the blood A case-control study of Guiguet et al.76 observed an increased risk for aplastic anaemia after exposure to paints (effective compounds in the paints are unknown), but no increased risk after exposure to solvents alone. Reduction of blood hemoglobin levels was shown after exposure to gasoline, car spray paint (xylene) and various solvents (house painters).25,77,78 An increased Hb level was noted, however, in a study of car and industrial spray painters.29 Alterations of several blood components (production of auto-antibodies) and of vascular endothelial cells were described for workers exposed to hydrocarbon-based paints or to mixed solvents.79 Even a group experiencing low exposure (house painters exposed to alkyd paints) showed hematological changes. Higher platelet counts and higher resistivity calculated from the the impedance of the whole blood were observed.73 Some studies showed decreases of thrombocyte counts in car painters and paint industry workers, other studies, however, recorded no changes.25 Studies concerning effects of paints on the white cell and thrombocyte counts were inconsistent. A decrease of white cells was described in several studies11 and a lymphocytosis was noted by Angerer and Wulf.77 Elofsson et al.,29 however, recorded no changes in white cell counts. Additionally, myelotoxic effects of solvents were shown, especially for benzene. 18.3.4.2.6 Toxic responses of the reproductive system In interviews with pregnant women, the effects of paints, varnishes and solvents were evaluated. No effects were found concerning congenital heart disease of infants with Down Syndrome (trisomie 21).80 The maternal risk was associated with smoking. Maternal toluene sniffing used as an organic solvent in acrylic paints, varnishes and other sources is associated with premature births and, in one case, renal tubular acidosis.81 Other studies of female painters showed tendencies towards an increase of spontaneous abortions82 or infant mortality.83 Other studies84 did not confirm these tendencies. McDowall reported an increase in malformations in children (polydactyly, syndactyly, spina bifida and anencephalus) whose parents were painters, assemblers or had related occupations.83 A study by Olson85 showed congenital malformations of the central nervous system for paternal exposure to paints. 18.3.4.2.7 Toxic responses on other organ systems Contradictory effects of solvents in paints on the kidneys have been described.25 Lauwerys et al., for example, have described some solvents used in paints as nephrotoxic (e.g., toluene).88 Minor effects such as slight hematuria and albuminuria or small effects on the glomeruli or even no effects were detected in other studies.25 In some studies of spray painters in the automobile and airplane industries and of house painters mortality from liver cirrhosis was increased.25,87 Another study investigating house painters, however, did not show an increase.42 There seem to be no increased risk for cardiovascular diseases according to some studies of paint industry workers.25,89 Nevertheless, effects of solvents on muscles and vessels could be shown (e.g., increased serum creatinine levels).25
1248
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
18.3.5 METHODS FOR THE EXAMINATION OF SOLVENTS IN PAINTS AND VARNISHES 18.3.5.1 Environmental monitoring There are several test procedures generally used in Europe. Procedures are still being developed to determine the presence of solvents in paints and varnishes. 18.3.5.1.1 Solvents in products Officially approved test procedures are: • Gravimetry: Determination of volatile organic compounds (DIN ISO 11 890-17 VOCs > 15 %). • Gas chromatography: Determination of volatile organic compounds (DIN ISO 11 890-27 - VOCs 0,1 % -15 %). 18.3.5.1.2 Emission of solvents Test chamber7,92,93 (volume 1 m3, defined conditions: temperature, humidity, air changing rates, air velocity). This is usually carried out 3 days after application of paints or varnishes by placing the products into a test chamber. The test series starts with sampling emissions of the products on defined absorption materials (Tenax, charcoal filters, activated charcoal), reconditioning and finally identification and classification of VOCs with gas chromatography. The test methods are repeated after 14 and 28 days. These test criteria permit determination of the behavior of emissions from finished products under defined conditions. Emission test cells.7,92,93 The product under investigation is hermetically sealed into the emission test cells. The emitted compounds are sampled by absorption materials and analyzed. This method can be used under laboratory conditions or in situ. 18.3.5.2 Biological monitoring of solvents in human body fluids 18.3.5.2.1 Solvents and metabolites in human body fluids and tissues The biological monitoring of solvents emitted from paints or varnishes on humans is not well developed. In two studies,94,95 solvents from paints and varnishes were determined in blood, urine and internal breath. Blood and urine analysis is less sensitive than internal breath measurements. This was carried out in a study on exposure to paints in aircraft maintenance.94 Kramer et al.95 found xylenes during paint production and paint-spraying in ambient air and in the blood and urine of workers. Threshold limit values (TLV) and biological exposure indices (BEI) were not exceeded.22,23 18.3.5.2.2 Biomarkers Even low air concentrations of solvents emitted from paints have an influence on the human organism, e.g. the induction of hepatic enzymes. This effect was shown in workers exposed to butyl glycol from paints in an electrophoresis painting plant where the exposure value was shown to be less than 0.3 times of the average limit. In these cases D-glucaric acid in urine, which reflects the D-glucarid acid enzyme pathway, was increased.96 In some studies the genetic effects on paint industry workers could not be detected when chromosomal aberrations and sister chromatid exchanges were studied.86,87 In other studies94,95 it was shown that the frequency of micronuclei and sister chromatid exchanges increased.
18.3 Indoor air pollution by solvents
1249
REFERENCES 1 2 3
4 5 6 7 8 9
10
11 12
13 14 15 16 17 18 19 20 21 22 23
24 25
26 27 28 29
30
CEN (European Committee for Standardization), Paints and varnishes. Terms and definitions for coating materials. Part 1: General terms. EN 971, 1996. Ullmann`s Encyclopedia of Industrial Chemistry, 1998. DIN 55945 (Deutsches Institut für Normung), Coating materials (Varnishes, Painting Materials and other Materials), Terms. Beschichtungsstoffe (Lacke, Anstrichstoffe und ähnliche Stoffe); Begriffe, Beuth Verlag, Berlin, 1988. VdL (Verband der Lackindustrie/ Association of the Varnish Industry in Europe)- VdL-Guidelines Building Coating Materials. VdL-Richtlinie Bautenanstrichstoffe, RL 01, 1997. Römpp Lexikon, Varnishes and Paints. Lacke und Druckfarben, Thieme, 1998. H. Zeh, M. Kroll, and K. Kohlhammer, VDI Berichte, 1122, pp. 455-475 (1994). DIN (Deutsches Institut für Normung), ISO/ DIN 11890. Paints and varnishes - Determination of volatile organic compound (VOC) content, Beuth Verlag, Berlin, 1998. European Commission, ECA-IAQ, Total Volatile Organic Compounds (TVOC) in Indoor Air Quality Investigations, EUR 17675 EN, Report No. 19, 1997. 1999/13/EC, Council Directive of 11 March 1999 on the limitations of emission of volatile organic compounds due to the use of organic solvents in certain activities and installations. ABl. EG Nr. L 085, 29.3.99, 1999. DIN (Deutsches Institut für Normung)(Normenausschuss Bauwesen NABau), Koordinierungsausschuss KOA 03 “Hygiene, Gesundheit und Umweltschutz”: Guidelines for Health Assessment of Building Products. Leitfaden zur Beurteilung von Bauprodukten unter Gesundheitsaspekten, KRdL-CEN/264/7 N 1999-07, 1999. GISBAU, WINGIS, Bau-Berufsgenossenschaften (Professional Associations of the Building Industry in Germany), 1999. Deutsches Lackinstitut Frankfurt a.M. (German Institute of Varnishes Frankfurt a.M.), Solvent reduction in the trade of painters and lacquerers. Documents concenring varnishes and paints. Magazine No. 7. Lösemittel-Reduzierung im Maler- und Lackiererhandwerk. Dokumente zu lacken und Farben. Heft Nr. 7, DeutschesLackinstitut, Frankfurt a.M., 1999. T. Salthammer, Chemie in unserer Zeit, 28, 280 (1994). DIN (Deutsches Institut für Normung), DIN ISO 11890-1,-2, Beuth Verlag, Berlin, 1998. VDI (Verein Deutscher Ingenieure), VDI 4300, VDI, Düsseldorf, 1999. L. Molhave, Indoor Air, 1, 357 (1995). Bundesgesundheitsamt (Public Health Department of the German Government), Bundesgesundheitsblatt, 36 (3), 117-118 (1993). SRU (Environmental Council of the German Government)(Sachverständigenrat für Umweltfragen), Air pollutions in rooms, Special Report, Metzler-Poeschel, Stuttgart, 1987. SRU (Environmental Council of the German Government)(Sachverständigenrat für Umweltfragen), Environmental Report 1996, Metzler-Poeschel, Stuttgart, 1996. Anonymous, Bundesgesundheitsblatt, 11, 422 (1996). ACGIH (American Conference of Governmental Industrial Hygienists), TLVs and BEIs. Threshold Limit Values for Chemical Substances and Physical Agents. ACGIH, 1998. DFG (Deutsche Forschungsgemeinschaft), MAK- und BAT-Werte-Liste 1999, Wiley-VCH, 1999. M. Fischer, and E. Böhm, Evaluation and assessment of emission of noxes caused by furniture varnishes. Erkennung und Bewertung von Schadstoffemissionen aus Möbellacken. Schadstoffe und Umwelt, Erich Schmidt Verlag, Berlin, 1994. W. Wesolowski, and J.P. Gromiec, Int. J. Occup. Med. Environ. Health, 10 (1), 79 (1997). IARC (International Agency for Research on Cancer), IARC Monographs on the evaluation of carcinogenic risks to humans. Some organic solvents, resin monomers, resin monomers and related compounds, pigments and occupational exposures in paint manufacture and painting. WHO, 47, IARC, Lyon, 1989. U. Ulfvarson, Int. Symp. on the Control of Air Pollution in the Working Environment, Stockholm, September 6-8, Swedish Work Environment Fund/ International Labour Office, Part II, 1977, 63-75. R.V. Winchester and V.M. Madjar, Ann. Occup. Med., 4, 221 (1986). G. Mancuso, R.M. Reggiani, and R.M. Berdandini, Contact Dermatitis, 34 (1), 17 (1996). S.A. Elofsson, F. Gamberale, T. Hindmarsh, A. Iregren, A. Isaksson, I. Johnsson, B. Knave, E. Lydahl, P. Mindus, H.E. Persson, B. Philipson, M. Steby, G. Struwe, E. Söderman, A. Wennberg, and L. Widen, Scand. J. Work Environ. Health, 6, 239 (1980). G. Wieslander, D. Norback, and C. Edling, Occup. Environ. Med., 51(3), 181 (1994).
1250
31 32 33 34 35 36 37 38 39 40 41
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland
J.A. Indulski, H. Sinczuk-Walczak, M. Szymczak, and W. Wesolowski, Int. J. Occup. Med. Environ. Health, 9(3), 235 (1996). M.W. Ruijten, J.T. Hooisma, J.T. Brons, C.E. Habets, H.H. Emmen, and H. Muijser, Neurotoxicology, 15 (3), 613 (1994). F. Fung, and R.F. Clark, J. Toxicol. Clin. Toxicol., 37(1), 91 (1999). A. Seeber, B. Sietmann, and M. Zupanic, Int. J. Occup. Med. Environ. Health, 9 (3), 235 (1996). NIOSH (National Institute for Occupational Safety and Health), Organic solvent neurotoxicity, Current Intelligence Bulletin No. 48, Cincinnati, 1987. A. Askergren, H. Beving, M. Hagman, J. Kristensson, K. Linroth, O. Vesterberg, and A. Wennberg, Arb. Hälsa, 4, 1 (1988). P. Oerbak, J. Risberg, I. Rosen, B. Haeger-Aronsen, S. Hagstadius, U. Hjortsberg, G. Regnell, S. Rehnström, K. Svensson, and H. Welinder, Scand. J. Work Environ. Health, 11 (Suppl. 2), 1 (1985). J.S. Mallov, MBK neuropathy among spray painters, J. Am. Med. Ass., 235, 1455 (1976). N.A. Maizlish, G.D. Langolf, L.W. Whitehead, L.J. Fine, J.W. Albers, J. Goldberg, and P. Smith, Br. J. Ind. Med., 42, 579 (1985). N. Cherry, H. Hutchins, T. Pace, H.A. Waldron, Br. J. Ind. Med., 42, 291 (1985). D. Triebig, D. Claus, I. Csuzda, K. Druschky, P. Holler, W. Kinzel, S. Lehrl, P. Reichwein, W. Weidenhammer, W.U. Weitbrechst, D. Weltle, K.H. Schaller, and H. Valentin, Int. Arch. Occup. Environ. Health, 60, 223 (1988). S. Mikkelsen, Scand. J. Soc. Med. Suppl., 16, 34 (1980). H. Rasmussen, J. Olsen, and L. Lawitsen, J. Occup. Med., 27, 561 (1985). C. Van Vliet, G. Swaen, J. Slangern, T. Border, F. Stirman, Int Arch. Occup. Environ. Health, 59, 493 (1987). E. Kiesswetter, Zbl. Hyg. Umweltmed., 202, 191 (1998). N. Fiedler and H. Kiepen, Environ. Health Perspect., 105 Suppl. 2, 409 (1997). S. Mikkelsen, M. Jorgensen, E. Browne, and C. Gyldensted, Acta Neurol. Scand. Suppl., 118, 1(1988). R.L. Chen, and A. Seaton, Cancer detection and prevention, 22 (6), 533 (1998). K.L. Milne, D.P. Sandler, R.B. Everson, S.M. Brown, Am. J. Ind. Med., 4, 565 (1983). S. Hernberg, P. Westerholm, K. Schultz-Larsen, R. Degerth, E. Kuosma, A. Englund, U. Engzell, H. Sand Hansen, and P. Mutanen, Scand. J. Work Environ. Health, 9, 315 (1983). H.S.R. Malker, J.K. McLaughlin, B.K. Malker, B.J. Stone, J.A. Weiner, J.L.E. Erickson, and W.J. Blot, J. Natl. Cancer Inst., 74, 61 (1985). L.M. Brown, T.J. Mason, L.W. Pickle, P.A. Stewart, BP. Auffler, K. Burau, R.G. Ziegler, and F. J. Fraumeni, Cancer Res., 48, 1960 (1988). H. Olssen, and L. Brandt, Scand. J. Work Environ. Health, 14, 246 (1980). H. Olssen, L. Brandt, Scand. J. Work Environ. Health, 6, 302 (1988). J. Cuzik, and B. De Stavola, Br. J. Cancer, 57, 516 (1988). N.J. Vianna, and A. Polan, Lancet, 1394 (1979). L. Davico, C. Sacerdote, G. Ciccone, L. Pegoraro, S. Kerim, G. Ponzio, and P. Vineis, Cancer Epidemiol. Biomarkers Prev., 7 (12), 1123 (1998). M.M. Crane, Cancer Epidemiol. Biomark. Prev., 5, 639 (1996). R. Piratsu, I. Iavarone, and P. Comba, Ann. Ist. Super Sanita, 32 (1), 3 (1996). J. Claude, E. Kunze, R. Frentzel-Beyme, K. Paczkowski, J. Schneider, H. Schubert, Am. J. Epidemiol., 124, 578 (1986). J. Claude, R. Frentzel-Beyme, E. Kunze, Int. J. Cancer, 41, 371-379 (1988). O.M. Jensen, J. Wahrendorf, J.B. Knudsen, B.L. Sorensen, Scand. J. Work Environ. Health, 13, 129 (1987). R.C. Brownson, J.C. Chang, J.R. Davis, and J.R. Bagby, J. Occup. Med., 30, 523 (1988). S. Norell, A. Ahlbom, R. Olin, R. Ewald, G. Jacobsen, I. Lindberg,-Navier, and K. Wiechel, Br. J. Ind. Med., 43, 775 (1986). H.S.R. Malker, J.K. McLaughlin, B.K. Malker, B.J. Stone, J.A. Weiner, J.L.E. Erickson, and W.J. Blot, Br. J. Ind. Med., 43, 257 (1986). M.L. McBride, Can. J. Public Health, 89 (1), 53 (1998). J.S. Colt, and A. Blair, Environ. Health Perspect., 106 (3), 909 (1998). R. Kishi, Y. Katakura, J. Yuasa, and H. Miyake, Sangyo Igaku, 35 (6), 5115 (1993). M.C. White and E.L. Baker, Br. J. Ind. Med., 45, 523 (1988). J.R. Beach, J. Raven, C. Ingram, M. Bailey, D. Johns, E.H. Walters, and M. Abramson, Eur. Respir. J., 10 (3), 563 (1997). J.E. Peterson, Occup. Med., 8(3), 533 (1993).
18.4 Solvent uses with exposure risks
72 73 74
75 76 77 78 79 80 81 82 83
84 85 86 87 88 89 90 91 92 93 94 95 96
1251
U. Ulfvarson, R. Alexandersson, M. Dahlqvist, U. Ekholm, B. Bergstrom, and J. Scullman, Scan. J. Work. Environ. Health, 18 (6), 376 (1992). H. Beving, B. Tedner, and L.E. Eriksson, Int. Arch. Occup. Environ. Health, 63 (6), 383 (1992). G. Engholm and A. Englund in Advances in Modern Environmental Toxicology, Vol. II, Occupational Health Hazards of Solvents, A. Englund, K. Ringen, and M.A. Mehlman, eds., Princeton Scientific Publishers, Princeton, N.J., pp. 173-185 (1982). H. Hellquist, K. Irander, C. Edling, and L.M. Ödkvist, Acta Otolaryngol., 96, 495 (1983). M. Guiguet, E. Baumelou, and J.Y. Mary, Int. J. Epidemiol., 24 (5), 993 (1995). J. Angerer, and H. Wulf, Int. Arch. Occup. Environ. Health, 56, 307 (1985). M. Hane, O. Axelson, J. Blume, C. Hogstedt, L. Sundell, B. Ydreborg, Scand. J. Work Environ. Health, 3, 91 (1977). A. Stevenson, M. Yaqoob, H. Mason, P. Pai, and G.M. Bell, QJM, 88 (1), 23 (1995). D.E. Fixler, N. Threlkeld, Teratology, 58 (1), 6 (1998). J. Erramouspe, R. Galvez, and D.R. Fischler, J. Psychoactive Drugs, 28 (2), 201 (1996). L.Z. Heidam, J. Epidemiol. Commun. Health, 38, 149 (1984). M.E. McDowall, Occupational Reproductive Epidemiology: The Use of Routinely Collected Statistics in England and Wales, 1980-82, Studies on Medical and Populations Subjects No. 50, Her Majesty´s Stationary Office, Office of Population Censuses and Surveys, London, 1980. W.E. Daniell, and T.L. Vaughan, Br. J. Ind. Med., 45, 193 (1988). J.H. Olsen, Dan. Med. Bull., 30, 24 (1983). U. Haglund, I. Lundberg, and L. Zech, Scand. J. Work Environ. Health, 6, 291 (1980). K.T. Kelsey, J.K. Wiencke, F.F. Little, E.L. Baker, J.B. Little, Environ. Mol. Mutagenesis, 11, 389 (1988). R. Lauwerys, A. Bernard, C. Viau, and J.P. Buchet, Scand. J. Work Environ. Health, 11 (1), 83 (1985). G. Engholm, A. Englund, H. Löwing, Scand. J. Work Environ. Health,. 13,. 181 (1987). R.W. Morgan, K.W. Claxton, S.D. Kaplan, J.M. Parsons, and O. Wong, J. Occup. Med., 27, 377 (1985). I. Lundberg, Scand. J. Work Environ. Health., 12, 108 (1986). European Collaborative Actions (ECA), Indoor Air Quality & Ist Impact On Man. Evaluation of VOC Emissions from Building Products - Solid Flooring Materials, ECA Report No. 18, 1997. Anonymous, Building products - Determination of volatile organic compounds Part 1: Emission test chamber method. prENV 13419-1, draft. G.K. Lemasters, J.E.Lockey, D.M. Olsen, S.G. Selevan, M.W. Tabor, and G.K. Livingston, Drug and Chemical Toxicology, 22 (1), 181 (1999). A. Kramer, M. Linnert, R. Wrbitzky, and J. Angerer, Int. Arch. Occup. Environ. Health, 72(1), 52 (1999). J.P. Collinot, J.C. Collinot, F. Deschamps, D. Decolin, G. Siest, and M.M. Galteau, J. Toxicol. Environ. Health, 48 (4), 349 (1996).
18.4 SOLVENT USES WITH EXPOSURE RISKS Pentti Kalliokoski University of Kuopio, Kuopio, Finland
Kai Savolainen Finnish Institute of Occupational Health, Helsinki, Finland
18.4.1 INTRODUCTION This chapter deals merely with exposure to common organic solvents which are used in large quantities to dissolve fats, resins, and other materials. Very dangerous chemicals, such as benzene, which are no more used as solvents due to their toxic properties, will not be discussed even though they may have had even extensive uses as solvents earlier and even though those may be still important chemicals as petroleum components or as intermediates for other chemicals. Also, solvents with very specialized uses, such as carbon disulfide the use of which is practically limited to viscose rayon industry and laboratories, are only
1252
Pentti Kalliokoski, Kai Savolainen
shortly considered. Exposure data presented originate mainly from literature; however, some unpublished data obtained from the Finnish Institute of Occupational Health (FIOH) has been added to indicate the order of magnitude of current exposure levels in industrialized countries (data cover the years 1994-1996). Exposure to organic solvents should be avoided mainly due to the risk of neurotoxic effects. Acute effects are narcotic resembling those caused by the use of alcohol. Those appear, for example, as a decrease in reaction time and impairment in psychological performance. Chronic neurotoxic effects are often called as the organic psychosyndrome (OPS) including memory disturbances, excessive tiredness, personality changes, irritability and affect lability. Intellectual reduction may occur but it seems that development of real dementia would require simultaneously heavy consumption of alcohol.1 Increased risk of sleep apnoea has been observed among men exposed occupationally to organic solvents.2 Exposure to toluene, xylene, and styrene has been found to contribute to the development of noise-induced hearing loss. This interaction has been suggested to be due to neurotoxic injuries caused by solvents in the cochlea.3 Long-term exposure to carbon disulfide and n-hexane may result in peripheral neuropathy. It should be noted that n-heptane and n-octane have not been shown to cause the type of peripheral neuropathy (numbness, weakness, and pain in extremities) associated with n-hexane.4 Otherwise than the other alkanes, n-hexane is metabolized to a reactive, toxic compound (2,5-hexanedione).5 Organic solvents are generally skin irritants. Repeated or prolonged contact may cause erythema and dryness of the skin. Defatting may lead to cracked skin. Many solvents also readily absorb through the skin (see 18.4.2). Reactive solvents, such as styrene and vinyltoluene, may cause contact dermatitis.6 The annual incidence of occupational diseases caused by organic solvents was 20.6 cases/10,000 exposed workers in Finland in the 1980’s. More than half of the cases (64%) were dermatoses.7 Benzene is a well-known human carcinogen. An association with exposure to benzene and leukemia was detected already in 1920’s.8 There is some evidence that exposure to other organic solvents may constitute a carcinogenic risk. The International Agency for Research on Cancer (IARC) has concluded that there is sufficient evidence for the carcinogenicity of painters’ occupational exposure.9 Epidemiological studies conducted among dry cleaning and metal degreasing workers suggest an increased cancer risk due to exposure to chlorinated solvents.10 The carcinogenicity of aromatic solvents (styrene, toluene, and xylene) has been investigated in many large epidemiological studies. Although the results have been inconclusive, certain site-specific associations (e.g., for lymphohematopoietic tissues) have appeared in some of the studies.11 Only styrene has shown to have some genotoxic activity in animal studies.9 18.4.2 EXPOSURE ASSESSMENT Inhalation is usually clearly the most important route for occupational exposure to solvents. Organic solvents also enter the body through the skin.12 For some solvents, especially for alkoxyalcohols (glycol ethers) and their acetates, this is even the main route of absorption. Even the skin uptake of vapor can be significant for these compounds.13 Dermal absorption of vapors is, however, usually negligible and contact with liquid is required. Increased workload and heat enhance both inhalation and dermal absorption.14 Skin contamination may also lead to oral uptake due to eating and smoking but this is of minor importance for volatile solvents.
18.4 Solvent uses with exposure risks
1253
Most industrial countries have occupational exposure limits (OELs) for airborne concentrations to prevent excessive exposures. The threshold limit values (TLVs) published by the American Conference of Governmental Industrial Hygienists (ACGIH15) are unofficial but have had a great impact on OELs in the Western countries. These are updated annually based on new epidemiological and toxicological data. Organic solvents are often used as mixtures. Because the OELs for most solvents base on their neurotoxic properties, their effect is considered to be additive and the combined exposure levels (or hygienic effects) are calculated as the sum of the fraction of the OEL that each solvent represents. If the sum is larger than one, a noncompliance situation exists. Compounds which may be dangerous by uptake through the skin carry a skin notation. However, no quantitative dermal exposure limits are available yet. Methanol, turpentine, glycol ethers, and many chlorinated and aromatic hydrocarbons have the skin notation. Exposures can also be assessed by analyzing biological specimens, such as blood or urine. Metabolites of compounds are usually analyzed from urine samples. The amount of biological exposure limits or indices (BEIs) is, however, much smaller than that for airborne concentrations. However, biological monitoring may be useful although no BEI has been established to ascertain effectiveness of personal protection or to follow exposure trend. Biological monitoring reflects exposure via all routes. This is beneficial for individual exposure assessment. Biological monitoring also reveals possible accumulation of a compound. This is, however, of minor importance for most solvents because of their short half-lives. The largest disadvantage of biological monitoring is that it does not provide any information on the reasons of exposure. About twenty solvents have an ACGIH BEI (e.g., acetone, carbon disulfide, 2-ethoxyethanol, n-hexane, methanol, methyl ethyl ketone, perchloroethylene, styrene, toluene, trichloroethylene, and xylene).15 Airborne concentrations of solvents are most commonly determined by taking samples on adsorbents, such as activated carbon. Sampling can be done with a pump or passively based on diffusion. In industrial working places where airborne concentrations of solvents are on ppm-level, samples are generally extracted with a solvent (carbon disulfide is the most common one). In offices and other nonindustrial environments where concentrations are at ppb-level, samples are taken onto Tenax adsorption tubes which are desorbed thermally. In both cases, gas chromatography is the most common analytical method. Because the sampling time generally varies, it is important to calculate the time-weighted average (TWA) concentration. TWA concentration is obtained by using the sampling times as weights. Sampling and analysis has been reviewed e.g. by Soule.16 Direct-reading instruments, such as infrared or photoionic analyzers are good if only one solvent is present but interferences may be a problem when solvent mixtures are analyzed. Monitoring methods for dermal exposure have not been standardized as well as the methods to assess inhalation exposure. Adsorptive pads of activated carbon cloth can be used for monitoring of dermal exposure to organic solvents. Patches are attached on various parts of the body under the clothing and the amounts analyzed are then multiplied with the areas of each body region.17 Exposure assessment is complicated by the contribution of solvent vapor on the samples. Because the adsorptive surface of activated carbon is much larger than that of the skin, pads will adsorb much more solvents than the skin. However, only exposure to liquid phase is usually significant; therefore, the adsorbed vapor should be subtracted from the total mass analyzed.18 However, this makes the exposure assessment
1254
Pentti Kalliokoski, Kai Savolainen
quite complicated, and if biological monitoring methods are available those usually provide a more practical alternative. Urine and blood are the most commonly used biological samples. Exhaled air samples can also be used. Sampling is usually carried out at the end of the shift. Because the amount of water consumed affects the concentration of the solvent or its metabolite in urine, creatinine (a normal constituent of urine) correction is often applied (mass analyzed is presented per gram of creatinine). Alternatively, the correction can be made for the relative density of urine. A density of urine of 1.024 is usually applied for this purpose. In principle, collection of 24-hour urine would be an ideal approach for biological monitoring because then the actually excreted amount of the metabolite would be measured. Because this is very difficult to carry out creatinine or density correction is generally chosen as a more practical alternative. Unfortunately, not only several ways of correction for the density of urine are being used but also several different units have been adopted. For example, the ACGIH BEIs are usually given as mg/g creatinine but the values of the FIOH and many other European organizations are presented as µmol/l of urine. In addition, other units, such as µmol/mol of creatinine, are used. The mass units given in mg can be converted to mmol by dividing by the molecular weight of the compound (molecular weight of creatinine is 113). On the other hand, because the concentration of creatinine in urine varies and it is determined separately for each individual sample, values given with and without creatinine correction (or density correction) cannot be compared directly. That is the reason why the BEIs given by different institutions or agencies have not been converted to the same units in the text. As a rule of thumb, the concentration per gram of creatinine can be obtained by multiplying the concentration per liter of urine with a factor of 0.5 - 1.0. Airborne solvent concentrations usually vary much with time. Although repeated random personal sampling is theoretically the optimal method for inhalation exposure assessment, it is very time-consuming and does not necessarily reveal the reasons for exposure. Often, occupational hygiene surveys are conducted only to make certain that the concentrations of air impurities are in compliance with the OELs. Also, the European standard (EN 689/95) for occupational exposure assessment is primarily issued for this purpose. It would, however, be practical if the measurements conducted would also provide useful information for planning of remedial measures if the measurements reveal those to be necessary. For this purpose, it is important to recognize the difference between manual tasks and process industries. In manual tasks, emissions are released very close to the worker. Most problematic solvent exposures occur while performing manual tasks, such as painting, gluing, degreasing, and cleaning. Batch processes in paint and printing ink manufacture also contain many manual tasks. Manual lay-up methods are common in the reinforced plastics industry. The workers often perform different tasks with different exposure levels. All major tasks should be investigated under various conditions. Smoke tube tests provide useful information on the spreading of solvent vapor and the efficiency of local exhausts. Good enclosure for the emission source is important for successful exposure control. Detailed instructions for ventilation arrangements in various industries are available (references will be given later). Rotogravure printing is an example of a process industry with solvent exposures. The process is, in principle, closed but emissions take place from the openings for the paper web. The workers do not need to stay in the immediate vicinity of the emission sources but can
18.4 Solvent uses with exposure risks
1255
often spend most of their time in the clean control room. Thus, the workers’ exposure levels depend mainly on their moving pattern during the work shift. In addition, the emission rate depends on the production rate, i.e. on the consumption rate of the solvent (toluene). Stationary sampling can be applied for exposure assessment when it is combined with questionnaire on use of time in various areas. This can be done with a direct-reading instrument because the airborne concentration level remains quite stable.20 The possible skin exposure should also be taken into consideration. If skin contamination seems possible adequate protective clothing should be used. Glove selection is an important but difficult issue. Glove materials often tolerate organic solvents poorly. Glove manufacturers have useful information for selecting gloves for individual solvents. Especially difficult is, however, to find gloves protecting efficiently penetration of solvent mixtures. Nitrile (butadiene-acrylonitrile copolymer) gloves are often chosen in such a situation. There is, however, considerable intermanufacturer and even batch lot variability in penetration of solvents through nitrile gloves.19 The workers should also be instructed not to use thinners for hand and skin cleaning. Total quality management (TQM) is an effective way to ensure also a high quality of the working environment. The guidance for right and safe working practices should be subjoined to all working instructions. Employee participation is an essential feature of a well working TQM, and also greatly assists the achievement of the hygienic goals set. 18.4.3 PRODUCTION OF PAINTS AND PRINTING INKS Painting technology has changed over the years. The exposure levels were generally highest between the mid-1950’s and mid-1960’s when solvent-based paints were used extensively both as construction and industrial paints, and exposure control technology was still undeveloped.21 Rotogravure and silkscreen printing inks contain organic solvents. Until 1950’s benzene was used as the solvent in rotogravure inks and it remained as an impurity in toluene until 1960’s.22 Today, toluene used in rotogravure inks does not contain benzene. Alkoxyalcohols and their acetates are used as silkscreen ink solvents. The development of safer solvents has been started quite recently. First, alkoxyethanols were replaced by their acetates which have lower vapor pressures. This did not, however, improve the safety much because the skin is the main route of absorption for both alkoxyethanols and alkoxyethylacetates, and they are considered to be equally toxic. Their substitution by alkoxypropanols and their acetates is, however, a significant improvement because those do not metabolize to toxic alkoxyacetic acids.23 High exposures remained common in paint and printing ink industry still in 1970’s; for example, most solvent measurements conducted in these industries in Finland exceeded the present OELs.24 Water-based paints are today clearly most common in construction painting. Alkyd-based construction paints with white spirit (Stoddard solvent, mineral spirit or solvent naphtha) as a solvent are, however, still produced. Solvent-based paints have remained most common in industrial painting, even though the solventless powder paints are also produced in large quantities. The main products are nowadays manufactured in automated processes. On the other hand, special products are also usually made in batches and include several manual tasks. In addition to paints and inks, thinners are often canned. When the processes are provided with proper enclosures and local exhausts, the airborne solvent concentrations can be kept well below the OELs. An easy but important control measure is to keep all solvent containing pots covered. Xylene (TLV 100 ppm, the Finnish OEL 100 ppm with skin notation), toluene
1256
Pentti Kalliokoski, Kai Savolainen
(TLV and the Finnish OEL 50 ppm with skin notation), butanol (TLV 50 ppm as ceiling value and with skin notation, the Finnish OEL 50 ppm with skin notation), and white spirit (TLV 100 ppm, the Finnish OEL 770 mg/m3 or 135 ppm) are usually the main concerns in paint manufacturing plants.24,25 In the 1980’s, the combined solvent exposure levels were generally below the OEL but high exposure levels were detected in pot cleaning.26 The situation has further improved to some extent in Finland in 1990’s. The measurements conducted in a large Finnish paint factory in 1996-7 yielded the following mean combined personal solvent exposure levels: batch production 0.34 (range 0.14 -0.72), automated production lines 0.17, filling 0.41 (range 0.14-1.0), and pot cleaning 1.51 (range 0.28-3.06). Thus, pot cleaning remains as a problematic task. Exposures to xylene and 1-methoxy-2-propanol caused the largest contributions in the hygienic effects. Xylene was the solvent with the highest airborne concentration (mean 26 ppm; range 0.1-359 ppm) compared to the OELs in the Finnish paint manufacturing plants in general in the mid 1990’s. It has been reported that no profound changes in exposures have occurred in the Swedish paint industry since the mid-1980’s.21 In the Finnish printing ink plants, the combined solvent levels were still in the 1980’s often out of compliance when compared to the OELs.26 Significant airborne concentrations were observed for toluene, ethyl acetate (TLV 400 ppm, the Finnish OEL 300 ppm), aromatic solvent naphtha ( the Finnish OEL 240 mg/m3), and acetone (TLV and the Finnish OEL 500 ppm). The cleaning of vessels of barrels was again an especially problematic task. If cleaning is done manually, it is difficult to control the exposure sufficiently well with local ventilation but respiratory protection is also needed. 18.4.4 PAINTING Painters are probably the largest worker group exposed to solvents. This may also be the reason why much data on occupational health risks due to solvent exposure originate from painting work although the exposure levels have been generally lower than e.g. in paint manufacture. On the other hand, chronic neurotoxic effects and especially cancer require long exposure time (more than 10 years) and thus the patients with solvent-related disorders have been exposed long ago when the exposure levels, especially in construction painting, were considerably higher than nowadays. In addition, a solvent induced mild toxic encephalopathy is often a progressive disease after the cessation of exposure. There is, however, a great individual variability both in susceptibility and prognosis of the disease.27 Several studies have indicated an interaction between solvent exposure and high alcohol consumption. It has even been suggested that solvent exposure and use of alcohol are acting synergistically.28 The changes in paint technology have had a great impact on exposure to solvents in construction painting where the influence has been much more significant than in paint production because of poor ventilation. Period of high solvent exposure lasted from mid-1950’s to mid-1970’s.28 At that time, even acute intoxications occurred. Painting and lacquering using epoxyester formulations often caused very high exposure levels.24 The average exposure level to white spirit was estimated to be 130 ppm for painters when alkyd paints were extensively used. During actual painting situations, the exposure levels could rise to 300 ppm. Levels of about 200 ppm are, however, more typical in conditions of poor ventilation. If doors and windows can be kept open, the airborne white spirit concentration is significantly reduced, about to 40 ppm. In early 1980’s, the average exposure levels to white spirit were reduced to about 40 ppm in Finland.29 Nowadays, water-based paints are
18.4 Solvent uses with exposure risks
1257
also commonly used in repair and maintenance painting and the average exposure levels have become generally low.28 Occasionally, high peak exposures continue to take place when solvent-based paints are used in poorly ventilated spaces. When small items are painted industrially, the worker can stay outside the spray booth and exposure to solvents can be effectively controlled. Measurements conducted in the 1970’s already indicated that no serious problems existed in that kind of work.29 Video terminal painting is a modern example of such a situation.30 On the other hand, when large metal products, such as cars and trailers, are painted, the worker must enter the booth and control is much more difficult. Solvents are not, however, the biggest hygienic problem in car painting but the paint mist (overspray). The measurements carried out in Finland in the 1980’s indicated that the combined solvent concentration was usually less than 50% of the OEL.26 On the other hand, the paint mist concentrations exceeded clearly the standards.29 If painting is done in a poorly ventilated booth, high solvent concentrations, however, appear.30 Isocyanate-based urethane paints are nowadays common in car painting. They contain 1,6-hexamethylene diisocyanate (HDI) based polyisocyanates, which have become a major cause of occupational diseases, especially asthma.31,32 Downdraft spray-painting booths provide the best overspray control. Air velocity, flow direction, and flow homogeneity are the ventilation parameters having the largest effect on booth performance. Spray-painting booths do not, however, completely control exposure to paint mist and isocyanates.33,34 In addition to actual painting, the workers may become exposed to solvents during other tasks, such as solvent cleaning. For example, isopropanol (TLV 400 ppm, the Finnish OEL 200 ppm) is used for wiping cars before the application of the primer. If this is performed in a well-ventilated room, the concentrations remain below the OEL. As high concentrations as 130 ppm were, however, measured in a room provided with a ventilation of 50 air changes per hour.35 Solvent containing formaldehyde resin paints and lacquers have been used extensively in Nordic furniture and wood product industry. In the early 1980’s, combined solvent concentration and especially formaldehyde levels often exceeded the OELs. The OEL violations became rare in the late 1980’s.36 The recent concentrations of other solvents than ethanol (even its mean concentration was only 17 ppm) have been below 10 ppm in Finland. Nowadays, solventless acrylics are mainly used for industrial wood coatings. This substitution has, however, created a new occupational health problem. The new products have caused many cases of dermal sensitization among exposed workers. 18.4.5 PRINTING The exposure to toluene has often been extensive in the past. As high exposure estimates as 450 ppm have been given for Swedish rotogravure printing workers in the 1950’s.37 A linear correlation has been observed between airborne toluene concentration in the pressroom air and the consumption rate of toluene.20 Large day-to-day variation in the airborne concentration is, therefore, common. The long-term mean concentrations were 63-186 ppm in two Nordic studies.24,38 At least in Finland, the exposure to toluene remained rather stable between 1960-1980 because the effect of ventilation improvements was outweighed by increased production. In the early 1980’s, the Finnish rotogravure printing plants were modernized and effective ventilation systems were installed. The new presses had better enclosures and the contaminated zones around the presses were separated from the other
1258
Pentti Kalliokoski, Kai Savolainen
parts of the pressroom.39 The mean exposure level of toluene can be less than 20 ppm in a modern rotogravure printing plant.37 As stated earlier, toxic alkoxyethanols and their acetates have been used commonly in silkscreen printing inks. Those have no warning odor and are, therefore, easily considered to be safe. Airborne concentrations of 2-ethoxyethanol and 2-ethoxyethyl acetate from 3 to 14 ppm (both have a TLV of 5 ppm, the Finnish OEL 2 ppm) have been detected during printing.40,41 Because the inhalation exposure consists only a part of the total exposure and absorption through the skin is of equal or greater importance, biological monitoring is the preferred exposure assessment method. The concentration of the toxic metabolite, 2-ethoxyacetic acid, in urine is the most commonly used method. Biomonitoring is also important because, otherwise than most other solvents, 2-ethoxyacetic acid has a long half-life and, therefore, it accumulates in the body during the workweek. Its ACGIH BEI is 100 mg/g creatinine (109 mmol/mol creatinine). A lower value of 30 mmol/mol creatinine has been proposed in Finland.42 The average urinary 2-ethoxyacetic acid concentrations were found to increase from 40 to near 90 mmol/mol of creatinine during a workweek.42 Other alkoxyethanols, especially 2-methoxyethanol and 2-butoxyethanol and their acetates have also been used in silkscreen printing inks. The TLV for 2-methoxyethanol and its acetate is 5 ppm (their Finnish OEL is 0.5 ppm) and 25 ppm for 2-butoxyethanol (the Finnish OEL is 20 ppm for 2-butoxyethanol and its acetate but a value of 5 ppm has been proposed.42 ACGIH has not set BEIs for these compounds. Limits of 3 mmol methoxyacetic acid/mol creatinine and 60 mmol butoxyacetic acid/mol creatinine has been recommended in Finland for the concentrations of the urinary metabolites.41 Mean inhalation exposure levels to 2-methoxyethyl acetate from 0.2 to 1.9 ppm have ben reported.40,42 The urinary methoxyacetic acid concentrations have ranged from 0.1 to 10.7 mmol/mol creatinine.42 Airborne mean 2-butoxyethanol and 2-butoxyethylacetate concentrations from 0.1 to 3.2 ppm have been observed.40,42 The use of safer alkoxypropanols and their acetates is rather new in silkscreen printing inks. Among them, only 1-methoxy-2-propanol has a TLV (100 ppm). Its German MAK value is 50 ppm (also given for its acetate). A German MAK value of 20 ppm have been set for 2-methoxy-1-propanol and its acetate (this is also the Norwegian OEL). These have no BEIs or official biological action limits in any country. A value of 3 mmol 2-methoxypropionic acid/mol creatinine has been found to correspond to inhalation exposure to 1-methoxy-2-propyl acetate at the German MAK level.42 The corresponding value for 1-ethoxypropionic acid is 40 mmol/mol creatinine.42 Even though alkoxypropanols are less toxic than alkoxyethanols and their dermal uptakes are lower, they have the disadvantage of having higher vapor pressures. However, if enclosed and automatic machines have been used airborne levels of alkoxypropanols have remained well below the German MAK levels.42 Screen printing inks also contain other solvents, e.g., toluene, xylene, and cyclohexanone (TLV 25 ppm with skin notation, the Finnish OEL 25 ppm as a ceiling value and with skin notation). The printers are also exposed to solvents while cleaning the press. The rollers are cleaned regularly. Occasionally, ink stains are also removed from other parts of the presses and floor (see 18.4.6). 18.4.6 DEGREASING, PRESS CLEANING AND PAINT REMOVAL Chlorinated solvents have traditionally been used as metal degreasing agents. Trichloroethylene (TLV 50 ppm; the Finnish OEL 30 ppm) and 1,1,1-trichloroethane (methyl chloro-
18.4 Solvent uses with exposure risks
1259
form; TLV 350 ppm; the Finnish OEL 100 ppm) have been the most common solvents for this purpose. The use of 1,1,1-trichloroethane has, however, recently been restricted in many countries due to environmental reasons. Urinary trichloroacetic acid and trichloroethanol concentrations can be used for biological monitoring for both solvents. The ACGIH BEI for trichloroacetic acid is 100 mg/g creatinine for trichloroethylene exposure and 10 mg/l for 1,1,1-trichloroethane.15 The BEI given by the Finnish Institute of Occupational Health (FIOH) is 360 µmol/l for trichloroethylene. In Finland, the concentration of 1,1,1-trichloroethane in blood (FIOH BEI 2 µmol/l) is recommended for biological monitoring of 1,1,1-trichloroethane because less than 10% of this solvent is metabolized. The ACGIH BEI for urinary trichloroethanol is 30 mg/l for 1,1,1-trichloroethane. For trichloroethylene, the sum of urinary trichloroethanol and trichloroethanol is used (ACGIH BEI 300 mg/g creatinine; the FIOH BEI 1000 µmol/l). Chlorofluorocarbons (CFCs) are also used but their use is decreasing due to their contribution to the ozone depletion in the stratosphere.43 1,1,2-trichloro-1,2,2-trifluoroethane (Freon 113; TLV and the Finnish OEL 1000 ppm) is used especially for ultrasonic cleaning of small metal parts. Alternative degreasing methods using either alkaline aqueous solutions or citrus oil (D-limonene) have been developed.43 Exposure to trichloroethylene and 1,1,1-trichloroethane should be avoided especially due to their neurotoxic properties.44,45 Some evidence is on the carcinogenicity of trichloroethylene.9 1,1,1-Trichloroethane was earlier stabilized with carcinogenic 1,4-dioxane. Nowadays, it has usually been replaced with other stabilizers.46 1,1,1-Trichloroethane has the skin notation in the Finnish OEL list. Metal degreasing is usually performed as vapor phase operation, where solvent is heated and the vapor condenses on the metal part. An extensive occupational hygiene survey on the US vapor degreasing operations has been performed in the 1970’s. Typical sample concentrations were 100-400 ppm.50 Much lower airborne trichloroethylene concentrations were observed in Finland in the 1980’s. The mean 8 h concentration was 7 ppm (range <1-30 ppm).46 The biological monitoring, however, revealed higher exposure levels. The mean concentration of trichloroacetic acid in urine was 153 µmol/l. The highest level was 860 µmol/l.46 Airborne 1,1,1-trichloroethane levels from 34 to 85 ppm has been detected in Finland during metal degreasing in the 1980’s.46 The mean airborne concentration of trichloroethylene was 22 ppm (range 9-45 ppm) in Finland in the mid 1990’s. The level of 1,1,1-trichloroethane has varied from less than 0.1 to 1.8 µmol/l (mean 0.3 µmol/l) in blood.46 The concentrations of 1,1,2-trichloro-1,2,2-trifluoroethane have been low in Finland. The mean airborne concentration was 25 ppm (range 5-85 ppm) in the 1980’s.46 Methylene chloride (dichloromethane) is most commonly used for paint removal. The stripping solution contains 55-85% methylene chloride.47 Ethanol (TLV and the Finnish OEL 1000 ppm), methanol (TLV and the Finnish OEL 200 ppm with skin notation), toluene and formic acid are other common constituents.46,47 US National Institute of Occupational Safety and Health has recommended that methylene chloride be regarded as a potential carcinogen and that it should be controlled to the lowest feasible limit.48 Its TLV is 50 ppm (the Finnish OEL 100 ppm). Very high airborne methylene chloride concentrations have been detected in the USA during furniture stripping. The TWA concentration was above 50 ppm in all five plants investigated. The highest TWA level was as high as 854 ppm. In individual samples, concentrations up to 2160 ppm were detected.51 In Finland, the 8 h TWA concentration has varied from 26 to 380 ppm. The highest sample concentration was 850 ppm.46 Local exhaust sys-
1260
Pentti Kalliokoski, Kai Savolainen
tem based on the ACGIH manual criteria for open surface tanks51 has been found to be effective for stripping tanks.47 When different versions of such a hood was installed in an existing stripping tank, considerable improvement in airborne concentrations was achieved. Methylene concentrations were reduced from 600-1150 ppm to 28-34 ppm. By the use of side baffles and larger room area which would allow more even supply of make-up air even lower airborne levels would be possible to achieve.47 Toluene is used to clean the rotogravure presses. High exposure levels, up to 700 ppm, may occur during this task.49 Offset presses are usually cleaned with chlorinated solvents. 1,1,1-Trichloroethane and Freons were earlier commonly used as cleaning solvents.46 The mean concentration of 1,1,1-trichloroethane has been 23 ppm. The highest concentration observed in Finland has been 117 ppm.46 Solvent cleaning is a task where exposure is difficult to control with ventilation arrangements but the use of a respirator is needed. Because it usually lasts only for a short time (typically about 1 hour/week) its use is neither too heavy physically. 18.4.7 DRY CLEANING Carbon tetrachloride was the first solvent used in dry cleaning. It was introduced in the 1920’s and used until the 1950’s. The use of trichloroethylene started in the 1930’s and it is still used to a limited extent. Perchloroethylene (tetrachloroethylene) was introduced in the 1950’s and is today by far the most common dry cleaning solvent.10 Perchloroethylene is neurotoxic and NIOSH considers it as potential human carcinogen.53 Its TLV is 25 ppm and the Finnish OEL 50 ppm. The concentration of perchloroethylene in blood is used for biological monitoring. The ACGIH BEI is 0.5 mg/l and FIOH BEI 6 mol/l (1 mg/l). Petroleum solvents are a safer alternative for dry cleaning.53 Emulsion cleaning, where an alkaline mixture of surfactants is used together with a small amount of nonchlorinated organic solvent in water, can be used for textiles that tolerate water.43 Dry cleaning machines have developed safer over the time. Five generations of machines have appeared. The first generation machines, where solvent-laden clothing was manually transferred to a separate dryer, were used until the late 1960’s. This transfer was eliminated in the machines of the second generation. Residual solvents were still vented to the atmosphere. Condensers were introduced to the non-vented, third generation machines in the late 1970’s. The fourth generation machines have in addition to the refrigerated condenser an activated carbon absorber to control solvent vapor at the cylinder outlet at the end of the drying cycle. The fifth generation machines are also provided with a monitor and an interlocking system to ensure that the concentration of perchloroethylene is below 300 ppm when the door is opened.54 The TWA exposure level was 40-60 ppm when the first generation machines were used. With the second and third generation machines the TWA exposure level was 15-20 ppm. All these machines, however, created short-term peak concentrations up to 1000-4000 ppm. The fourth generation machines reduced the TWA exposure level below 3 ppm, and when the fifth generation machines are used TWA concentration remains below 2 ppm. With the fourth and fifth generation machines, the peak concentrations are 10-300 ppm.54 Local ventilation hoods can be used above the door as an additional control measure.55 In addition to the cleaners, pressers, sewers, counters, and maintenance workers may be exposed to some extent.56 The mean 8 h TWA concentration of perchloroethylene was 13 ppm (range 3-29 ppm) in Finland. The mean concentration of perchloroethylene in the blood of Finnish dry clean-
18.4 Solvent uses with exposure risks
1261
ing workers was 1.3 µmol/l (range <0.1 - 14.0 µmol/l).46 The average value for maintenance men was 0.3 µmol/l. Their highest value was 1.3 µmol/l.46 18.4.8 REINFORCED PLASTICS INDUSTRY Exposure to styrene is the main occupational hygiene problem in reinforced plastics industry, where it is used as a crosslinking agent and solvent in unsaturated polyester resins. In addition, workers are exposed to acetone which is used as a clean-up solvent. Other solvents, such as methylene chloride, toluene, xylene, heptane (TLV 400 ppm, the Finnish OEL 300 ppm), methylcyclohexane (TLV and the Finnish OEL 400 ppm), and butyl acetate (TLV and the Finnish OEL 150 ppm) may also be used.57 Styrene is neurotoxic.58 Styrene is also a suspected carcinogen because it is metabolized via styrene-7,8-oxide.59 The TLV and the Finnish OEL of styrene is 20 ppm. Urinary mandelic acid concentration is the most common biological monitoring method for styrene. The ACGIH BEI is 800 mg/g creatinine and the FIOH BEI 3.2 mmol/l. Open mold methods (hand lay-up and spray-up methods) are most commonly used. In addition, the products often have large surface areas and large styrene vapor emissions take place. The emission depends on the type of resin used. While the mean evaporation loss of styrene was 11.6% for conventional resins, it was only 4.4% for the low styrene emission (LSE) resins containing volatilization inhibitors.60 The reinforced plastics industry began in 1950’s. The airborne concentrations of styrene were typically 100-200 ppm until the 1980’s.61 Styrene concentrations of about 50 ppm were found possible to achieve during the production of small and medium-sized products in the 1980’s by using a mobile fan or a set of successive air jets to blow fresh air from behind the worker towards the mold. The prevention of eddy formation over the mold was found to be of crucial importance.62 When a large occupational hygiene survey was conducted in Finland in the early 1990’s, it appeared that styrene concentrations had remained high. As many as 77% of the 8-h TWA concentrations exceeded the Finnish OEL of 20 ppm. The mean TWA concentration was 43 ppm and the highest TWA value 123 ppm.63 The average concentrations of styrene oxide were low, on an average one pre mil of the styrene concentration (mean concentration was 0.1 ppm).63 The mean concentration of acetone was 78 ppm (range 3-565 ppm).63 Airborne levels of styrene were still high in the Finnish boat manufacturing plants in the mid 1990’s (mean concentration 51 ppm, range 15-89 ppm). The mean acetone concentration was 25 ppm (range 9-53 ppm). Lower concentrations of about 10 ppm were reported from Sweden.64 However, no detailed process and ventilation description was given. The mean urinary mandelic acid concentration was about 5 mmol/l in the late 1970’s in Finland. The level was reduced to about 4 mmol/l in mid-1980’s and to 35 mmol/l in late 1980’s. However, the maximum levels remained above 30 mmol/l throughout the whole period.65 Recent studies have indicated that sophisticated ventilation arrangements are needed to reduce the concentration of styrene below 20 ppm. In hand lay-up work, this is possible by the use of a mobile supply ventilation and exhaust ventilation.66 In the spray-up work, concentrations below 20 ppm can be achieved by using a booth provided with mobile curtains to reduce the open face area and by maintaining a control velocity of 0.35 m/s.66
1262
Pentti Kalliokoski, Kai Savolainen
18.4.9 GLUING Shoe manufacturing is an industry where toxic solvents have been used. Benzene containing glues were used until mid-1960’s, at least in Italy.67 A significant part of the information on the toxic and carcinogenic effects of benzene on the hematopoietic system originates from this industry.68 After benzene, n-hexane became the most important solvent.67 Epidemiological evidence on neuropathy caused by exposure to n-hexane comes again largely from shoe manufacturing.5 The TLV and the Finnish OEL of hexane is 50 ppm. The concentration of the toxic metabolite, 2,5-hexanedione is used for biological monitoring of exposure to n-hexane. The BEI given by the FIOH is 12 µmol/l. The ACGIH BEI is 5 mg/g creatinine. Shoe glues may also contain toluene, methyl ethyl ketone, and acetone.69 The mean airborne concentration of n-hexane was 38 ppm (range 4-90 ppm) in the Finnish shoe industry in the 1980’s.26 The mean concentrations of acetone and toluene were 131 ppm and 14 ppm.26 The mean exposure levels in two US shoe manufacturing plants in the late 1970’s were: toluene 22-50 ppm, methyl ethyl ketone 133-153 ppm, acetone 46-223 ppm, and n-hexane 22-55 ppm.69 The exposure is, however, easy to control by doing the gluing in enclosure hoods. Contact glues contain usually white spirit with low boiling point range (60-90oC). Its Finnish OEL is 350 mg/m3. The solvent contains n-hexane, earlier usually 5% or more. In the 1990’s, white spirits containing only ca. 2% n-hexane have become available. Contact glues are used to fasten seals and strips e.g., in construction and transportation vehicle industries. Levels have usually been below the OEL in Finland. If contact glues are used for floor coverings, very high exposures may occur. These glues have, however, largely been replaced by ethanol- and water-based glues.24 18.4.10 OTHER In the production of viscose, the workers are exposed to carbon disulfide. In addition to inhalation exposure, absorption through the skin may be important.70 It is strongly neurotoxic.71 Its TLV is 10 ppm and the Finnish OEL 5 ppm. Both carry the skin notation. Urinary 2-thiothiazolidine-4-carboxylic acid concentration can be used for biological monitoring of exposure to carbon disulfide. Its ACGIH BEI is 5 mg/g creatinine (4 mmol/mol creatinine) and the BEI given by the FIOH 2.0 mmol/mol creatinine. Airborne concentrations from less than 0.2 to 65.7 ppm were detected in measurements conducted in a German plant in the 1990’s. The urinary 2-thiothiazolidine4-carboxylic acid levels ranged from less than 0.16 to 11.6 mg/g creatinine.70 A long time average of about 8 ppm has been reported in a Dutch factory.72 In acetate fiber plants, the workers are exposed to acetone. The mean concentration in the air of three Japanese factories was 372 ppm. Some workers’ exposure levels exceeded 1000 ppm.73 Exposure to various solvents takes place in pharmaceutical industry. The airborne 8-h TWA concentrations of methylene chloride were 16-167 ppm in a Finnish factory in the 1980’s.46 Even though the active ingredients cause the main health risk while pesticides are used, but the exposure to solvents may also be important. As many as 71 different solvents were used in the 8000 pesticide formulations sold in Italy in the 1990’s. Among the solvents, carcinogens such as benzene, carbon tetrachloride, and chloroform were found.74 Cyclohexanone is used as a coating solvent in audio and video tape production. Airborne TWA concentrations from 2 to 20 ppm were detected in factory in Singapore.75 Uri-
18.4 Solvent uses with exposure risks
1263
nary cyclohexanol level of 54.5 mg/l (23.3 mg/g creatinine) was found to correspond inhalation exposure to 25 ppm of cyclohexanone.75 18.4.11 SUMMARY Even though organic solvents have been increasingly replaced by water-based and solid formulations, they are still widely used. About 3% of the Finnish working population has estimated to have significant solvent exposure.26 It should also be noted that solventless products may cause new health problems and the knowledge of the health risks of the substitutions is often poor.43 An extensive survey of chemical exposure conducted by FIOH in Finland in the 1980’s. The highest exposure levels were detected in shoe gluing, painting of furniture and other wood products, silkscreen and rotogravure printing, floor lacquering, reinforced plastics production, paint removing, metal degreasing, and various solvent cleaning operations.25,45,64 Mostly, inhalation exposure can be effectively controlled by enclosures and ventilation. Open mold methods in the reinforced plastics industry and paint stripping, however, require sophisticated ventilation arrangements. High exposures often occur during solvent cleaning operations. Exposure levels may be very high while cleaning pots in paint and printing ink industry. The use of a proper respirator is necessary during such a work. The avoidance of high peak exposures is important also to prevent development of addiction. The workers may adopt working habits that cause unnecessary exposure. In extreme causes, even this may lead to solvent sniffing.24 Recent studies have revealed that percutaneous absorption is often an important route of solvent exposure. As the control of respiratory exposures develops, the relative importance of dermal exposures increases. REFERENCES 1 2 3 4 5. 6 7 8 9 10 11 12 13 14 15 16 17
18 19 20 21
C. Hogstedt, Scand. J. Work Environ. Health, 20, 59 (1994). C. Edling, A. Lindberg, and J. Ulfberg, Br. J. Ind. Med., 50, 276 (1993). L. Barregård and A. Axelsson, Scand. Audiol., 13, 151 (1984). L. Low, J. Meeks, and C. Mackerer in Ethel Browning’s Toxicity and Metabolism of Industrial Solvents, R. Snyder, Ed., Elsevier, Amsterdam, 1987, pp. 297-311. D. Couri and M.Milks, Annual Review of Pharmacology and Toxicology, 22, 145 (1982). S. Sjöberg., J .Dahlqvist, S. Fregert, Contact Dermatitis, 8, 207 (1982). A. Tossavainen and J. Jaakkola, Appl. Occup. Environ. Hyg., 9, 28 (1994). P. Delore and C. Borgomano, J. Med. Lyon, 9, 227 (1928). IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 47, Lyon, 1989. D. McGregor, E. Heseltine, and H. Moller, Scand. J. Work Environ. Health, 21, 310 (1995). A. Anttila, E. Pukkala, R. Riala et al., Int. Arch. Occup. Environ. Health, 71, 187 (1998). H-W. Leung and D. Paustenbach, Appl. Occup. Environ. Hyg., 9, 187 (1994). G. Johanson and A. Boman, Br. J. Ind. Med., 48, 788 (1991). V. Fiserova-Bergerova, Ann. Occup. Hyg., 34, 639 (1990). ACGIH, TLVs and BEIs, 1999 R. Soule in Patty’s Industrial Hygiene and Toxicology, G. Clayton and F. Clayton, Eds, Vol.1, Part A, John Wiley and Sons, New York, 1991, pp. 137-194. E. Andersson, N. Browne, S. Duletsky et al. Development of Statistical Distributions of Ranges of Standard Factors Used in Exposure Assessments, National Technical Information Service, U.S. Department of Commerce, 1985. B-S. Cohen and W. Popendorf, Am. Ind. Hyg. Assoc. J., 50, 216 (1989). J. Perkins and B. Pool, Am. Ind. Hyg. Assoc. J., 58, 474 (1997). P. Kalliokoski, Am. Ind. Hyg. Assoc. J., 51, 310 (1990). I. Lundberg and R. Milatou-Smith, Scand. J. Work Environ.& Health, 24, 270 (1998).
1264
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69
Pentti Kalliokoski, Kai Savolainen
B. Svensson, G. Nise, V. Englander et al., Br. J. Ind. Med., 47, 372 (1990). J. Laitinen, Int. Arch. Occup. Health, 71, 117 (1998). P. Kalliokoski in Safety and Health Aspects of Organic Solvents, V. Riihimäki and U. Ulfvarson, Eds., Alan R. Liss, New York, 1986, pp. 21-30. D. Rees, N. Soderlund, R. Cronje et al., Scand. J. Work Environ. Health, 19, 236 (1993). H. Riipinen, K. Rantala, A. Anttila, Organic Solvents, Finnish Institute of Occupational Health, Helsinki, 1991 (in Finnish). C. Edling, K. Ekberg, G. Ahlborg et al., Br. J. Ind. Med., 47, 75 (1990). I. Lundberg, H. Michelsen, G. Nise et al., Scand. J. Work Environ. Health, 21, suppl.1 (1995). R. Riala, P. Kalliokoski, G. Wickström et al., Scand. J. Work Environ. Health, 10, 263 (1984). J-D. Chen, J-D. Wang, J-P. Jang et al., Br. J. Ind. Med., 48, 696 (1991). A. Musk, J. Peters, D. Wegman, Am. J. Ind. Med., 13, 331 (1988). M. Janko, G. McCarthy, M. Fajer et al., Am. Ind. Hyg. Assoc. J., 53, 331 (1992). W. Heitbrink, M. Wallace, C. Bryant, Am. Ind. Hyg. Assoc. J., 56, 1023 (1995). N. Goyer, Am. Ind. Hyg. Assoc. J., 56, 258 (1995). D. George, M. Flynn, R. Harris, Am. Ind. Hyg. Assoc. J., 56, 1187 (1995). E. Priha, Exposure to Gaseous and Material-bound Formaldehyde in the Wood and Textile Industries, Ph.D Thesis, University of Kuopio, 1996. B-G. Svensson, G. Nise, E. Erfurth et al., Br. J. Ind. Med., 49, 402 (1992). P. Övrum, M. Hultengren, T. Lindqvist, Scand. J. Work Environ. Health, 4, 237 (1978). P. Kalliokoski in Ventilation ‘88, J. Vincent, Ed., Pergamon Press, London, 1988, pp. 197-200. H. Veulemans, D. Groeseneken, R. Masschelein et al., Scand. J. Work Environ. Health, 13, 239 (1987). R. Vincent, B. Rieger, I. Subra et al., Occup. Hyg., 2, 79 (1996). J. Laitinen, Biomonitoring and Renal Effects of Alkoxyalcohols and Their Acetates Among Silkscreen Printers, PhD Thesis, University of Kuopio, 1998. A-B. Antonsson, Am. Ind. Hyg. Assoc. J., 56, 394 (1995). A. Seppäläinen and M. Antti-Poika, Scand. J. Work Environ. Health, 9, 15 (1983). M Ruijten, M. Verberk, H. Salle, Br. J. Ind. Med., 48, 87 (1990). K. Rantala, H. Riipinen, A. Anttila, Halogenated Hydrocarbons, Finnish Institute of Occupational Health, Helsinki, 1992 (in Finnish). C. Estill and A. Spencer, Am. Ind. Hyg. Assoc. J., 57, 43 (1996). NIOSH, Current Intelligence Bulletin 46: Methylene Chloride, DHHS (NIOSH) Publication No. 86-114, Cinicinnati, 1986. P. Kalliokoski, Toluene Exposure in Finnish Publication Rotogravure Plants, PhD Thesis, University of Minnesota, 1979. W. Burgess, Recognition of Health Hazards in Industry, John Wiley & Sons, New York, 1981. C. McCammon, R. Glaser, V. Wells et al., Appl. Occup. Environ. Hyg., 6, 371 (1991). ACGIH, Industrial Ventilation, A Manual of Recommended Practice, Cincinnati, 1988. NIOSH, Hazard Controls, HC 17, DHHS (NIOSH) Publication No. 97-155, Cincinnati,1997. NIOSH, Hazard Controls, HC 18, DHHS (NIOSH) Publication No. 97-156, Cincinnati, 1997. NIOSH, Hazard Controls, HC 19, DHHS (NIOSH) Publication No. 97-157, Cincinnati, 1997. A. Blair, P. Stewart, P. Tolbert et al., Br. J. Ind. Med., 47, 162 (1990). G. Triebig, S. Lehrl, D. Weltle et al., Br. J. Ind. Med., 46, 799 (1989). N. Cherry and D. Gautrin, Br. J. Ind. Med., 47, 29 (1990). M. Severi, W. Pauwels, P. Van Hummelen et al., Scand. J. Work Environ. Health, 20, 451, 1994. P. Kalliokoski, R. Niemelä, A. Säämänen et al. In Ventilation ‘94, A. Janson and L. Olander, Eds., Stockholm, 1994, pp. 207-210. WHO, Styrene, Environmental Health Criteria 26, Geneva, 1983. P. Kalliokoski, A. Säämänen, L. Ivalo et al., Am. Ind. Hyg. Assoc. J., 49, 6, 1988. P. Pfäffli, L. Nylander, A. Säämänen et al., Työ ja Ihminen, 2, 213, 1992 (in Finnish, English summary). U. Flodin, K. Ekberg, L. Andersson, Br. J. Ind. Med., 46, 805, 1989. A. Säämänen, A. Anttila, P. Pfäffli, Styrene, Finnish Institute of Occupational Health, Helsinki, 1991 (in Finnish). A. Säämänen, Methods to Control Styrene Exposure in the Reinforced Plastics Industry, Publications of the Technical Research Centre in Finland No. 354, Espoo, 1998. E. Paci, E. Buiatti, A. Constantini et al., Scan. J. Work Environ. Health, 15, 313, 1989. E. Vigliani and G. Saita, N. Engl. J. Med., 271, 872, 1964. J. Walker, T. Bloom, F. Stern et al., Scand. J. Work Environ. Health, 19, 89, 1993.
18.4 Solvent uses with exposure risks
70 71 72 73 74 75
H. Drexler, Th. Göen, J. Angerer, Int. Arch. Occup. Environ. Health, 67, 5, 1995. WHO, Carbon disulfide, Environmental Health Criteria 10, Geneva, 1979. M. Ruijten, H. Salle, M. Verberk, Br. J. Ind. Med., 50, 301, 1993. A. Fujino, T. Satoh, T. Takebayashi et al., Br. J. Ind. Med., 49, 654, 1992. G. Petrelli, G. Siepi, L. Miligi et al., Scand. J. Work Environ. Health, 19, 63, 1993. C-N. Ong, S-E Chia, W-H. Phoon et al., Scand. J. Work Environ. Health, 17, 430, 1991.
1265
19
Regulations Carlos M. NuNez U.S. Environmental Protection Agency National Risk Management Research Laboratory Research Triangle Park, NC, USA
19.1 INTRODUCTION Since the Clean Air Act Amendments of 1970, when a war against pollution was declared, solvents have been on the U.S. regulatory radar screen.1 This represented the beginning of new environmental policies which led to today’s more stringent regulations with a greater focus on toxic substances.2 In the U.S., historically, these policies and regulations have been divided by medium (air, water, and land), which has represented a challenge to the regulated industries and the Federal, State, and local regulatory communities. In recent years, however, the U.S. Congress and the Environmental Protection Agency (EPA) have recognized the need to address environmental problems from a more holistic approach considering multimedia and innovative environmental management strategies.3 Hence, various programs and initiatives have emerged, which have proven to be extremely successful, encouraging voluntary industry participation rather than the old command and control approach.4 Almost all solvents are volatile organic compounds (VOCs) and hazardous air pollutants (HAPs), and their evaporation creates environmental problems that have become the focus of many domestic and international regulations and initiatives. A VOC solvent is defined by EPA as any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates, and ammonium carbonate, which is emitted or evaporated into the atmosphere.5 In 1996, an estimated 19 million short tons of VOCs were emitted in the U.S. of which 33 percent came from solvent utilization (see Table 19.1).6 In the lower atmosphere VOC solvents participate in photochemical reactions to form, to varying degrees, ground level ozone and other oxidants which affect health, as well as cause damage to materials, crops, and forests. Ozone impairs normal functioning of the lungs and reduces the ability to perform physical exercise. Such effects are more severe in individuals with sensitive respiratory systems. Even healthy adults can experience symptoms and reduction in lung function during moderate exercise at ozone levels below the current ozone standard. Some solvents are also toxic and/or carcinogenic which contributes to direct health problems. They are pollutants that have been associated with serious health effects such as cancer, liver or kidney damage, reproductive disorders, and developmental or neurological problems.7,8 They also have detrimental environmental effects on wildlife and degrade wa-
1268
Carlos M. NuÔez
ter or habitat quality. The Clean Air Act of 1990 in section 112(b) lists 188 HAPs,9 some of which are solvents (see Table 19.2),10-14 for which sources are identified and regulated. The contents of Table 19.2 will be discussed in more detail later. Table 19.1. Major categories in the solvent utilization sector and their estimated VOC emissions6 Estimated Emissions, 1996 Source category
Quantity, tons
distribution in solvent utilization sector, %
Degreasing
661,000
11
Graphic arts
389,000
6
Dry cleaning
190,000
3
Surface coating
2,881,000
46
Other industrial and non-industrial processes
2,153,000
34
Total emissions
6,274,000
100
Table 19.2. Classification of solvents by their codes, exposure limits, and environmental effects
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv K009; K010; K040; K156; K157
mg/m3
TLV14 ppmv
see 29 CFR 1910.1048
Environmental Effect
O T D G ✓
50-00-0
Formaldehyde
50-21-5
Lactic acid
✓
50-70-4
Sorbitol
✓
56-23-5
Carbon tetrachloride
F001; F024; F025; K016; K019; K020; K021; K073; K116; K150; K151; K157; U211; D019
56-38-2
Parathion
P089
56-81-5
Glycerine
56-93-9
Benzyltrimethylammonium chloride
57-14-7
1,1-Dimethyl hydrazine
57-55-6
Propylene glycol
57-57-8
β-Propiolactone
60-29-7
Ethyl ether
✓
✓
✓ ✓ ✓
3
✓
✓
3
✓
(2)
5
(S) 0.1
0.1 mg/m 10 mg/m
✓ K017; K108; K109; K110; U098
(S) 0.5
✓
(S) 1
✓
✓
see 29 CFR 1910.1013 U117
400
1200
0.5
✓ ✓
✓
19 Regulations
1269
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv 60-34-4
Methyl hydrazine
P068
62-53-3
Aniline
K083; K103; K104; U012
62-75-9
N-Nitrosodimethylamine P082
64-17-5
Ethyl alcohol
64-18-6
Formic acid
64-19-7
Acetic acid
64-67-5
Diethyl sulfate
67-56-1
Methanol
67-63-0
Isopropyl alcohol
67-64-1
Acetone
U002
67-66-3
Chloroform
K009; K010; K019; K020; K021; K029; K073; K116; K149; K150; K151; U044; D022
67-68-5
Dimethyl sulfoxide
68-12-2
Dimethyl formamide
71-23-8
n-Propyl alcohol
71-36-3
n-Butyl alcohol
U031
71-43-2
Benzene
F005; F024; F025; F037; F038; K085; K104; K105; K141; K142; K143; K144; K145; K147; K151; K159; K169; U019; D018
71-55-6
1,1,1-Trichloroethane
F001, F002, F024, F025, K019, K020, K028, K029, U226
74-88-4
Methyl iodide
U138
74-96-4
Ethyl bromide
75-00-3
Ethyl chloride
75-04-7
Ethylamine
K009; K010; U123
U154
mg/m3
(C) (S) 0.2 (C)(S)0.35
(S) 5
(S) 19
see 29 CFR 1910.1016
TLV14 ppmv
Environmental Effect
O T D G
0.01
✓
✓
2
✓
✓
(S)
✓
✓
✓
1000
1900
5
9
5
✓
10
25
10
✓ ✓
✓
200
260
200
✓
✓
400
980
400
✓
1000
2400
500
ε
(C) 50
(C) 240
10
✓
✓
✓
K018
10
✓
(S) 10
(S) 30
200
500
✓
100
300
✓
✓
see 29 CFR 1910.1028 Table Z-2
0.5
✓
✓
350
1900
350
ε
✓ ✓ ✓
(S) 5
(S) 28
2
✓
✓
200
890
5
✓
1000
2600
100
✓
10
18
5
✓
✓
1270
Carlos M. NuÔez
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
40
70
200 (C) 10
TLV14 ppmv
Environmental Effect
O T D G
75-05-8
Acetonitrile
K011; K013; K014; U003
75-07-0
Acetaldehyde
U001
75-08-1
Ethyl mercaptan
75-09-2
Dichloromethane
F001; F002; F024; F025; K009; K010; K156; K158; U080
(2)
50
ε
✓
75-15-0
Carbon disulfide
F005; P022
(2)
10
✓
✓
75-18-3
Dimethyl sulfide
75-25-2
Bromoform
75-31-0
Isopropylamine
75-33-2
Isopropyl mercaptan
75-34-3
1,1-Dichloroethane
F024, F025; U076
75-35-4
Vinylidene chloride
F024; F025; K019; K020; K029; U078; D029
75-52-5
Nitromethane
75-55-8
Propyleneimine
75-56-9
Propylene oxide
75-64-9
tert-Butylamine
75-65-0
tert-Butyl alcohol
75-68-3
1-chloro-1,1-difluoroethane F001 (HCFC-142b)
ε
✓ ✓
75-69-4
Trichlorofluoromethane (CFC-11)
ε
✓ ✓
75-83-2
Neohexane
76-13-1
1,1,2-Trichloro-1,2,2trifluoroethane (CFC-113)
76-14-2
1,2-Dichloro-1,1,2,2F001 tetrafluoroethane(CFC-114)
77-47-4
Hexachlorocyclopentadiene
F024; F025; K032; K033; K034; U130
77-78-1
Dimethyl sulfate
K131; U103
78-59-1
Isophorone
✓
✓
360
✓
✓
(C) 25
✓
40
✓
U225
(S) 0.5
(S) 5
0.5
✓
5
12
5
✓
✓
✓
P067
100
400
100
✓
✓
5
✓
✓
100
250
20
✓
2
5
2
✓
✓
100
240
20
✓
✓
✓
100
300
100
F001; F002; F024; F025; U121
✓
✓
F001; F002
1000
7600
1000
ε
✓ ✓
1000
7000
1000
ε
✓ ✓
0.01
✓
✓
0.1
✓
✓
✓
✓
(S) 1
(S) 5
25
140
19 Regulations
1271
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
78-78-4
Isopentane
✓
78-81-9
Isobutylamine
✓
78-83-1
Isobutyl alcohol
78-84-2
Isobutyraldehyde
78-87-5
1,2-Dichloropropane
U083
78-88-6
2,3-Dichloropropene
F024; F025
78-92-2
sec-Butyl alcohol
78-93-3
Methyl ethyl ketone
78-96-6
Monoisopropanolamine
79-01-6
Trichloroethylene
79-04-9
Chloroacetyl chloride
79-09-4
Propionic acid
79-10-7
Acrylic acid
79-20-9
Methyl acetate
79-21-0
Peracetic acid
79-24-3
Nitroethane
F005; U140
100
300
✓
50
✓
F005; U159; D035
75
350
✓
75
✓
✓
150
450
100
✓
200
590
200
✓
✓
✓ F001; F002; F024; F025; K018; K019; K020; U228; D040
✓
(2)
U008 200
610
0.05 (S)
✓
10
✓
2
✓
250
ε
✓
✓
✓
100
310
100
✓
5
35
1
ε
✓
✓
✓
79-34-5
1,1,2,2-Tetrachloroethane
F001; F024; F025; K019; K020; K030; K073; K095; K150; U209
79-44-7
Dimethyl carbamoyl chloride
U097
79-46-9
2-Nitropropane
F005; U171
25
90
10
✓
✓
80-62-6
Methyl methacrylate
U162
100
410
100
✓
✓
84-66-2
Diethyl phthalate
U088 U069
✓
5 5
3
✓
84-74-2
Dibutyl phthalate
85-68-7
Butyl benzyl phthalate
87-68-3
Hexachlorobutadiene
90-02-8
Salicyl aldehyde
✓
91-17-8
Decahydronapthalene
✓
91-22-5
Quinoline
✓
5 mg/m
✓
✓ F024; F025; K016; K018; K030; D033
0.02
✓
✓
✓
1272
Carlos M. NuÔez
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
TLV14 ppmv
Environmental Effect
ppmv
mg/m3
100
435
100
✓
(C) 50
(C) 300
25
✓
O T D G
95-47-6
o-Xylene
U239
95-50-1
1,2-Dichlorobenzene
F024; F025; K042; K085; K105; U070
96-09-3
Styrene oxide
96-12-8
1,2-Dibromo-3-chloropropane
96-33-3
Methyl acrylate
96-37-7
Methyl cyclopentane
✓
97-63-2
Ethyl methacrylate
✓
97-64-3
Ethyl lactate
✓
97-88-1
n-Butyl methacrylate
✓
98-00-0
Furfuryl alcohol
98-01-1
Furfural
U125
98-07-7
Benzotrichloride
K015; K149; U023
98-56-6
p-Chlorobenzotrifluoride
98-82-8
Cumene
U055
98-95-3
Nitrobenzene
F004; K083; K103; K104; U169; D036
99-87-6
p-Cymene
✓
100-37-8
Diethylethanolamine
✓
100-39-0
Benzyl bromide
✓
100-41-4
Ethyl benzene
100-42-5
Styrene
100-44-7
Benzyl chloride
100-46-9
Benzylamine
✓
100-47-0
Benzonitrile
✓
100-51-6
Benzyl alcohol
✓
100-52-7
Benzaldehyde
✓
100-61-8
N-Methylaniline
101-84-8
Diphenyl ether
✓
103-11-7
2-Ethylhexyl acrylate
✓
U066
✓
✓
✓
see 29 CFR 1910.1044
✓
✓
10
✓
35
50
200
10
✓
(S) 5
(S) 20
2
✓ ✓
✓
ε (S) 50
(S) 245
50
✓
✓
(S) 1
(S) 5
1
✓
✓
100
K015; K149; K085; P028
1
2
435
100
✓
✓
(2)
20
✓
✓
✓
✓
5
9
0.5
✓
19 Regulations
1273
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
103-23-1
Dioctyl adipate
✓
104-72-3
Decylbenzene
✓
104-76-7
2-Ethyl hexanol
✓
105-39-5
Ethyl chloroacetate
✓
105-46-4
sec-Butyl acetate
105-54-4
Ethyl butyrate
✓
105-58-8
Diethyl carbonate
✓
106-35-4
Ethyl butyl ketone
106-42-3
p-Xylene
106-43-4
p-Chlorotoluene
✓
106-63-8
Isobutyl acrylate
✓
106-88-7
1,2-Epoxybutane
106-89-8
Epichlorohydrin
106-91-2
Glycidyl methacrylate
106-93-4
Dibromoethane
106-99-0
Butadiene
107-02-8
Acrolein
107-03-9
n-Propyl mercaptan
107-05-1
Allyl chloride
F024, F025
107-06-2
1,2-Dichloroethane
F024; F025; K019; K020; K029; K030; U077; D028
107-07-3
Ethylene chlorohydrin
107-10-8
n-Propylamine
107-13-1
Acrylonitrile
107-15-3
Ethylenediamine
107-18-6
Allyl alcohol
107-21-1
Ethylene glycol
107-30-2
Chloromethyl methyl ether
107-31-3
Methyl formate
200
U239
950
200
50
230
50
✓
100
435
100
✓
0.3 K017; U041
✓
(S) 5
(S) 19
✓
✓
✓
✓
✓
✓
(2)
✓
✓
1000
2200
✓
✓
0.1
0.25
✓
✓
U067
P003
✓
1
5
3
1
✓
✓
(2)
10
✓
✓
16
(S)
✓ ✓
K011; K013; U009
see 29 CFR 1910.1045
2
✓
✓
✓
P005
U046
(S) 2
(S) 5
(2)
see 29 CFR 1910.1006 100
250
100
✓ ✓
✓
✓
✓
✓
1274
Carlos M. NuÔez
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
107-39-1
Diisobutylene
✓
107-41-5
Hexylene glycol
✓
107-66-4
Dibutylphosphate
107-83-5
Isohexane
107-87-9
Methyl propyl ketone
107-92-6
n-Butyric acid
107-98-2
Propylene glycol methyl ether
108-03-2
1-Nitropropane
108-10-1
Methyl isobutyl ketone
108-11-2
1
5
1
✓ ✓
200
700
200
✓ ✓
100
✓
25
✓
25
90
100
410
✓
Methylamyl alcohol
(S) 25
(S) 100
✓
108-18-9
Diisopropylamine
(S) 5
(S) 20
5
✓
108-20-3
Isopropyl ether
500
2100
250
✓
108-21-4
Isopropyl acetate
108-38-3
m-Xylene
U239
108-39-4
m-Cresol
F004; U052; D024
108-82-7
Diisobutylcarbinol
108-83-8
Diisobutyl ketone
108-84-9
Methyl amyl acetate
✓
108-86-1
Bromobenzene
✓
108-88-3
Toluene
F005; F024; F025; K015; K027; K036; K037; K149; U220
108-90-7
Chlorobenzene
F002; F024; F025; K015; K085; K105; U037; D021
108-91-8
Cyclohexylamine
108-93-0
Cyclohexanol
108-94-1
Cyclohexanone
U057
108-95-2
Phenol
K001; K022; K050; K060; K087; U188
109-06-8
2-Methylpyridine
U191
109-60-4
n-Propyl acetate
200
840
200
✓
109-66-0
Pentane
1000
2950
600
✓
U161
✓
250
950
250
✓
100
435
100
✓
✓
(S) 5
(S) 22
5
✓
✓
✓
50
75
290
25
✓
(2)
50
✓
✓
350
10
✓
✓
10
✓
50
200
50
✓
50
200
25
✓
(S) 5
(S) 19
5
✓ ✓
✓
19 Regulations
1275
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G ✓
109-67-1
1-Pentene
109-73-9
Butylamine
109-78-4
Ethylene cyanohydrin
109-79-5
n-Butyl mercaptan
109-86-4
Ethylene glycol monomethyl ether
109-86-4
Methyl cellosolve
(S) 25
(S) 80
109-87-5
Dimethoxymethane
1000
3100
109-87-5
Methyl formal
109-89-7
Diethylamine
25
75
109-94-4
Ethyl formate
100
300
110-12-3
Methyl isoamyl ketone
100
475
50
✓
110-19-0
Isobutyl acetate
150
700
250
✓
110-43-0
Methyl n-amyl ketone
100
465
50
✓
110-49-6
Methyl cellosolve acetate
(S) 25
(S) 120
5
✓
110-54-3
Hexane
500
1800
50
✓
110-63-4
1,4 - Butanediol
✓
110-66-7
n-Amyl mercaptan
✓
110-71-4
Ethylene glycol dimethyl ether
✓
110-80-5
Cellosolve
F005; U359
110-82-7
Cyclohexane
U056
110-86-1
Pyridine
F005; K026; K157; U196; D038
110-87-2
Dihydropyran
110-91-8
Morpholine
(S) 20
(S) 70
20
✓
111-15-9
Cellosolve acetate
(S) 100
(S) 540
5
✓
111-15-9
Ethylene glycol monoethyl ether acetate
✓
111-27-3
n-Hexanol
✓
111-40-0
Diethylenetriamine
111-42-2
Diethanolamine
✓
(C) (S) 5 (C) (S) 15
✓
10
35
✓
0.5
✓ ✓
5
✓ ✓ ✓
5 (S)
✓
(S) 200
(S) 740
300
1050
(300)
✓
5
15
5
✓
✓
✓
✓
✓
1 (S) 2 mg/m
3
✓
✓
1276
Carlos M. NuÔez
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
111-46-6
Diethylene glycol
✓
111-49-9
Hexamethyleneimine
✓
111-55-7
Ethylene glycol diacetate
✓
111-65-9
Octane
111-66-0
1-Octene
111-76-2
2-Butoxyethanol
111-76-2
Ethylene glycol monobutyl ether
✓
111-77-3
Diethylene glycol monomethyl ether
✓
111-84-2
Nonane
111-90-0
Diethylene glycol monoethyl ether
✓
111-92-2
Di-n-butylamine
✓
111-96-6
Diethylene glycol dimethyl ether
✓
112-07-2
Ethylene glycol monobutyl ether acetate
✓
112-30-1
n-Decyl alcohol
✓
112-34-5
Diethylene glycol monobutyl ether
✓
112-41-4
1-Dodecene
✓
112-50-5
Ethoxy triglycol
✓
112-55-0
Lauryl mercaptan
✓
112-80-1
Oleic acid
✓
115-10-6
Dimethyl ether
✓
17-81-7
Bis(2-ethylhexyl)phthalate
U028
121-44-8
Triethylamine
K156; K157; U404
121-69-7
N,N-dimethylaniline
123-38-6
Propionaldehyde
see 29 CFR 1910.1013
123-42-2
Diacetone alcohol
50
240
50
✓
123-51-3
Isoamyl alcohol
100
360
100
✓
500
2350
300
✓ ✓
(S) 50
(S) 240
(25)
200
5
✓
✓
✓
✓
25
100
1
✓
✓
(S) 5
(S) 25
5
✓
✓
✓
✓
19 Regulations
1277
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G ✓
123-62-6
Propionic anhydride
123-63-7
Paraldehyde
123-72-8
n-Butyraldehyde
123-86-4
n-Butyl acetate
123-91-1
1,4-Dioxane
123-92-2
Isoamyl acetate
124-11-8
1-Nonene
✓
124-17-4
Diethylene glycol monobutyl ether acetate
✓
124-40-3
Dimethylamine
126-33-0
Sulfolane
126-99-8
Chloroprene
127-18-4
Tetrachloroethylene
127-19-5
N,N-Dimethylacetamide
131-11-3
Dimethyl phthalate
131-18-0
Di-n-amyl phthalate
140-88-5
Ethyl acrylate
141-32-2
n-Butyl acrylate
141-43-5
Ethanolamine
141-43-5
Monoethanolamine
141-78-6
Ethyl acetate
141-79-7
Mesityl oxide
141-97-9
Ethyl acetoacetate
142-82-5
Heptane
142-84-7
Di-n-propylamine
142-96-1
Di-n-butyl ether
151-56-4
Aziridine
287-92-3
Cyclopentane
K009; K010; K026; U182
✓ ✓
U108
U092
✓
150
710
(S) 100
(S) 360
✓
100
525
✓
10
18
150
✓
✓
5
✓
(S) 25 F001; F002; F024; F025; K116; K019; K020; K073; K150; K151; U210; D039
(S) 10 U102
(S) 90
10
✓
✓
(2)
25
ε
✓
(S) 35
10
5
✓ 3
5 mg/m
✓
✓
✓
U113
(S) 25
3
(S) 100
5
✓
10
✓
✓
✓
6
✓
U112
400
1400
400
✓
25
100
15
✓ ✓
500
2000
400
✓ ✓
U110
✓
P054
see 29 CFR 1910.1012
0.5
✓
600
✓
✓
1278
Carlos M. NuÔez
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
298-07-7
Di-(2-ethylhexyl)phosphoric acid
302-01-2
Hydrazine
306-83-2
2,2-Dichloro-1,1,1-trifluoro F001 ethane (HCFC-123)
ε
502-56-7
Di-n-butyl ketone
✓
510-15-6
Chlorobenzilate
513-37-1
Methallyl chloride
540-59-0
1,2-Dichloroethylene
540-84-1
2,2,4-Trimethylpentane
540-88-5
tert-Butyl acetate
541-41-3
Ethyl chloroformate
541-85-5
Ethyl amyl ketone
542-75-6
1,3-Dichloropropene
F024; F025; U084
542-88-1
Bis(chloromethyl)ether
K017; P016
543-59-9
n-Amyl chloride
584-84-9
2,4-Toluene diisocyanate K027
591-78-6
Methyl n-butyl ketone
592-41-6
1-Hexene
592-76-7
1-Heptene
594-42-3
Perchloromethyl mercaptan
624-83-9
Methyl isocyanate
628-63-7
n-Amyl acetate
100
626-38-0
sec-Amyl acetate
125
629-14-1
Ethylene glycol diethyl ether
✓
629-76-5
Pentadecanol
✓
680-31-9
Hexamethylphosphoramide
✓
696-28-6
Phenyldichloroarsine
✓
763-29-1
2-Methyl-1-pentene
✓
822-06-0
Hexamethylene-1,6diisocyanate
872-05-9
1-Decene
✓
U133
(S) 1
(S) 1.3
0.01
✓
✓
U038
✓ ✓ ✓
✓
✓ F024; F025; K073
200
790
200
✓ ✓
200
950
200
✓
✓ ✓
25
130
25
✓
1
✓
✓
✓
✓
✓ (C) 0.02 (C) 0.14
100
410
0.005
✓
10
✓
30
✓
✓
✓
0.1
0.1
✓
0.02
✓
525
100
✓
650
125
✓
0.8
(S) 0.02 (S) 0.05
0.005
✓ ✓
✓
✓
✓
19 Regulations
1279
CAS No.
Chemical name
10
EPA Code
Permissible Expo13 sure Limits (PEL)
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
872-50-4
N-Methylpyrrolidinone
✓
1191-17-9
2,2-Dichloroethyl ether
✓
1300-71-6
Xylenol
✓
1330-20-7
Xylene (isomer mixtures)
1336-36-3
✓
✓
Polychlorinated biphenyls
✓
✓
1338-24-5
Naphthenic acids mixtures
✓
1634-04-4
Methyl tert-butyl ether
✓
1717-00-6
1,1-dichloro-1-fluoroethane F001 (HCFC-141b)
ε
✓ ✓
2837-89-0
2-chloro-1,1,1,2-tetrafluoro ethane (HCFC-124)
F001
ε
✓ ✓
4170-30-3
Crotonaldehyde
U053
✓
7664-39-3
Hydrofluoric acid
U134
✓
7785-26-4
Dipentene
8030-30-6
Naphtha: coal tar
8032-32-4
Petroleum naphtha or naphtha: VM & P
8052-41-3
Naphtha: Stoddard solvent
13360-63-9
N-Ethyl-n-butylamine
✓
13952-84-6
sec-Butylamine
✓
25154-52-3
Nonylphenol
✓
25265-71-8
Dipropylene glycol
✓
25322-69-4
Polypropylene glycol
✓
25340-17-4
Diethylbenzene
✓
25378-22-7
Dodecene
✓
26761-40-0
Diisodecyl phthalate
✓
26952-21-6
Isooctyl alcohol
27215-95-8
Nonene
✓
28473-21-3
Nonanol
✓
29063-28-3
Octanol
✓
34590-94-8
Dipropylene glycol methyl ether
U239
100
435
100
✓
K022
100
500
✓
400
2900
300
✓
100
✓
50
(S) 100
(S) 600
100
✓
✓
✓
✓
1280
Carlos M. NuÔez
CAS No.
Chemical name
Permissible Expo13 sure Limits (PEL)
10
EPA Code
ppmv
mg/m3
TLV14 ppmv
Environmental Effect
O T D G
53535-33-4
Heptanol
✓
64742-81-0
Kerosene
✓
74806-04-5
Carene
✓
(ε) Exempt VOC. Permissible Exposure Limits (PELs) are 8-hour time weighted averages (TWAs) unless a number is preceded with a (C) which denotes a ceiling limit. When entry is in the mg/m3 column only, the value is exact; otherwise, it is approximate. Threshold Limit Values (TLVs) are the time weighted average concentrations for a conventional 8-hour workday and a 40-hour workweek. (C) Indicates ceiling limit, or a concentration that should not be exceeded during any part of the working exposure. (S) Denotes skin designation as the basis for the value. O Classified as a volatile organic compound (VOC). VOC can react in the lower atmosphere to form ozone and other oxidants. VOC means any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates, and ammonium carbonate, which participates in atmospheric photochemical reactions. Some compounds are specifically exempted from this definition which is found in 40 C.F.R. 51.100(s). T Considered a hazardous air pollutant (HAP) and listed in Title III of the Clean Air Act Amendments of 1990. D A regulated stratospheric ozone layer depleter. G May impact climate change.
Other solvents, known as chlorofluorocarbons (CFCs), cause stratospheric ozone depletion. They gradually release chlorine and other halogens into the atmosphere, which are effective at destroying the ozone layer,15 our natural protection from damaging ultraviolet light (UV). They also have high global warming potentials (GWPs).16 These solvents are generally not considered significant precursors of ground level ozone formation. The production and use of many CFCs have been banned17,18 and new chemicals have been synthesized and are becoming available for various applications (see Table 19.3).19,20 Table 19.3. List of acceptable replacement chemicals for ozone-depleting solvents (ODS)20 Substitutes Alternative Technologies (e.g., powder, hot melt, thermoplastic plasma spray, radiation-cured, moisture-cured, chemical-cured, and reactive liquid)
CFC-113a C
E
M
P
c
c
c
d
d
d
d
HFC-123
E
M
P
c
c
c
d
d
d
d
c
HFC-225ca/cb
d
HFC-4310mee
d
High-solids formulations
C
a
HCFC-141bb C
E
M
P
c
c
c
c
Aqueous cleaners Benzotrifluorides
Methyl chloroform
c
c
c
d
d
c
d
d c
c
19 Regulations
1281
Substitutes
CFC-113a C
Methyl chloroform
E
M
P
Hydrofluorether (HFE)
c
c
Methoxynanofluorobutane, iso and normal
c
E
M
P
c
c
c
c
c
c
Methylene chloride
c
c
c
c
Monochlorotoluenes
d
d
d
d
No clean alternatives
C
d
c
Oxygenated solvents (alcohols, ketones, esters, ethers)
c
Perchloroethylene
c
c
M
P
c
c
c
c
c
c
c
c
c
c
c
c
d
d
d
C
c
c
c
c
c
c
e
e
e
e
Perfluoropolyethers
e
e
e
e
Plasma cleaning
c
c
c
c
Semi aqueous cleaners
c
c
c
c
c
c
Straight organic solvent cleaning
c
c
c
c
c
c
Supercritical fluids
c
c
c
c
c
c
c
c
c
c
c
c
Terpenes
c
Trans-1,2-dichloroethylene
d
c
c
c
Trichloroethylene
c
c
c
c
c
c
c
UV/ozone cleaning
c
Vanishing oils
d
c
Volatile methyl siloxanes Water-based formulations
HCFC-141bb E
Perfluorocarbons (C5F12, C6F12, C6F14, C7F16, C8F18, C5F11NO, C6F13NO, C7F15NO, and C8F16)
Petroleum hydrocarbons
a
c
c
c c
c
c
c
c
c
where: a b c d e C E M P
Class I ODS - substance with an ozone depletion potential of 0.2 or higher Class II ODS - substance with an ozone depletion potential of less than 0.2 Acceptable Acceptable, subject to use conditions Acceptable, subject to narrowed use limit Adhesive, coating, and ink applications Electronics cleaning Metals cleaning Precision cleaning
Solvents are used commercially and industrially for numerous applications. Some of those applications include polymer synthesis, coating formulations, cleaning (e.g., elec-
1282
Carlos M. NuÔez
tronic circuit board, fabrics, ink supply lines, application equipment, and degreasing), and chemical processes. More than 300 solvents are currently used in the U.S.21 Organic chemical solvents are also widely used as ingredients in household products. Paints, varnishes, and wax all contain organic solvents, as well as many cleaning, disinfecting, cosmetic, degreasing, and hobby products.21 These consumer products can release organic compounds while in use and even when stored. Their release indoors can cause health problems, especially considering that we spend around 90 percent of our time indoors.22 All of the environmental and health problems aforementioned have prompted a number of regulations specifically designed to promote and maintain a cleaner and sustainable environment as well as a healthier and safer workplace. This chapter discusses the environmental and safety laws and regulations and the regulatory process in the U.S. Some international regulatory perspectives will also be provided. The purpose of the chapter is to present some key elements of relevant laws and regulations affecting the use of solvents. Solvent users can use this information to understand the range and basic structure of laws and regulations that may affect solvent use. State and federal regulators can use this information to gain a better understanding of the full scope of requirements across all media. In sections 19.2 to 19.4, the discussion follows the single-medium-based regulatory approach used in the U.S. Recognizing, however, that environmental problems may cross medium lines, section 19.5 of this chapter provides a brief discussion of multimedia laws and regulations. Section 19.6 discusses laws and regulations to protect workers from exposure to hazardous chemicals. An overview of international environmental laws and regulations is given in section 19.7.
19.2 AIR LAWS AND REGULATIONS 19.2.1 CLEAN AIR ACT AMENDMENTS OF 19909 19.2.1.1 Background In the seventies, the Air Quality Act of 1967 was amended twice: first in 1970 and then in 1977. These amendments brought, among other things, the establishment of national ambient air quality standards (NAAQS), new source performance standards (NSPS) for major new or modified stationary sources, and the national emission standards for hazardous air pollutants (NESHAPS). Also in 1970, the decentralized environmental quality programs and oversight were integrated with the founding of EPA.23 EPA became the federal agency with primary environmental regulatory responsibility in the U.S with 10 Regional Offices located in major cities (see Figure 19.1 and Table 19.4). The Clean Air Act was amended once again in 1990, building on earlier editions while incorporating some significant changes. The old command and control policy changed to a more cooperative effort on the part of the federal government, state agencies, local municipalities, and industry. Similar approaches were implemented as early as 1979. Then, EPA tried to develop market-based concepts in an attempt to limit and control pollution within a given area, also known as a bubble, while permitting some level of flexibility for the affected facilities. Facilities achieving greater control than required, were allowed to sell their excess pollution reduction credit to other facilities within that area provided the overall pollution level was not exceeded.24 These attempts did not achieve the desired pollution reduction, which prompted revisions and amendments to the CAA.
19 Regulations
1283
Figure 19.1. EPA organizational structure.
Despite substantial revisions and amendments to the CAA, prior EPA, state, or local laws or regulations were not repealed. In fact, even today, many states still follow prior CAA amendments.25 Table 19.4. List of EPA’s regional offices EPA Region
States
Address
Phone Number
1
Connecticut, Maine, Massachu- John F. Kennedy Federal Bldg., One Consetts, New Hampshire, Rhode Is- gress Street, Suite 1100, Boston, MA land, Vermont 02114-2023
(617) 918-1111
2
New Jersey, New York, Puerto 290 Broadway, New York, NY 10007-1866 Rico, Virgin Islands
(212) 637-3000
3
Delaware, District of Columbia, Maryland, Pennsylvania, Virginia, West Virginia
1650 Arch Street, Philadelphia, PA 19103-2029
(215) 814-5000
4
Alabama, Florida, Georgia, Kentucky, Mississippi, North Carolina, South Carolina, Tennessee
61 Forsyth Street, SW, Atlanta, GA 30303-3104
(404) 562-9900
5
Illinois, Indiana, Michigan, Minnesota, Ohio, Wisconsin
77 West Jackson Boulevard, Chicago, IL 60604-3507
(312) 353-2000
1284
EPA Region
Carlos M. NuÔez
States
Address
Phone Number
6
Arkansas, Louisiana, New Mexico, Oklahoma, Texas
1445 Ross Avenue, 12th Floor, Suite 1200, Dallas, TX 75202-2733
(214) 665-6444
7
Iowa, Kansas, Missouri, Nebraska
726 Minnesota Avenue, Kansas City, KS 66101
(913) 551-7000
8
Colorado, Montana, North Dakota, South Dakota, Utah, Wyoming
999 18th Street, Suite 500, Denver, CO 80202-2466
(303) 312-6312
9
American Samoa, Arizona, California, Guam, Hawaii, Nevada
75 Hawthorne Street, San Francisco, CA 94105
(415) 744-1702
10
Alaska, Idaho, Oregon, Washington
1200 Sixth Avenue, Seattle, WA 98101
(206) 553-1200
The CAA of 1990 contains six titles and related provisions designed to “encourage” air pollution abatement and reduction. These provisions address several environmental pollution problems that affect us all, such as tropospheric ozone, hazardous pollution, mobile emissions, urban pollution, acid deposition, and stratospheric ozone depletion. Because the scope of this chapter is on solvents and the regulations that impact their use, only Titles I, III, V, and VI and their relevance to solvents will be discussed. 19.2.1.2 Title I - Provisions for Attainment and Maintenance of National Ambient Air Quality Standards Under this Title (Section 108), EPA has issued National Ambient Air Quality Standards (NAAQS) for six criteria pollutants - referred to as traditional pollutants in Canada and the European Union (EU):26 ground level ozone, nitrogen oxides, carbon monoxide, sulfur dioxide, particulate matter less than 10 µm in aerodynamic diameter (PM10), and lead. They are called criteria pollutants because EPA has developed health-based criteria or science-based guidelines as the basis for establishing permissible levels. The air quality criteria established by EPA must “accurately reflect the latest scientific knowledge useful in indicating the kind and extent of all identifiable effects on public health and welfare which may be expected from the presence of such pollutants in the ambient air.”9 These ambient air pollutants or NAAQS pollutants are generally present nationally and may be detected through the use of ambient monitoring detectors. Other air pollutants may be regulated as HAPs under Section 112 of Title III if they pose danger to human health but their effect is confined to a localized area.25 NAAQS include standards to allow for an adequate protection of health (primary) and standards to prevent environmental and property damage from anticipated adverse effects (secondary). In July 1997, EPA published its revision to the ozone NAAQS. The new ozone NAAQS limit is 0.08 ppm and is based on an 8-hour air measurement average. Areas that do not meet the established NAAQS or that affect the ambient air quality in a nearby area that does not meet NAAQS for the pollutant are known as non-attainment areas.9 In the U.S., an estimated 90 million people live in these areas. These areas may be further classified based on the severity of non-attainment and the availability and flexibility of the pollution control measures believed necessary to achieve attainment.9 A State Imple-
19 Regulations
1285
Figure 19.2. Designation and classification of ozone areas.
mentation Plan (SIP) must be prepared and submitted by each state with a non-attainment area. The SIP must contain strategies to achieve compliance within the state’s borders and “Air Quality Control Regions.” Once a non-attainment area meets the NAAQS for a given pollutant, it is then classified as a maintenance area. Ozone non-attainment areas are classified according to the severity of the pollution problem (see Figure 19.2). These areas also have to achieve VOC reduction as determined by the non-attainment classification. A 15 percent VOC reduction in non-attainment areas, classified as moderate and above, was required by 1996. Also, Section 182(c)(2)(B) requires that non-attainment areas classified as serious and above follow with a 3 percent VOC reduction per year.9 Each state with a non-attainment area selects reasonable control measures, known as Reasonably Available Control Technologies (RACTs), for major sources to achieve the required reduction. Specific information about RACTs can be found in the EPA published Control Technology Guideline (CTG) documents, which are designed to assist state agencies achieve VOC reduction and ozone NAAQS compliance. States are primarily responsible for meeting NAAQS following measures laid out in their SIPs. States work with stationary and mobile sources to ensure that criteria pollutant levels (ozone in the case of solvents) allow the states to meet NAAQS requirements. If a state is not doing an adequate job in improving and maintaining air quality through the activities identified in the SIP, EPA may step in and insist on more stringent measures. EPA promulgates Federal Implementation Plans (FIPs) to correct inadequacies in SIPs. A FIP includes enforceable emission limitations or other control measures and techniques or economic incentives so compliance with the NAAQS is achieved. In the case of new major stationary sources, the states try to stay or come into attainment through the New Source Review (NSR) program. The NSR program, sometimes called preconstruction permitting, addresses emissions from new major stationary sources;
1286
Carlos M. NuÔez
i.e., entire facilities or major modifications to existing facilities. Requirements depend on whether the facility is in a non-attainment area or is in a prevention of significant deterioration (PSD) area. Major sources for these areas must obtain a permit before any construction or modification is started. CAA defines a major VOC source as one that emits 9.1 Mg/year in an area classified as severe.26 Facilities in PSD areas are required to apply Best Available Control Technology (BACT) which takes into account economic impact. BACT is analyzed by regulators and applied to facilities on a case by case basis. BACT determination must result in emissions reduction as stringent as the federal New Source Performance Standard (NSPS) rule. Facilities located in non-attainment areas or that are increasing emissions in such areas must apply for a state permit prior to starting construction. The permit must codify that they are applying Lowest Achievable Emissions Rate (LAER). LAER is also established on a case by case basis. They also need to obtain equivalent offsetting emissions reductions (Offsets) from existing sources within the non-attainment area. Unlike BACT, a LAER determination does not consider economic, energy, or other cross-media impacts. Information about emission control technology, work practices, and reformulations that have been determined to be BACT and LAER for various types of sources is made available to the states through a central database called the RACT/BACT/LAER Clearinghouse. Most organic solvents in this book are classified as VOCs and are found in commercial and consumer products such as paints, coatings, containers, and packaging (see Table 19.2 for a list of these VOC solvents and exempt solvents). They contribute to the formation of ozone and are regulated under this Title, Section 183(e).9 Fuels and fuel additives are not the focus of Section 183(e). VOCs, nitrogen oxides, and oxygen undergo a photochemical reaction in the presence of sunlight resulting in the formation of tropospheric or ground level ozone, the primary component of smog. Heat also increases the reaction rates of these reactions. The contribution of some solvents to photochemical reaction may be considered negligible. As a result, some VOC solvents may be reclassified as exempt solvents; e.g., acetone. This reclassification begins with industry submitting a petition to EPA for a solvent or solvents to be reclassified as exempt. As of December 3, 1999, EPA had a list of 18 compounds for which petitions requesting VOC exempt status (see Table 19.5) had been received.27 Table 19.5. List of VOCs with requested VOC exempt status Submitting
Compound Chlorobromomethane
Organization
Date
ICF Kaiser
11/10/95
Enviro Tech International
05/10/96
Albemarle Corp.
11/18/97
Methyl bromide
Chemical Manufacturers Association
07/19/96
n-Alkanes (C12 - C18)
The Aluminum Association
11/27/96
The Printing Industries of America
12/20/96
Penzoil Products Company
12/20/96
Bromopropane
Technical white oils
Proposed date
19 Regulations
1287
Submitting
Compound
Organization
Date
t-Butyl acetate
ARCO Chemical Company (now Lyondell)
01/17/97
Benzotrifluoride
Occidental Chemical Company
03/11/97
E.I. du Pont de Nemours and Company
08/11/97
Texas Mid-Continent Oil & Gas Association
12/05/97
3M Corporation
10/08/97
Proposed date 9/30/99 64 FR 52731
Carbonyl sulfide trans-1,2-Dichloroethylene
Dibasic Esters Group, affiliated with the SynDimethyl succinate and dimethyl thetic Organic Chemical Manufacturers Assoglutarate ciation, Inc.
10/14/97
Carbon disulfide
Texas Mid-Continent Oil & Gas Association
12/05/97
BP Chemicals
01/21/98
GNI Chemicals Corporation
01/21/98
Toluene diisocyanate (TDI)
Chemical Manufacturers Association
01/22/98
1,1,1,2,3,3,3-heptafluoropropane (HFC-227)
Great Lakes Chemical Corporation
02/18/98
Acetonitrile
Methylene diphenyl diisocyanate Chemical Manufacturers Association (MDI)
08/19/98
1,1,1,2,2,3,3-heptafluoro-3methoxy-propane
3M Performance and Fluids Division
02/05/99
Propylene carbonate
Huntsman Corporation
07/27/99
Methyl pivalate
Exxon Chemical Company
11/22/99
Note: As of December 3, 1999, EPA had not published final actions on these chemicals.
In 1995, under Section 183(e), EPA generated a list of consumer and commercial products categories to be regulated which accounted for at least 80 percent of the VOC emissions from such products in ozone non-attainment areas.28 The list was divided into four groups, and each group is to be regulated every 2 years (starting in 1997) based on the best available controls (BACs). These four groups are listed in Table 19.6.29 The term control is broad, including not only end-of-pipe control technology, but also source reduction alternatives. The product categories were selected for regulation based on eight criteria developed to address five factors outlined in Section 183(e)(2): 1. utility 2. commercial demand 3. health or safety functions 4. content of highly reactive compounds 5. availability of alternatives 6. cost effectiveness of controls 7. magnitude of annual VOC emissions 8. regulatory efficiency and program considerations
1288
Carlos M. NuÔez
Table 19.6. Consumer and commercial products categories Group I (1997) Aerospace Coatings, Architectural Coatings, Autobody Refinishing Coatings, Consumer Products (24 categories), Shipbuilding and Repair Coatings, Wood Furniture Coatings Group II (1999) Flexible Packaging Printing Materials Group III (2001) Aerosol Spray Paints, Industrial Cleaning Solvents, Flat Wood Paneling Coating, Lithographic Printing Materials Group IV (2003) Auto and Light Duty Truck Assembly Coatings, Fiberglass Boat Manufacturing Materials, Large Appliance Coatings, Letterpress Printing Materials, Metal Furniture Coatings, Miscellaneous Industrial Adhesives, Miscellaneous Metal Products Coatings, Paper, Film, and Foil Coatings, Petroleum Drycleaning Solvents, Plastic Parts Coatings This table reflects the regulatory schedule revised on March 18, 1999 Federal Register (64 FR 13422).
19.2.1.3 Title III - Hazardous Air Pollutants Section 112 of Title III of the CAA of 1990 lists 188 HAP compounds or groups of compounds (the original list had 189, but caprolactam was removed from the list) which are believed to be carcinogens or can otherwise pose serious health effects. HAP solvents can cause a number of health problems such as cancer, kidney and liver damage, developmental problems in children, nervous system problems, respiratory irritation, birth defects, and miscarriages.7,8 It may take minutes or years for these health effects to manifest themselves. Approximately half (90 compounds) of the HAP compounds are solvents and are found in many of the source categories regulated under Title III. EPA has the authority to delete chemicals from or add new chemicals to this original list, but no criteria pollutant can be added, only its precursors. If a petition to add or delete a chemical is submitted to EPA, EPA has 18 months to respond with a written explanation of its denial or concurrence. Before the 1990 CAA, EPA listed only eight substances as HAPs (beryllium, mercury, vinyl chloride, asbestos, benzene, radionuclides, arsenic, and coke oven emissions) and issued standards for seven (coke oven emissions were not regulated before the 1990 CAA). After the enactment of the CAA of 1990, a substantially greater number (174) of source categories were identified to be regulated within a period of 10 years. In other words, EPA would promulgate regulations for 174 source categories in half the time that it took to promulgate regulations for only seven. This number of categories to be regulated has been reduced because of subcategorization or deletion of categories. Many of these categories are major stationary sources which emit HAP or toxic solvents to the atmosphere (see Table 19.2) and are scheduled to be regulated by the year 2000. These regulations are technology-based standards, known as Maximum Achievable Control Technology (MACT) standards. MACT standards are EPA’s regulatory means to establish emission limits for air toxics. MACT standards have already been promulgated for 39 source categories (see Table 19.7). EPA will not be developing MACT standards for seven of the originally listed major source categories. There are 66 source categories that remain to be regulated.
19 Regulations
1289
Table 19.7. Categories of sources of HAPs for which Maximum Achievable Control Technology (MACT) standards have already been developed30 Schedule
Source Categories Dry Cleaning
2-Year Hazardous Organic NESHAP
Compliance Date 09/23/96 F/G - 05/14/01; H - 05/12/99; New Sources - 05/12/98
Aerospace Industry
09/01/98
Asbestos
Delisted
Chromium Electroplating Coke Ovens
Deco - 0 1/25/96; Others -01/25/97 Not available
Commercial Sterilizers
12/06/98
Degreasing Organic Cleaners (Halogenated Solvent Cleaning)
12/02/97
Gasoline Distribution (Stage 1)
12/15/97
Hazardous Waste Combustion
06/19/01
Industrial Cooling Towers
03/08/95
Magnetic Tape Marine Vessel Loading Operations
w/o new control devices -12/15/96; w/ new control devices - 12/15/97 MACT - 09/19/99; RACT -09/19/98
Off-Site Waste Recovery Operations
07/01/00
Petroleum Refineries
08/18/98
Polymers & Resins I: Butyl Rubber, Epichlorohydrin Elastomers, Ethylene Propylene Rubber, Hypalon (TM) Production, Neoprene Production, Nitrile Butadiene Rubber, Polybutadiene Rubber, Polysulfide Rubber, Styrene-Butadiene Rubber & Latex
07/31/97
Polymers & Resins II: Epoxy Resins Production, Non-Nylon Polyamides Production
03/03/98
Polymers & Resins IV: Acrylontrile-Butadiene-Styrene, Methyl Methacrylate-Acrylonitrile, Methyl Methacrylate-Butadiene, Polystyrene, Styrene Acrylonitrile, Polyethylene Terephthalate
07/31/97
4-Year
Printing & Publishing
05/30/99
Secondary Lead Smelters
06/23/97
Shipbuilding & Ship Repair
12/16/96
Wood Furniture
11/21/97
Chromium Chemical Manufacturing
Delisted
7-Year Electro Arc Furnace: Stainless & Non-Stainless Steel
Delisted
1290
Schedule
7-Year
10-Year
Carlos M. NuÔez
Source Categories
Compliance Date
Ferroalloys Production
05/20/01
Flexible Polyurethane Foam Production
10/08/01
Generic MACT: Acetal Resins, Hydrogen Fluoride, Polycarbonates Production, Acrylic/Modacrylic Fibers
06/29/02
Mineral Wool Production
06/01/02
Nylon 66 Production
Delisted
Oil & Natural Gas Production
06/17/02
Pesticide Active Ingredient Production: 4-Chlor-2-Methyl Acid Production; 2,4 Salts & Esters Production; 4,6-dinitro-o-cresol Production; Butadiene Furfural Cotrimer; Captafol Production; Captan Production; Chloroneb Production; DacthalTM Production; Sodium Pentachlorophenate Production; TordonTM Acid Production
06/30/02
Pharmaceutical Production
09/21/01
Phosphoric Acid / Phosphate Fertilizers
06/10/02
Polyether Polyols Production
06/01/02
Portland Cement Manufacturing
Not available
Primary Aluminum Production
10/07/99
Primary Lead Smelting
06/04/02
Pulp & Paper (non-combust) - MACT I
04/15/01
Pulp & Paper (non-combust) - MACT III
04/16/01
Steel Pickling - HCl Process
06/22/01
Tetrahydrobenzaldehyde Manufacture (formerly known as Butadiene Dimers Production)
05/12/01
Wood Treatment MACT
Delisted
Wool Fiberglass Manufacturing
06/14/01
Cyanuric Chloride Production
Delisted
Lead Acid Battery Manufacturing
Delisted
Natural Gas Transmission & Storage
06/17/02
Under Section 112 of Title III of the CAA of 1990, EPA will regulate all major sources and area sources deemed appropriate. A major source is a stationary source located within a contiguous area under common control that emits or has the potential to emit ≥ 9.1 Mg/yr (10 tons/yr) of a single HAP or ≥23.7 Mg/yr (25 tons/yr) of a combination of HAPs. An area source is a non-major stationary source. Under Section 112, EPA is required to promulgate 40 categories by 1992, 25 percent of the listed categories by 1994, an additional 25 percent
19 Regulations
1291
by 1997, and all categories by the year 2000. A list of sources was developed by EPA in 199231 and modified 4 years later.32 Each MACT standard allows existing sources 3 years to comply. Additional years may be granted to complete the installation of controls, to dry and cover mining waste, and for coke ovens. A new source that began construction after the date a MACT standard was proposed will begin compliance at startup or by the date the final standard is issued, whichever is later. Under the early reduction program, additional time may be given to a facility to meet HAP reduction requirements if about 90 percent of the HAP emissions are reduced before a MACT standard is in effect. Once MACT standards are promulgated, the CAA requires EPA to conduct a risk assessment, if considered necessary for public health, to: • assess the level of risk or “residual risk” remaining to public health, • estimate the public health significance of the estimated residual risk, • identify the technological and commercially available methods and costs to reduce the residual risk, • estimate the actual health effects in the vicinity of sources, • identify any health studies and uncertainties in the available risk assessment methodologies, • determine any negative health and environmental impacts resulting from the reduction of the residual risk, and • produce recommendations leading to legislation for the residual risk. This may result in even more stringent regulations for solvent users. Figure 19.3 outlines basic steps involved in the development of a MACT regulatory standard.28
Figure 19.3. Steps for the development of MACT regulations.
1292
Carlos M. NuÔez
The CAA provides for potential delays in the promulgation of MACTs for major HAP sources; i.e., Section 112(j). This is known also as the MACT Hammer.25 This provision requires major HAP sources to apply for a permit 18 months after the MACT standard scheduled promulgation date. The State must then make a case-by-case determination of MACT for the source. If the delayed MACT standard is later promulgated, the permit will be revised to reflect any new emission limitation requirements and the affected source may be given up to 8 years to comply.9 Air toxics have also been regulated at the state level under various air toxic state programs. In the U.S., the State of California has one of the more stringent and complex air toxics programs. California leads with air toxic programs such as Assembly Bill 1807 (AB 1807) that targets emissions that pose greatest risk to public health and Senate Bill 1731 (SB 1731) that regulates air toxics through facility-based controls rather than risk-based controls.33 19.2.1.4 Title V - Permits Title V was added to the CAA, in the 1990 CAA Amendments, to establish a permit program, federally mandated, to be implemented by the states. Under this Title, EPA has established 10 minimum “elements” for an operating permit program to be administered by the states. These elements serve as basic guidelines to be used when establishing an operating permit program. They include requirements for applications, monitoring and reporting, and annual fees to the owner or operator of sources subject to a permit. They also include the following elements to ensure the effectiveness of an operating permit program: • adequate personnel, funding, and permitting authority to administer the program; • procedures to expeditiously evaluate permit applications, prevent unreasonable delay by the permitting authority, and allow for revisions and needed changes to existing permits; and • provisions to allow permitted facilities to change without requiring permit revision.9 Each state is required to develop and submit a permit program. By 1997, EPA had approved nearly all states’ permit programs, and some sources have already applied for Title V permits.25 The idea of an operating permit is to consolidate all CAA requirements for a source in one regulatory vehicle. This includes all applicable SIP air emission limitations, monitoring, and reporting requirements along with CAA regulations (HAP regulations, NSPS, etc.).25 19.2.1.5 Title VI - Stratospheric Ozone Protection Title VI of the CAA deals with chemicals, CFCs, and other ozone depleting substances (ODS) that can cause deterioration of the stratospheric ozone layer. Some of these chemicals are used as solvents in cleaning operations (cleaning of metals, electronics, and precision equipment), coatings, adhesives, and inks. Their production and use have been banned both domestically and internationally34 (except for a few countries). An unprecedented international effort by 20 countries and the Commission of the European Communities resulted in the 1987 Montreal Protocol. The Montreal Protocol became effective on January 1, 1989. In the U.S., the CAA Amendments of 1990 were the domestic response to such a critical environmental issue of global proportions and provided EPA with the regulatory agenda. ODS solvents are required to be replaced with non-depleting chemicals. Controlling these substances is not an alternative. ODS solvents are listed in Table 19.2. Substitute sol-
19 Regulations
1293
vents deemed acceptable have been identified for some of these ODS solvents under EPA’s Significant New Alternatives Policy (SNAP), but other health concerns have limited their use. Some exemptions to the use of phase out ODS exist for essential uses as long as they are consistent with the Montreal Protocol (to date, none has been authorized by EPA).25 The SNAP program was established to control the commercialization of ODS substitutes and to ensure that such substitutes do not pose greater harm to the environment than the original chemicals. Through the SNAP program, EPA identifies, classifies, restricts, or prohibits the use of ODS substitutes.
19.3 WATER LAWS AND REGULATIONS 19.3.1 CLEAN WATER ACT 19.3.1.1 Background In 1972, Congress enacted the Federal Water Pollution and Control Act (FWPCA), also known as the Clean Water Act (CWA).35 The CWA is the primary federal legislation that regulates the quality of streams, lakes, estuaries, and coastal waters in the U.S. The CWA has been amended three times: 1977, 1982, and 1987.36,37,38,39 The CWA is divided into six titles. Its purpose is to restore and maintain the chemical, physical, and biological integrity of the Nation’s waters by eliminating the discharge of pollutants into navigable waters or municipal sewer systems.39 Waters can become contaminated by pollutants discharged from point sources, non-point sources, and storm water and by pollutants from the degradation of wetlands. Point source discharges originate from industrial and sewage treatment plants while non-point discharges are generated from diffuse sources such as sediment runoff from construction sites, leaks from faulty septic systems, and fertilizer and herbicide runoff from agricultural and residential activities. Point sources are “discernable, confined, and discrete conveyances.”38 Examples of a point source include pipes, ditches, channels, tunnels, conduits, discrete fissures, or containers. These sources are regulated and control through the National Pollutant Discharge Elimination System (NPDES) permit program. In 1976, EPA’s focus changed from control of conventional pollutants (biological oxygen demand and suspended solids) to control of toxic pollutants. This shift in EPA’s focus was the result of a lawsuit in 1976 which resulted in the Flannery Decree and in the 1977 amendments.36,40 These amendments established a permit program for point source discharges to focus on 21 major industrial categories and 65 priority pollutants (see Table 19.2 for list of priority pollutant solvents). Today’s list of priority pollutants has been expanded to include 129 toxic pollutants from 34 industrial categories. In the 1987 amendments of the CWA, Congress incorporated Section 319 to establish a national program to control non-point source discharges which provided a regulatory schedule for storm water. These amendments also established a revolving loan fund for construction of sewage treatment plants and provided EPA with enhanced enforcement tools.38 19.3.1.2 Effluent Limitations Under Title III of the CWA, effluent limitations provide the control conditions for a facility’s wastewater discharge under an NPDES permit which regulates the quantity and rate of discharges to navigable waters. Guidelines, called the effluent limitations guidelines, have been developed to establish technology-based limits for all types of industrial discharges. Over 50 industry-specific technology-based effluent guidelines have already been promulgated.40 There are five technology-based treatment standards currently established by the
1294
Carlos M. NuÔez
CWA to establish the minimum level of control to be required by a permit. They are the Best Practicable Control Technology (BPT), Best Available Technology Economically Achievable (BAT), Best Conventional Pollutant Control Technology (BCT), Best Available Demonstrated Technology (BDT), and Best Management Practices (BMPs). Industries that do not discharge directly into surface waters but into publicly operated treatment works (POTWs) are subject to pretreatment standards under Section 307(b) of the CWA.38 19.3.1.3 Permit Program The CWA, under Section 301 (effluent limitations), prohibits the discharge of any pollutant unless the source obtains a permit and ensures compliance. The NPDES permit program, pursuant to Section 402 of the CWA, requires that commercial and industrial facilities and POTWs apply for permits issued by EPA or a designated state. A permit regulates the amount, concentration, and rate of discharge of each regulated pollutant. According to the CWA, a pollutant includes dredged soil, solid waste, incinerator residue, sewage, garbage, sewage sludge, munitions, chemical wastes, biological and radioactive materials, wrecked or discarded equipment, rock, sand, cellar dirt, and industrial, municipal, and agricultural waste.39 These pollutants are further classified in three categories: conventional, toxic, and non-conventional. Virtually, any material and characteristics such as toxicity or acidity is considered a pollutant.40 Solvents fall under the category of toxic pollutants. An NPDES permit can be issued by EPA or by a state that has received EPA permitting approval. As of 1996, 40 states and territories had received EPA’s approval.40 States and territories that are not authorized to administer the NPDES program are Alaska, Arizona, District of Columbia, Idaho, Louisiana, Maine, Massachusetts, New Hampshire, New Mexico, the Pacific Territories, Puerto Rico, and Texas. For these states and territories, the 10 EPA Regional Offices will issue and administer the NPDES permits. The permit process is elaborate and requires extensive information and numerous steps. First a number of forms need to be completed to provide information about the facility, its operation, and the nature of the discharges. This is followed by a period of discussion between EPA and the discharging facility, and the draft permit is announced for comments from the public for 30 days. Once the comments are reviewed, responded to, and implemented, then the final NPDES permit is issued. 19.3.2 SAFE DRINKING WATER ACT 19.3.2.1 Background Enacted in 1974, the Safe Drinking Water Act (SDWA) was established to ensure safe drinking water in public water systems. It required that EPA identify substances in drinking water which could adversely affect public health.41 SDWA safeguards drinking water with two standards. Primary drinking water standards with a maximum contaminant level (MCL) designed to protect human health and secondary drinking water standards to protect public welfare. The secondary standards involve physical characteristics of drinking water such as color, taste, and smell. A total of 83 contaminants are regulated under the SDWA: 49 volatile and synthetic organic chemicals, 23 inorganic chemicals, 6 microbiological contaminants, and 5 radiological contaminants. In 1986, the SDWA was amended to greatly increase the responsibilities of EPA and state agencies. Under the 1986 amendments, EPA is directed to schedule the promulgation of primary public drinking water regulations, impose civil and criminal penalties for tampering with public water systems, and enforce more stringent standards.2 Also, under the
19 Regulations
1295
1986 amendments, the wellhead protection program was established to focus on potential contamination of surface and subsurface surrounding a well.42 In 1988, Congress passed the Lead Contamination Control Act (LCCA) as an additional amendment to the SDWA to protect the public against the contamination of drinking water by lead. In 1996, the SDWA was amended again to place emphasis on sound science and a risk-based standard approach, small water supply systems, and water infrastructure assistance through a revolving loan fund.43 19.3.2.2 National Primary Drinking Water Regulations National Primary Drinking Water Regulations (NPDWRs) are enforceable, unlike secondary regulations that are only considered advisable, and established by EPA for contaminants that may cause adverse health effects. The promulgation of the NPDWRs begins with the publication of the advanced notice of the proposed rule. This notice presents the current scientific and technical knowledge about the contaminant and the approach EPA is taking toward its regulation. During this period, EPA seeks comments and any additional information through public meetings, workshops, etc. Comments are then reviewed, and a proposed rule is announced in the Federal Register which includes MCLs and non-enforceable health goals called maximum contaminant level goals (MCLGs). After additional public comments, the final regulation is announced and promulgated. The schedule for this regulatory development is contained in the SDWA amendments of 1986. The first 9 contaminants were to be regulated by EPA within 1 year of enactment, followed by another 40 contaminants within 2 years of enactment, and 34 contaminants within 3 years of enactment for a total of 83 by 1989. The SDWA of 1986 also required EPA to regulate 25 additional contaminants every 3 years. EPA regulated these 83 contaminants except for arsenic, sulfate, and radionuclides. These three are addressed under the 1996 amendments. With the 1996 amendments of the SDWA, beginning in 2001, EPA is required to make regulatory decisions for a minimum of five listed contaminants every 5 years. Along with the regulatory development, EPA prepares criteria documents which provide technical support for the final rules. These criteria documents identify fundamental information regarding health effects of contaminants in drinking water. Health-effects-related data such as physical and chemical properties, toxicokinetics, human exposure, health effects in animals and humans, mechanisms of toxicity, and quantification of toxicological effects are evaluated and serve as supporting documentation for the MCLGs.44
19.4 LAND LAWS & REGULATIONS 19.4.1 RESOURCE CONSERVATION AND RECOVERY ACT (RCRA)10 19.4.1.1 Background The law that regulates solid and hazardous waste management is called RCRA. It was passed as a law in 1976 as an amendment to the Solid Waste Disposal Act (SWDA), and amended in 1980 and 1984, to regulate solid hazardous waste from its creation all the way to its disposal. The 1984 amendments are called the Hazardous and Solid Waste Amendments (HSWAs). These amendments made RCRA a more stringent law giving EPA more increased authority to enforce tighter hazardous waste management standards. They also provided for EPA to revisit problems associated with underground storage tanks. More specifically, Subtitle C regulates hazardous waste while Section D regulates non-hazardous waste such as municipal solid waste and sewage sludge.45 RCRA requirements were also
1296
Carlos M. NuÔez
amended to include small businesses that manage specified amounts of hazardous waste and show an increase in the number of hazardous wastes.46 19.4.1.2 RCRA, Subtitle C - Hazardous Waste Subtitle C is RCRA’s regulatory arm to deal with management of hazardous solid waste. A waste is considered hazardous if it appears on EPA’s list of hazardous wastes or if hazardous characteristics can be identified. However, the person generating the solid waste is responsible for determining if the generated solid waste is hazardous. Over 500 solid wastes have been identified and listed by EPA as hazardous. Once a waste appears on the EPA hazardous list (wastes classified as F, K, P, U, or so-called “listed” wastes), such a waste is always considered hazardous regardless of its chemical composition. However, EPA is currently working on developing “exit criteria” for certain chemicals in listed wastes that would allow these wastes to be managed as non-hazardous solid wastes if the chemical concentrations are below specified levels. In Part 26, EPA classifies hazardous wastes by the following codes10 (see Table 19.2 for EPA’s hazardous codes for solvents): F wastes from non-specific sources K wastes from specific sources P discarded acutely hazardous commercial chemical products U discarded commercial chemical products D wastes for which the above codes do not apply − considered hazardous only if they exhibit ignitability (I), corrosivity (C), reactivity (R), and/or toxicity (E/H/T). A solid waste considered hazardous will always maintain its listed classification even if it is found mixed with any other waste or a residue derived from a listed hazardous waste.10 In addition, land disposal of a hazardous waste is prohibited by RCRA unless it is demonstrated that hazardous constituents will not migrate from the disposal unit, causing harm to the environment and public health (40 CFR 268).10 Hazardous solid waste generators are classified as small quantity generators (SQGs) (i.e., 100 to 1000 kg/month) or large quantity generators (LQGs) (i.e., ≥ 1000 kg/month or > 1 kg of an acutely toxic hazardous waste). These waste quantities apply to each month, not the average over the year. For example, if a facility generates 5 kg in each of the first 11 months and 1200 kg in the last month, such a facility would be considered a large quantity generator (LQG) even though its average is less than 1000 kg/month. Generators can be further classified as conditionally exempt small quantity generators (CESQGs) if they produce < 100 kg/month of hazardous waste and do not accumulate more than 1000 kg of hazardous waste. CESQGs are exempt from all RCRA notification, reporting, and manifesting requirements. Waste from CESQGs must be sent to properly permitted, licensed, or registered treatment, storage, and disposal (TSD) facilities, recycling/reclamation facilities, or facilities authorized to manage industrial or municipal solid waste.2 RCRA regulations also provide for states to have primary responsibility for managing and implementing the RCRA hazardous waste program (40 CFR 271). Currently, most states and territories have been granted this authority. States can receive an interim authorization (40 CFR 271, Subpart B) as a temporary vehicle while developing a program for final authorization, which is fully equivalent with the federal program. Hence, for a state to obtain final authorization, it must also be consistent with and no less stringent than the Federal program.
19 Regulations
1297
Figure 19.4. Definition of a solid waste (Figure 1 - 40 CFR 260, Appendix I).
RCRA also addresses the transportation of hazardous waste (40 CFR 263). Any transporter of hazardous waste must also comply with the Department of Transportation (DOT) rules under the Hazardous Materials Transportation Act (HMTA) regarding proper labeling, packaging, handling, and placarding.47 Finally, a facility generating waste can determine if its waste is regulated under Subtitles C or D by following the steps outlined in the Code of Federal Regulations, Title 40 Part 260, Appendix I (see Figures 19.4, 19.5, and 19.6). Other provisions exist if the solid waste will be incinerated.
19.5 MULTIMEDIA LAWS AND REGULATIONS 19.5.1 POLLUTION PREVENTION ACT OF 19903 19.5.1.1 Background Before 1990, most regulations in the U.S. were issued based on the use of conventional pollution management practices usually considering waste management and end-of-pipe control technologies. Only a handful of government activities were considering reduction of pollution at the source. In the area of waste minimization, recycling approaches were employed in addition to add-on control technology to achieve pollution reduction. However, reducing pollution at the source refers to the top of the hierarchy. Although add-on control
1298
Carlos M. NuÔez
technology has its place in many applications, new approaches need to be considered to obtain emissions reduction required by new regulations. In 1990, Congress passed the Pollution Prevention (P2) Act, recognizing that to attain greater levels of pollution reduction, reduction or prevention of pollution at the source “through cost-effective changes in production, operation, and raw materials use” needed to be encouraged and implemented.3 A fundamental belief expressed in the P2 Act, is that these changes would not only help the environment but would benefit industry’s bottom line through the reduction of raw material, pollution control, and liability costs.3 19.5.1.2 Source Reduction Figure 19.5. Definition of a hazardous waste (Figure 2 - 40 CFR 260, Provisions Appendix I). The P2 Act is an attempt by the U.S. to establish a framework for integrated multimedia environmental initiatives. The P2 Act provides for a multimedia, integrated, and cost-effective approach to solving environmental problems while encouraging sustainable development. P2 or source reduction includes any practice that reduces or eliminates the creation of pollutants through increased efficiency in the use of raw materials, energy, water, or other resources, or protection of natural resources by conservation. Pollution may be reduced by modifying equipment or technology, modifying processes, reformulating products, substituting raw materials, and improving housekeeping, maintenance, training, or inventory control.3 The P2 Act encourages the reduction or elimination of wastes of all types and requires that facilities reporting to the Toxic Release Inventory (TRI) provide documentation of their P2/waste minimization or reuse efforts for TRI-reportable chemicals. The P2 Act required EPA to establish an Office of Pollution Prevention with the authority of promoting source reduction through a multimedia perspective. The Office of Pollution Prevention already existed under The Office of Policy, Planning, and Evaluation (OPPE). It was established by EPA 2 years prior the enactment of the P2 Act. In response to the P2 Act, in 1991, this office developed a source reduction strategy. The P2 Act also includes provisions aimed at improving the collection and public access of data. Under the Emergency Planning and Community Right-to-Know Act of 1986, industrial facilities are required to report on their annual releases of toxic chemicals to the
19 Regulations
Figure 19.6. Special provisions for certain hazardous wastes (Figure 3 - 40 CFR 260, Appendix I).
1299
environment. In response to the P2 Act, the report must include information on their efforts on source reduction and recycling. In response to the P2 Act and to help promote source reduction, EPA established various prevention programs and initiatives: • 33/50 Program - A very successful EPA experimental program based on voluntary industry participation. Companies that reported to TRI any of the 17 target chemicals were encouraged to participate voluntarily to meet a 33 and 50 percent reduction by 1992 and 1995, respectively. Over 1000 companies participated in this program achieving a 55 percent reduction, which translates to an overall reduction of 340 x 106 kg (750 x 106 lb) of toxic chemical releases and transfers.4 More information about this program can be obtained by contacting the EPA’s Office of Pollution Prevention and Toxics website at
http://www.epa.gov/opptintr/3350/.
• Common Sense Initiative (CSI) − This initiative consisted of six pilot sectors: auto-
•
•
mobile manufacturing, computer and electronics, iron and steel, metal finishing, printing, and petroleum refining. The CSI idea was to focus on environmental management by industrial sector rather than by environmental medium.48 More information can be found at http://www.epa.gov/commonsense. Source Reduction Review Project (SRRP)49 − This was a pilot program designed to ensure that source reduction measures and multimedia issues were considered during the development of air, water, and hazardous waste regulations affecting 17 industrial categories. The SRRP was established to provide a model for the regulatory development process throughout EPA. Some progress has been made, but most regulations are still developed primarily based on a single medium and, while source reduction is encouraged, add-on control technologies are the basis of the regulations. Project XL (eXcellence and Leadership) − This is another pilot program recently established by EPA to promote the use of innovative approaches rather than conventional regulatory and policy strategies to achieve more cost-effective
1300
Carlos M. NuÔez
environmental results. The XL project has three key elements: superior environmental performance (SEP) to anticipate superior environmental performance, regulatory flexibility to encourage participation and maximize success, and stakeholder involvement to assist in development and implementation of the project.50 Visit http://www.yosemite.epa.gov/xl/xl_home.nsf/all/homepage for more information. Other examples of voluntary programs and initiatives include: AgStar, Climate Wise, Coalbed Methane Outreach, Design for the Environment, Energy Star Builidings, Energy Star Residential, Energy Star Office Equipment, Energy Star Transformer, Environmental Accounting, Environmental Leadership Program, Green Chemistry, Green Lights, Indoor Environments, P2 Information Clearinghouse, P2 Grant Program, Transportation Partners, U.S. Initiative on Joint Implementation, WAVE, Waste Minimization National Plan, and WasteWi$e. 19.5.2 TOXIC SUBSTANCES CONTROL ACT 19.5.2.1 Background Also referred to by some as the sleeping giant of environmental regulations,51 the Toxic Substances Control Act (TSCA) was enacted in 1976 to regulate chemicals in commerce that may cause adverse environmental and health effects.52 TSCA requires testing of manufactured substances to determine the character of their effect and regulates the manufacture, distribution, use, and disposal of new and existing substances. In 1986, TSCA was amended to include the Asbestos Hazardous Emergency Response Act (AHERA). The 1990 amendments provided for coverage of all public and commercial buildings. TSCA does not regulate food and food products, tobacco or tobacco products, and pesticide manufacturing, processing, or distribution in commerce. These areas are regulated under the Federal Food, Drug, and Cosmetic Act (FFDCA) and the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA). TSCA’s regulatory responsibility is not delegated to states, as may be the case with the other regulations previously mentioned. Since the focus of this book is on solvents, only some key provisions of Title I will be briefly discussed. 19.5.2.2 Controlling Toxic Substances Under Title I of TSCA (Section 4), manufacturers, importers, and processors of chemical substances and mixtures may be required by EPA to obtain health and environmental data on their health and environmental effects if: their manufacture, processing, use, distribution in commerce, and disposal, or any combination of such activities, poses an unreasonable risk to human health or the environment; • they are or will be produced and enter the environment in substantial quantities; • human exposure is significant; and • data and experience to determine the chemicals’ potential impact on human health and the environment are insufficient. TSCA (Section 5) also provides EPA with the authority to regulate and control the introduction of new chemicals either through manufacture or import, or the processing of an existing chemical for a significant new use. Such manufacturers and importers must file a Premanufacture Notification (PMN) 90 days before producing or importing the chemical. Once a PMN is filed, EPA assesses the information and determines if the chemical poses an unreasonable risk of injury to health or the environment. An additional 90 days may be re-
19 Regulations
1301
quired by EPA to complete the revision of the PMN. If during the first 90 days, EPA does not respond to the PMN, the manufacture or import of the chemical can begin. EPA has the authority to limit, prohibit, or ban the production of chemicals based on insufficient evidence to conduct a reasonable evaluation of the risk or because the chemical poses an unreasonable risk to human health and the environment (Section 6). Examples of prohibited substances by TSCA include asbestos, halogenated chlorofluoroalkanes, and polychlorinated biphenyls (PCBs). TSCA provides specific guidance and requirements for data gathering and reporting. This includes information such as production volume, plant size, amount lost to the environment during production or import, quantity of releases, and worker exposure information pertaining to controlled or uncontrolled releases. Companies are also required to maintain records of allegations of significant health or environmental effects (Section 8). Since 1979, EPA has maintained and published a list of chemical substances manufactured, imported, or processed for commercial purposes. This list, known as the TSCA Inventory, currently has almost 80,000 chemicals.
19.6 OCCUPATIONAL LAWS AND REGULATIONS 19.6.1 OCCUPATIONAL SAFETY AND HEALTH ACT53 19.6.1.1 Background In April 1971, the Occupational Safety and Health (OSH) Act of 1970 was enacted and resulted in the creation of the Occupational Safety and Health Administration (OSHA) and the National Institute for Occupational Safety and Health (NIOSH). The purpose of this Act is to ensure that workers are provided with workplaces free from recognized safety, health, and environmental hazards. OSHA is part of the Department of Labor (DOL) and responsible for developing and enforcing rules to ensure workplace safety and health. NIOSH is a research agency, part of the Centers for Disease Control and Prevention (CDC), under the Department of Health and Human Services. NIOSH is responsible for developing and establishing new and improved OSH standards and identify potential hazards of new work technologies and practices. Since 1970, OSHA has issued more than 500 rules dealing with all aspects of worker safety and protection in the workplace. Recent amendments to the 1970 Act require employers to notify employees of potential workplace health hazards, including chemicals. OSHA regulates exposure to about 400 hazardous and toxic substances in the workplace that can cause harm.54 Under Section 18 of the OSH Act, states are allowed to assume the responsibility for developing and enforcing their own safety and health programs.53 To date, 25 states have OSHA-approved state plans.55 Under the OSH Act, employers are required to provide employees with a workplace free from “recognized hazards.” Employers are also required to maintain accurate records of employees’ exposure to potentially toxic materials required to be monitored or measured, conduct periodic inspections, and notify employees if they are exposed to toxic materials at higher levels than those prescribed by existing health and safety regulations. 19.6.1.2 Air Contaminants Exposure Limits12-14 Information on exposure limits to solvent emissions in the workplace is provided in Table 19.2. These permissible exposure limit (PEL) values were obtained from OSHA’s Tables Z-1, Z-2, and Z-3.13 They are the maximum allowed PELs legally enforceable by OSHA.
1302
Carlos M. NuÔez
For comparison purposes, Table 19.2 provides exposure limit values generated by the American Conference of Governmental Industrial Hygienists (ACGIH) which for many of the listed chemicals are the only values currently available.14 Usually, the ACGIH values are more stringent than the OSHA values. Both ACGIH and NIOSH have proposed more stringent PELs for over 35 and 25 solvents, respectively.14,56 Although Table Z-1 has not been formally updated since 1987, new toxicological data on specific chemicals could be used by OSHA to produce directives intended to amend old exposure limit requirements or to add new chemicals to the regulated list. An OSHA PEL is based on an 8-hour, time weighted average (TWA) concentration. These were initially based on ACGHI TLV/TWA values in place since 1971, and many have not been revised since.57 The lower the value, the more toxic the solvent. PEL values in Table 19.2 are given in parts per million by volume (ppmv) of contaminated air volume and the equivalent milligrams of solvent in a cubic meter of air (mg/m3). These values are focused only on inhalation exposure, which has been strongly criticized.54 Although this is the most common exposure route to solvent emissions in the workplace, other exposure risks exist such as skin absorption and eye contamination. It is then very important to observe toxicological data compiled in the Material Safety and Data Sheets (MSDSs). Chemical manufacturers and importers are responsible to produce or obtain an MSDS for each hazardous chemical they produce or import, distributors must provide their customers with a copy of MSDSs, and employers are responsible to maintain a copy of MSDSs for each hazardous chemical used in their facility and make them available to their employees. There are various sources of additional helpful information on the safe use of solvents in the workplace. The Hazard Evaluation System and Information Service (HESIS) of California’s Department of Health Services has a 50-page document (Using Solvent Safely) that describes methods of using solvents in the workplace. This document includes information on solvents that pose reproductive problems to both women and men. For more information, visit their website on http://www.ohb.org/solvmenu.htm.58 19.6.1.3 Hazard Communication Standard The Hazard Communication Standard (HCS) was established to provide workers the “right-to-know” of potential hazards associated with their jobs. HCS is a proactive measure to disseminate information to workers and employers about the health hazards pertaining to the chemicals they handle and the protection required. Employers are required to prepare and implement a hazard communication (HAZCOM) program, ensure that all containers have appropriate labels, provide employees easy access to MSDSs, and conduct training programs. The HAZCOM program is the written plan of action that describes the standard implementation strategy for a given facility.
19.7 INTERNATIONAL PERSPECTIVE Similar environmental efforts have occurred in other countries as they have made the environment and human health a priority. Many countries around the world have developed policies, laws, and regulations in an effort to improve and maintain a cleaner environment and human health while providing for a sustainable future. This section provides a view of such efforts.
19 Regulations
1303
19.7.1 CANADA59 In Canada, Environment Canada, which was created in 1971, is the organization responsible for environmental regulation and protection. Environment Canada is divided into five Regional Offices: Ontario, Atlantic, Prairie and Northern Region, Quebec, and Pacific and Yukon. Its environmental services are administered through the Canadian Environment Protection Act (CEPA) which includes the Air Quality Act, the Canadian Water Act, the Ocean Dumping Act, the Environment Contaminant Act, and the Department of the Environment Act. CEPA was recently amended (September 1999), placing more emphasis on pollution prevention and toxic substances. It also provides an increased enforcement authority and resources to carry out necessary research and development activities. In 1978, Canada signed the Great Lakes Water Quality Agreement with the U.S. which identified specific objectives for minimum levels of water quality for more than 35 substances and compounds.60 Although this is an international agreement, it has certainly impacted and, to some extent, shaped Canada’s domestic environmental laws. As a result of these statutes, the federal government establishes objectives, guidelines, and emission standards for five national air quality pollutants (carbon monoxide, nitrogen oxides, ozone, sulfur dioxide, and suspended particulate matter). In the case of stationary sources, the regulatory responsibility usually falls under the jurisdiction of 10 provinces. The jurisdiction of water quality management is divided between the provinces and the federal government. The provinces’ primary responsibility is to regulate fresh water resources, while the federal government’s jurisdiction is over seacoast and inland fisheries, navigation and fisheries, and trade and commerce. The 10 provinces are also primarily responsible for regulating the management of household, non-hazardous, and hazardous solid waste. 19.7.2 EUROPEAN UNION In Europe, the European Community (EC) was formed by the union of three organizations: the European Coal and Steel Community (ECSC), established in 1951; the European Economic Community (EEC), established by the Treaty of Rome in 1957; and the European Atomic Energy Community (EURATOM), established in 1957.60,61 In 1967, with the Merger Treaty, these organizations merged to form the EC. After the Maastricht Treaty in 1992, the EC became the European Union (EU) which consists of the European Commission, the European Council, the European Parliament, and the European Court of Justice.61 Environmental policy was structured as part of the formation of the integrated European states. This effort was extended and refined later with the Maastricht Treaty of 1992. Over the years, however, EC has taken steps to provide a better environment and health for its member states and citizens. Some of these steps have been through the implementation of comprehensive Environmental Action Programmes (EAPs) and specific legislative measures or directives.62,63 The fifth EAP started in 1993 with an emphasis on sustainable environmental development and a variety of environmental issues such as climate change, acidification and air pollution, depletion of natural resources and biodiversity, depletion and pollution of water resources, deterioration of the urban environment and coastal zones, and waste.60 Also in 1993, the European Environment Agency (EEA) was established to collect, organize, and disseminate technical, scientific, and economic information pertaining to the quality of the environment in Europe. Air legislation is targeting emissions from industrial operations, greenhouse gases, lead, motor vehicle emissions, nitrogen oxides, ODS, sulfur dioxide, and suspended particulate matter. Water legislation in the EU is divided into effect- and source-oriented direc-
1304
Carlos M. NuÔez
tives. Four effect-oriented directives provide objectives for water with specific end uses: Bathing Water, Drinking Water, Fish Water and Shellfish Water, and Fresh Water Quality Information Exchange. Seven source-oriented directives focus on the elimination or reduction of pollution at the source: Asbestos, Dangerous Substances, Groundwater, Nitrate, Titanium Dioxide, Urban Wastewater, and Water Pollution Information Exchange.64 Solid and hazardous waste has received great attention in recent years, and directives have been developed to address their definition, classification, generation, management, and transport across frontiers. In 1993, EU implemented the European Waste Catalogue which defines 15 categories of waste and a residual category intended to capture any materials, substances, or products not included in those categories. There are 27 substances defined as toxic and hazardous waste for which specific information had to be provided during transport: nature, composition, quantity of waste, and sender’s and receiver’s name.65
19.8 TOOLS AND RESOURCES FOR SOLVENTS Various software tools have been developed to identify environmentally benign alternative solvents or equipment modifications to reduce the amount of toxic and volatile solvents. Some tools have attempted to consider life cycle impact in their selection methodologies, but these are the exceptions rather than the rule. This section provides a brief description of some of the tools and resources available to assist in solvent and equipment replacement. In addition, a hotline listing is provided (Table 19.8). Some of these tools are the result of years of research and development by the National Risk Management Research Laboratory of EPA’s Office of Research and Development. This laboratory has been researching solvent abatement,66 replacement, reduction,67 complete elimination for cleaning and coating operations, and developing tools68 to assess the overall environmental impact of alternative approaches. Table 19.8. List of hotlines in the U.S.69 Name
Description
E-mail
Phone No.
Air RISC Hotline
Information on health effects, urban toxics, risk [email protected] assessment, human health, and exposure.
919-541-0888
Hazardous Waste Ombudsman
Assists public and regulatory community in resolving problems associated with the Hazard- N/A ous Waste Program.
800-262-7937 202-260-9361
Pollution Prevention Information Clearinghouse (PPIC)
Provides answers and referrals in response to [email protected] questions about pollution prevention.
800-424-9346
Office of Pollution Prevention and Toxics (OPPT)
OPPT has the Non-confidential Information Center with public dockets for TSCA and Toxic Release Inventory (TRI) rulemaking actions, N/A TSCA administrative record, and non-confidential case files for documents submitted under TSCA.
202-260-7099
19 Regulations
Name
1305
Description
E-mail
Phone No.
Provides information on air pollution prevention and control technologies, permit requireRACT/BACT/LAER ments at stationary air pollution sources, and N/A Clearinghouse related Federal air pollution emission standards.
919-541-0800
RCRA Information Center
Provides access to all regulatory materials supN/A porting EPA’s actions under RCRA.
703-603-9230 800-424-9346
RCRA Information Hotline
Responds to requests about hazardous waste concerning identification, generators, transN/A porters, treatment, storage and disposal facilities, recycling sites, export, and import.
703-603-9230
Provides information and assistance regarding Safe Drinking Water drinking water regulations, the wellhead pro- hotlineHotline tection program, source water protection and [email protected] guidance, and education materials.
800-426-4791
Subsurface Remediation Information Center
Provides technical and scientific information N/A on groundwater protection and remediation.
580-436-8651
TSCA Assistance Information Service
Provides TSCA regulation information.
WasteWi$e Helpline
WasteWi$e is a voluntary program to [email protected] age businesses to reduce solid waste.
[email protected]
202-554-1404 800-372-9473
• Solvent Alternatives Guide (SAGE): SAGE is an Internet-based tool developed by
•
•
the Surface Cleaning Program at Research Triangle Institute (RTI) in cooperation with EPA’s Air Pollution Prevention and Control Division to identify solvent and process alternatives for parts cleaning and degreasing. SAGE works as an expert system asking the user a series of questions concerning the part’s size and volume, nature of the soil to be removed, production rate, etc. SAGE then produces a ranked list of candidate processes and chemistries most likely to work for a given situation. Since SAGE is based on readily available and proven processes and chemistries, it does not assist in the design of new solvents. SAGE is available at http://clean.rti.org/. Coating Alternatives Guide (CAGE): CAGE is based on similar principles found in SAGE but focuses on identifying alternatives for paint and coating formulations. CAGE was also developed by RTI in cooperation with EPA. It is an expert system designed to provide recommendations on low VOC/HAP coating alternatives for various substrates. It also provides the user with links to other useful websites. CAGE is available at http://cage.rti.org/. Enviro$en$e, first opened to the public in 1994, is a central reservoir of pollution prevention and cleaner production information and databases. It provides the user with a solvent substitution data system, compliance and enforcement assistance information, and a site for communication and exchange of information (Enviro$en$e Cooperatives). Enviro$en$e can be reached at http://es.epa.gov/.
1306
Carlos M. NuÔez
• Program for Assisting the Replacement of Industrial Solvents (PARIS): PARIS is a
•
•
•
solvent design software system developed by EPA, the National Research Council, and RTI to substitute offending solvents with a single chemical or a mixture of solvents based on physical and chemical properties and activity coefficients. PARIS also evaluates and considers the environmental properties of the substitute chemicals. The program’s recommended alternatives may require testing to validate their performance.70 Computer-Aided Molecular Design (CAMD): CAMD was developed by the Department of Chemical Engineering’s Computer-Aided Process Engineering Centre at the Technical University of Denmark. CAMD can be used to select and design new solvents based on thermodynamic properties. It contains a database with thousands of chemicals which can be accessed to select the desired chemical. If the chemical does not exist, CAMD uses computational chemistry to build the chemical configuration of the new chemical.70 EPA’s Environmental Technology Verification (ETV) program was established to verify the performance of innovative technical solutions to environmental and human health problems. Companies with new commercial-ready environmental technologies can participate in this program. For more information visit ETV’s website at http://www.epa.gov/etvprgrm/index.htm. Canada also instituted an ETV program to foster Canada’s environmental technologies (http://www2.ec.gc.ca/etad/etv_e.html). South Coast Air Quality Management District (SCAQMD) has established a program to certify clean air solvents for industrial cleaning. Large and small industries can be exempted from record-keeping requirements and emission fees if they use clean air solvents. For more information about SCAQMD’s Clean Air Solvent (CAS) Certification program visit http://www.aqmd.gov/tao/cas/cas.html.
19.9 SUMMARY Solvent releases can affect the quality of air, water, and soil which can then have adverse effects on human health and the environment. More stringent environmental laws and regulations have been established to control their utilization and ensure a safer and healthier environment and a sustainable future. They are placing greater emphasis on the elimination or reduction of such releases at the source and the preservation of limited natural resources. However, replacing offensive solvents requires a comprehensive assessment of their overall environmental impact. This will ensure that substitute chemicals will not impose more stress on our environment and human health by transferring the problem to other media. This chapter was intended to provide a “bird’s eye view” of key environmental laws and regulations for solvents. This chapter will not serve as a replacement of the laws discussed herein.
REFERENCES 1 2 3
Clean Air Act Amendments of 1970, Public Law 91-604, December 31, 1970. N.P. Cheremisinoff and M.L. Graffia, Environmental and Health & Safety Management: A Guide to Compliance, Noyes Publications, New York, NY, 1995. Pollution Prevention Act of 1990,42 U.S.C. §13101, et seq.
19 Regulations
4 5 6
7 8
9 10 11 12
13 14 15 16
17 18 19
20
21 22 23 24 25 26 27 28 29 30
1307
U.S. Environmental Protection Agency, 33/50 Program: The Final Record, EPA-745/R-99-004, Office of Pollution Prevention and Toxics, Washington, DC, 1999. Requirements for Preparation, Adoption, and Submittal for Implementation Plans, 40 Code of Federal Regulations § 51.100(s). S.V. Nizich, T. Pierce, A.A. Pope, P. Carlson, and B. Barnard, National Air Pollutant Emission Trends, 1900-1996, EPA-454/R-97-011, NTIS PB98-153158, Office of Air Quality Planning and Standards, Research Triangle Park, NC, December 1997. National Institute for Occupational Safety and Health Pocket Guide to Chemical Hazards (Publication No. 94-116), Cincinnati, OH (1994). I. Lundberg, G. Nise, G. Hedenberg, M. Högberg, and O. Vesterberg, Liver function tests and urinary albumin in house painters with previous heavy exposure to organic solvents, Occup. and Environ. Med., 51, 347 (1994). Clean Air Act Amendments of 1990, Public Law 101-549, November 15, 1990. Identification and Listing of Hazardous Waste, 40 Code of Federal Regulations 261. 1996 North American Emergency Response Guidebook, J.J. Keller & Associates, Inc., Washington, DC, 1996. U.S. Environmental Protection Agency, Title III List of Lists, Consolidated List of Chemicals Subject to the Emergency Planning and Community Right-to-Know Act (EPCRA) and Section 112(r) of the Clean Air Act, as Amended, EPA-550-B-98-017, Office of Solid Waste and Emergency Response, Washington, DC, November 1998. Occupational Safety and Health Standards, 29 Code of Federal Regulations, Chapter XVII, Part 1910. 1999 TLVs and BEIs, Threshold Limit Values for Chemical Substances and Physical Agents Biological Exposure Indices, American Conference of Governmental Industrial Hygienists, Cincinnati, OH (1999). S. Solomon and D. Albritton, Time-Dependent Ozone Depletion Potentials for Long- and Short-Term Forecasts, Nature, 357, 33 (1992). J.T. Houghton, L.G. Meira Filho, B.A. Callander, N. Harris, A. Kattenberg, and K. Maskell, Eds, Climate Change 1995, The Science of Climate Change, Contribution of Working Group 1 to the Second Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK (1996). Montreal Protocol, The International Lawyer, 26, 148 (1992). The Montreal Protocol, Harvard Environmental Law Review, 15, 275 (1991). J.S. Nimitz, Development of Nonflammable, Environmentally Compliant Fluoroiodocarbon Solvents: Phase I Final Report, Prepared for U.S.A.F. Wright Laboratories under contract F33615-94-C-5003, ETEC 95-2 (1995). U.S. Environmental Protection Agency, List of Substitutes for Ozone-Depleting Substances, last update May 1999, http://www.epa.gov/spdpublc/title6/snap/lists/index.html. Stratospheric Protection Division, Washington, DC, June 1999. Technical Guide No. 6, Handbook of Organic Industrial Solvents, American Mutual Insurance Alliance, Chicago, IL, 1980. J.P. Robinson and W.C. Nelson, National Human Activity Pattern Survey Data Base, U.S. EPA, Atmospheric Research and Exposure Assessment Laboratory, Research Triangle Park, NC, 1995. Reorganization Plan No. 3 of 1970, 3 CFR 1072. M. Seralgeldin, Symp. Environmentally Acceptable Inhibitors and Coatings, The Electrochemical Society, Impact of the Regulatory Environment on Green Technologies, Pennington, NJ, 1997, pp. 1-16. R.J. Martineau, Jr. and D.P. Novello, The Clean Air Act Handbook, American Bar Association, Chicago, IL, 1998. P.E.T. Douben and M.A. Serageldin, Pollution Risk Assessment and Management, P.E.T. Douben, Ed., John Wiley & Sons Ltd., New York, NY, 1998, pp. 49-91. Communication with William Johnson on December 7, 1999, EPA’s Office of Air Quality Planning and Standards, Research Triangle Park, NC (1999). Coating Regulations Workshop, April 8-9, 1997, EPA’s Office of Air Quality Planning and Standards, Durham, NC. Federal Register 64: 13422, Consumer and Commercial Products: Schedule for Regulation, Notice of revisions to schedule for regulation (1999). U.S. Environemental Protection Agency’s Technology Transfer Network, Unified Air Toxics Website: Rules and Implementation, http://www.epa.gov/ttn/uatw/eparules.html, Office of Air Quality Planning and Standards, Research Triangle Park, NC, December 21, 1999.
1308
31
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66
Carlos M. NuÔez
Federal Register 57: 31576, Initial list of Categories of Sources Under Section 112(c)(1) of the Clean Air Act Amendments of 1990, Notice of initial list of categories of major and area sources. U.S. Environmental Protection Agency (1992). Federal Register 61: 28197 and 37542, Revision of list of Categories of Sources and Schedules for Standards Under 112(c) and (e) of the Clean Air Act Amendments of 1990, Notice of revisions of initial list. G.S. Koch and P.R. Ammann, Current Trends in Federal Regulation of Hazardous Air Pollutants, J. Environ. Reg., 4, 25 (1994). P. Zurer, Ozone Treaty Tightened, CFC Substitutes Controlled, Chemical & Engineering News, 70, 5 (1992). Federal Water Pollution Control Act, Public Law 92-500 (1972). Clean Water Act Amendments of 1977, Public Law 95-217 (1977). Clean Water Act Amenments of 1982, Public Law 97-117 (1982). Clean Water Act of 1987 (Clean Water Act), Public Law 100-4 (1987). Navigation and Navigable Waters, 33 U.S.C. 1251-1387. L.M. Gallagher and L.A. Miller, Clean Water Handbook, Government Institutes, Inc., Rockville, MD, 1996. Safe Drinking Water Act, 42 U.S.C 300f et seq. (1974). Safe Drinking Water Act Amendments of 1986, Public Law 99-339 (1986). Safe Drinking Water Act Amendments of 1996, Public Law 104-182 (1996). E.J. Calabrese, C.E. Gilbert, and H. Pastides, Safe Drinking Water Act: Amendments, Regulations and Standards, Lewis Publishers, Chelsea, MI, 1989. RCRA: Reducing Risk from Waste, EPA-530/K-97-004, Office of Solid Waste and Emergency Response, Washington, DC, September 1997. Understanding the Hazardous Waste Rules: A Handbook for Small Businesses - 1996 Update, EPA-530/K-95-001, Office of Solid Waste and Emergency Response, Washington, DC, June 1996. Regulations Relating to Transportation, 49 Code of Federal Regulations 171-195. U.S. Environmental Protection Agency, The Common Sense Initiative, Lessons Learned About Protecting the Environment in Common Sense, Cost Effective Ways, EPA-100/R-98-011, Office of Reinvention, Washington, DC, December 1998. U.S. Environmental Protection Agency, Source Reduction Review Project, EPA-100/R-92/002, Office of Pollution Prevention and Toxics, Washington, DC, August 1992. U.S. Environmental Protection Agency, Project XL 1999 Comprehensive Report, EPA-100/R-99-008, Office of the Administrator, Washington, DC, 1999. J.W. Vincoli, Basic Guide to Environmental Compliance, Van Nostrand Reinhold, New York, NY, 1993. Commerce and Trade, 15 U.S.C. 2601, et seq. Occupational Safety and Health Act of 1970, 29 U.S.C. 651 et seq. B. Plog, Fundamentals of Industrial Hygiene, National Safety Council, Itasca, IL, 1996. OSHA. State Occupational Safety and Health Plans. http://www.osha-slc.gov/fso/osp/. January 18, 2000. National Institute for Occupational Safety and Health recommendations for occupational safety and health: Compendium of policy documents and statements, Publication No. 92-100, Cincinnati, OH (1992). G. Noll, M. Hildebrand, and J. Yvirra, Hazardous Materials: Managing the Incident, Oklahoma State University, Stillwater, OK, 1995. California’s Department of Health Services. http://www.ohb.org/solvmenu.htm, Oakland, CA, June 1, 1997. Commission for Environmental Cooperation. Summary of Environmental Law in North America, Last update October 1, 1995, htttp://www.cec.org/infobases/law, Montreal, Canada. L.K. Caldwell, International Environmental Policy, Duke University Press, Durham, NC, 1996. B.J.S. Hoetjes, Environmental Law and Policy in the European Union and the United States, R. Baker, Ed., Praeger Publishers,Westport, CT, 1997, pp. 31-44. K.Hanf, Environmental Law and Policy in the European Union and the United States, R. Baker, Ed., Praeger Publishers, Westport, CT, 1997, pp. 125-146. J. Ebbesson, Compatibility of International and National Environmental Law, Kluwer Law International, 1996. C.H.V. de Villeneuve, Environmental Law and Policy in the European Union and the United States, R. Baker, Ed., Praeger Publishers, Westport, CT, 1997, pp. 159-168. I.J. Koppen, Environmental Law and Policy in the European Union and the United States, R. Baker, Ed., Praeger Publishers, Westport, CT, 1997, pp. 234-247. C.M. Nunez, G.H. Ramsey, W.H. Ponder, J.H. Abbott, L.E. Hamel, and P.H. Kariher, Corona Destruction: An Innovative Control Technology for VOCs and Air Toxics, J. Air & Waste Management Assoc., 43, 242-247 (1993).
19 Regulations
67
68
69 70
1309
C.M. Nunez, A.L. Andrady, R.K. Guo, J.N. Baskir, and D.R. Morgan, Mechanical Properties of Blends of PAMAM Dendrimers with Poly(vinyl chloride) and Poly(vinyl acetate), J. Polymer Sci., Part A - Polymer Chemistry, 36, 2111-2117 (1998). C.M. Nunez, G.H. Ramsey, M.A. Bahner, and C.A. Clayton, An Empirical Model to Predict Styrene Emissions from Fiber-Reinforced Plastics Fabrication Processes, J. Air & Waste Management Assoc., 49, 1168-1178 (1999). U.S. Environmental Protection Agency, Headquarters Telephone Directory, EPA/208-B-99-001, Office of Information Resources Management, Washington, DC, April 1999. U.S. Environmental Protection Agency, Headquarters Telephone Directory, EPA/625/R-99/005, Office of Research and Development, Washington, DC, July 1999.
ACRONYMS ACGIH AEC AHERA BAC BACT BAT BCT BDT BMP BPT CAA CAGE CAMD CAS CDC CEPA CESQG CFC CSI CTG CWA DEPH DOL DOT DV EAP EC ECSC EEA EEC ENVIRO$EN$E EPA ETV EU EURATOM FFDCA FIFRA FIP FWPCA GWP HAP HAZCOM HCFC HCS HESIS HFE HMTA
American Conference of Governmental Industrial Hygienists Atomic Energy Commission Asbestos Hazardous Emergency Response Act Best Available Control Best Available Control Technology Best Available Technology Economically Achievable Best Conventional Pollutant Control Technology Best Available Demonstrated Technology Best Management Practices Best Practicable Control Technology Clean Air Act Coating Alternatives Guide Computer-Aided Molecular Design Clean Air Solvent Centers for Disease Control and Prevention Canadian Environment Protection Act conditionally exempt small quantity generator chlorofluorocarbon Common Sense Initiative Control Technology Guideline Clean Water Act bis (2-ethylhexyl)phthalate Department of Labor Department of Transportation Design Value Environmental Action Programme European Community European Coal and Steel Community European Environment Agency European Economic Community public environmental information system U.S. Environmental Protection Agency Environmental Technology Verification European Union European Atomic Energy Community Federal Food, Drug, and Cosmetic Act Federal Insecticide, Fungicide, and Rodenticide Act Federal Implementation Plan Federal Water Pollution and Control Act global warming potential hazardous air pollutant hazard communication hydrochlorofluorocarbon Hazard Communication Standard Hazard Evaluation System and Information Service hydrofluoroether Hazardous Materials Transportation Act
1310
HSWA LAER LCCA LQG MACT MCL MCLG MSDS NAAQS NESHAP NIOSH NPDES NPDWR NSPS NSR ODS OPPE OPPT OSH OSHA P2 PARIS PCB PCBTF PEL PM10 PMACT PMN POTW PPIC PSD RACT RCRA RISC SAGE SCAQMD SDWA SEP SIP SNAP SQG SRRP SWDA TLV® TRI TSCA TSD TWA UN UV VOC XL
Carlos M. NuÔez
Hazardous and Solid Waste Amendments Lowest Achievable Emissions Rate Lead Contamination Control Act large quantity generator Maximum Achievable Control Technology maximum contaminant level maximum contaminant level goal Material Safety and Data Sheet national ambient air quality standards national emission standards for hazardous air pollutants National Institute for Occupational Safety and Health National Pollutant Discharge Elimination System National Primary Drinking Water Regulations new source performance standards New Source Review ozone depleting substance Office of Policy, Planning, and Evaluation Office of Pollution Prevention and Toxics Occupational Safety and Health Occupational Safety and Health Administration Pollution Prevention Program for Assisting the Replacement of Industrial Solvents polychlorinated biphenyl p-chlorobenzotrifluoride permissible exposure limit particulate matter of 10 micrometers in aerodynamic diameter or smaller Preliminary Maximum Achievable Control Technology Premanufacture Notification publicly operated treatment works Pollution Prevention Information Clearinghouse prevention of significant deterioration Reasonable Available Control Technology Resource Conservation and Recovery Act Risk Information Support Center Solvent Alternatives Guide South Coast Air Quality Management District Safe Drinking Water Act superior environmental performance State Implementation Plan Significant New Alternatives Policy small quantity generator Source Reduction Review Project Solid Waste Disposal Act Threshold Limit Value Toxic Release Inventory Toxic Substances Control Act treatment, storage, and disposal time weighted average United Nations ultraviolet light volatile organic compound eXcellence and Leadership
19.10 Regulations in Europe
1311
19.10 REGULATIONS IN EUROPE Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Thbingen, Thbingen, Germany
The legislative and executive regulations in Europe are generally based on EEC regulations which are converted to and coordinated by national regulations as it is a rule in Europe. EEC regulations do not exist concerning all aspects, or they are not always converted effectively in the EEC Member States. In Europe environmental law, especially concerning chemical and hazardous substances, was first realized about 30-40 years ago.1 Main aims of EEC regulations are the registration and the classification of a wide variety of chemical substances, the environmental protection, the health protection, harmonization of national laws, and the liberalization of the market. For these reasons several EEC regulations were created and completed (see 19.10.1). Changes of regulations intended to optimize some aspects, e.g., environmental and health aspects by the obligation of registration and test procedures.7 The list 19.10.1 shows all relevant regulations in Europe concerning solvents. Relevant German regulations concerning solvents are listed in 19.10.2 as an example of the conversion of EEC regulations in Member States. In Germany the application of chemical and hazardous substances are based on a few regulations.28,30 Special and practical applications of these basic regulations are put in concrete terms in different standards, e.g., TRGS29 or DIN.31-56 19.10.1 EEC REGULATIONS 1
2
3 4 5
6
7
8
9 10
67/548/EEC, Council Directive of 27 June 1967 on the approximation of the laws, regulations and administrative provisions of the Member States relating to the classification, packaging and labeling of dangerous substances, ABl. EG Nr. L 196/1, 1967. 73/173/EEC, Council Directive of 4 June 1973 on the approximation of the laws, regulations and administrative provisions of the Member States relating to the classification, packaging and labeling of dangerous preparations (solvents), ABl. EEC Nr. L 189, 1973. Council Declaration (EC) and of the Member States of 22 November 1973 concerning an environmental program (EC), ABl. EG Nr. C 112/1, 1973. 75/442/EEC, Council Directive of 15 July 1975 on waste, ABl. EEC Nr. L 194/47, 1975. 76/769/EEC, Council Directive of 27 July 1976 on the approximation of the laws, regulations and administrative provisions of the Member States relating to limitations of the use of certain dangerous substances and preparations, ABl. EEC Nr. L 262/201, 1976. 77/728/EWG, Council Directive of 7 November 1977 on the approximation of the laws, regulations and administrative provisions of the Member States relating to the classification, packaging and labeling of paints, varnishes, printing paints, adhesives etc., ABl. EEC Nr. L 303/23, 1977. 79/831/EEC, Council Directive of 18 September 1979 on the approximation of the laws, regulations and administrative provisions of the Member States relating to the classification, packaging and labeling of dangerous preparations, ABl. EEC Nr. L 154/1, 1979. 88/379/EEC, Council Directive of 7 June 1988 on the approximation of the laws, regulations and administrative provisions of the Member States relating to the classification, packaging and labeling of dangerous preparations, ABl. EEC Nr. L 187/14, 1988. 88/344/EEC, Council Directive of 13 June 1988 on the approximation of the law of the Member States on extraction solvents used in the production of foodstuffs and food ingredients, ABl. EEC Nr. L 57/28, 1988. 89/349/EEC, Commission Recommendation of 13 April 1989 on the reduction of chlorofluoro-carbons by the aerosol industry, Abl. EG Nr. L 144/56, 1989.
1312
11
12 13 14 15
16 17 18 19
20 21 22 23 24 25
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg
91/155/EEC, Commission Directive of 5 March 1991 defining and laying down the detailed arrangements of the system of specific information relating to dangerous preparations in implementation of Article 10 of Directive 88/379/EEC, ABl. EEC Nr. L 76/35, 1991. 91/689/EEG, Council Directive of 12 December 1991 on hazardous waste, ABl. EG Nr. L 23/29, 1993. 92/2455/EEC, Council Regulation (EEC) of 23 July 1992 concerning the export and import of certain dangerous chemicals, ABl. EEC Nr. L 251/13, 1992. 93/793/EEC, Corrigendum to Council Regulation (EEC) No. 793/93 of 23 March 1993 on the evaluation and control of the risks of existing substances, ABl. EEC Nr. L 224/34, 1993. 93/67/EEC, Commission Directive of 20 July 1993 laying down the principles for assessment of risks to man and the environment of substances notified in accordance with Council Directive 67/548/EEC, ABl. EEC Nr. L 227/9, 1994. Council Regulation (EEC) No. 259/93 of 1 February 1993 on the supervision and control of waste within, into and out of the European Community, ABl. EG Nr. L 30/1, 1993, Commission decision of 20 December 1993 concerning a list of hazardous waste pursuant to Article 1(a) of Council Directive 91/689/EEC on hazardous waste, ABl. EG Nr. L5/15, 1994. 94/1179/EEC, Commission Regulation of 25 May 1994 concerning the first list of priority substances as foreseen under Council Regulation (EEC) No. 793/93, ABl. EEC Nr. L 161/3, 1994. 94/1488/EEC, Commission Regulation of 28 June 1994 laying down the principles of the assessment of risks to mass and the environment of existing substances in accordance with Council Regulation No. 793/93, ABl. EEC Nr. L 161/3, 1994. Council Regulation (EEC) No. 3093/94 of 15 December 1994 on substances that deplete the ozone layer, ABl. EG Nr. L 333/1, 1994. 94/904/EEC, Council Decision of 22 December 1994 establishing a list of hazardous waste pursuant to Article 1(4) of Council Directive 91/689/EEC on hazardous waste, ABl. EG Nr. L 356/4, 1994. 95/365/EEC, Commission Decision of 25 July 1995 establishing the ecological criteria for the award of the Community eco-label to laundry detergents, ABl. EG Nr. L 217/14, 1995. 95/2268/EEC, Commission Regulation of 27 September 1995 concerning the second list of priority substances as foreseen under Council Regulation No. 793/93, ABl. EEC Nr. L 231/18, 1995. 96/61/EEC, Council Directive of 24 September 1996 concerning integrated pollution prevention and control, ABl. EG Nr. L 257/26, 1996. 97/142/EEC, Commission Regulation of 27 January 1997 concerning the delivery of information about certain existing substances, ABl. EEC Nr. L 25/11, 1997.
19.10.2 GERMAN REGULATIONS 26
27
28 29
30 31 32 33 34 35 36
Verordnung über Höchstmengen an bestimmten Lösemitteln in Lebensmitteln (Lösemittel-HöchstmengenVerordnung - LHmV) vom 27. Juli 1989. Regulation concerning maximum quantities of certain solvents in foods (LHmV) of 27 July 1989, BGBl., 1568 (1989). Verordnung über die Entsorgung gebrauchter halogenierter Lösemittel (HKWAbfV) vom 23. Oktober 1989. Regulation concerning the waste management of halogenated solvents (HKWAbfV) of 23 October 1989, BGBl., 1918 (1989). Verordnung zum Schutz vor gefährlichen Stoffen (Gefahrstoffverordnung - GefStoffV) vom 26.10.1993. Regulation concerning the protection against hazardous substances of 26 October 1993. Technische Regeln für Gefahrstoffe (TRGS). Technical Regulations concerning hazardous substances (TRGS), e.g. TRGS 002, TRGS 003, TRGS 101, TRGS 102, TRGS 150, TRGS 220, TRGS 222, TRGS 400, TRGS 402, TRGS 403, TRGS 404, TRGS 415, TRGS 420, TRGS 512, TRGS 519, TRGS 531, TRGS 900, TRGS 903, TRGS 905, TRGS 906, TRGS 910. Gesetz zum Schutz vor gefährlichen Stoffen (Chemikaliengesetz - ChemG) vom 25. Juli 1994. Law concerning the protection against hazardous substances of 25 July 1994, BGBl. I, 1703 (1994). DIN 53169, Solvents for paints and varnishes; determination of density, refractive index, flash point, acid value, saponification value, olefinic and aromatic content in hydrocarbons, 1991. DIN 53170, Solvents for paints and similar coating materials; determination of the evaporation rate, 1991. DIN 53171, Solvents for paints and varnishes; determination of distillation characteristics (boiling range and boiling temperature as a function of distilled volume), 1991. DIN 53172, Solvents for paints and varnishes; determination of evaporation residue, 1993. DIN 53173, Solvents for paints and varnishes; determination of carbonyl value, 1991. DIN 53174-1, Solvents for paints and varnishes - Methods of test for solvent mixtures - Part 1: General references and survey, 1995.
19.10 Regulations in Europe
37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56
1313
DIN 53174-2, Solvents for paints and varnishes and similar coating materials; methods of analysis for solvent mixtures; gas chromatographic method, 1992. DIN 53175, Binders for paints, varnishes and similar coating materials; determination of the solidification point (titer) of fatty acids, 1991. DIN 53245, Solvents for paints and varnishes; alcohols; supply specification, further properties and methods of test, 1994. DIN 53246, Solvents for paints and varnishes - Acetic esters - Delivery specification, further requirements and methods of test, 1997. DIN 53247, Solvents for paints and varnishes - Ketones - Supply specifications, further requirements and methods of test, 1997. DIN 53248, Solvents for paints, varnishes and similar coating materials - Gum spirit of turpentine and wood turpentines - Requirements and methods of test, 1995. DIN 53249, Solvents for paints, varnishes and similar coating materials - Dipentene - Requirements and methods of test, 1995. DIN 55651, Solvents for paints and varnishes - Symbols, 1997. DIN 55681, Solvents; stability testing of trichloroethylene, 1985. DIN 55682, Solvents for paints and varnishes - Determination of solvents in water-thinnable coating materials - Gas chromatographic method, 1994. DIN 55682/A1, Solvents for paints and varnishes - Determination of solvents in water-thinnable coating materials - Gas chromatographic method; Amendment A1, 1998. DIN 55683, Solvents for paints and varnishes - Determination of solvents in coating materials containing organic solvents only - Gas chromatographic method, 1994. DIN 55685, Solvents for paints and varnishes; alcohols; gas chromatographic determination of the degree of purity, 1992. DIN 55686, Solvents for paints and varnishes; acetic esters; gas chromatographic determination of the degree of purity, 1992. DIN 55687, Solvents for paints and varnishes; ketones; gas chromatographic determination of the degree of purity, 1992. DIN 55688, Solvents for paints and varnishes - Ethylene glycol ethers - Gas chromatographic determination of the degree of purity, 1995. DIN 55689, Solvents for paints and varnishes - Propylene glycol ethers - Gas chromatographic determination of the degree of purity, 1995. DIN 55997, Solvents for paints and varnishes - Deionized water - Requirements and methods of test, 1998. DIN 55998, Solvents for paints and varnishes - Propylene glycol ethers - Supply specification, further requirements and methods of test, 1998. DIN 55999, Solvents for paints and varnishes, ethylene glycol ethers; supply specification, further properties and methods of test, 1994.
20
Toxic Effects of Solvent Exposure 20.1 TOXICOKINETICS, TOXICODYNAMICS, AND TOXICOLOGY Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Thbingen, Thbingen, Germany
20.1.1 TOXICOKINETICS AND TOXICODYNAMICS 20.1.1.1 Exposure Highest exposures can be found in workplace (e.g., evaporation of solvents) or during special processes (e.g., leaks of normally closed systems). Acute and severe solvent accidents often happen in workplaces (high solvent concentrations, intermittent high-level exposures, high duration of exposure). Apart from working sites, various other emission sources of solvents should be considered, e.g., consumer products. The description of exposure parameters (type of solvents, concentrations, duration, routes of exposure) are important for the evaluation of toxicokinetics. Solvents and other chemicals are usually emitted as a mixture of various substances. Therefore, the risk assessment of emitted solvents is difficult to ascertain.1,2 Solvent concentrations and duration of exposure vary in most cases (intermittent high-value peaks, periods of low exposure). The exposure is influenced essentially by surrounding occupational and environmental conditions, such as working climate, protective equipment and by individual parameters such as eating habits. The exposure to solvents is regulated by relevant threshold limit values.1,2 Exposure and exposure values can be controlled by defined methods (e.g., ambient and biological monitoring). 20.1.1.2 Uptake Relevant uptake routes of solvents are absorption from the lung and percutaneous absorption. The intestinal uptake is usually caused by accidents or by intent. The absorption rate is influenced by various factors.
1316
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg
20.1.1.2.1 Inhalation Inhalation is the most common pathway of solvent absorption, especially at working sites. The pulmonal absorption of solvents depends on the following parameters:3-6 • Exposure (concentrations and concentration fluctuations in the ambient air, exposure time, physical exertion). The alveolar concentration of solvents or the difference between air and blood concentration levels determine the diffusion process into alveolar blood vessels. Physical exertion influences lung parameters, especially ventilation, and consequently alveolar and blood concentrations. • Lung parameters (pulmonary and alveolar ventilation, pulmonary perfusion, air-blood coefficient, blood-tissue coefficient). These coefficients describe the amount of solvents which can diffuse. The blood-tissue partition coefficient influences the tissue equilibrium concentrations. Solvents with stronger hydrophobic properties (e.g., toluene) reach equilibrium more rapidly because of a low tissue-blood coefficient. Intraindividual differences such as child/adult are also of significance. • Physicochemical characteristics of solvents (solubility such as hydrophobic and hydrophilic properties, state such as liquid or gaseous and degree of volatility). 20.1.1.2.2 Dermal uptake Dermal uptake of solvents requires skin contact and depends on the area of contact, skin thickness, dermal state (e.g., eczema and defects in the stratum corneum), exposure parameters (contact time, etc.) and solvent properties.7,8 The main barrier against percutaneous uptake of solvents are structures of the stratum corneum, especially intercellular lipids and fibrous keratin. Removal of lipids by polar solvents such as ethanol or hydration in the stratum corneum is associated with an increase of skin permeability. Defects or lack of stratum corneum that may occur in skin diseases, at particular skin locations such as hair follicles or glandula regions enhance the percutaneous movement of solvents. The absorption through mucosa membranes is facilitated because of the lack of the stratum corneum. Skin defects or diseases can be provoked by solvents which cause irritation, cellular hyperplasia and swelling, or removal of lipids. Skin defects are provoked mainly by frequent use of solvents thus enhancing their absorption. Other characteristics, which influence percutaneous absorption, are solvent concentration gradients, solvent partitioning (water/lipid partition coefficient) and permeability constants. Lipophilic chemicals are absorbed most easily (for example, benzene). These can include liquid solvents or solvents having low vapor pressure.9-11 Vapors absorbed by dermal uptake can significantly contribute to the body burden as a result of the whole body exposure: e.g. 1-2 % of xylene or toluene, up to 5-10 % 1-methoxypropane-2-ol.10 For other substances, much higher skin absorption rates were measured after the whole body exposure: 2-methoxyethanol up to 55 %, 2-ethoxy-ethanol up to 42 %.12 It is important to consider that the dermal uptake of vapors is especially significant when using a gas-mask.10 In addition to inhalation measurements, measurement of percutaneous absorption is an important method for assessing health or environmental risks.
20.1 Toxicokinetics, toxicodynamics, and toxicology
1317
Dermal absorption of solvents is shown in Table 20.1.1. Table 20.1.1. Dermal uptake of solvents according to the German MAK-list.2,64 Benzene Bromomethane 2-Butanone 2-Butoxyethanol Carbon disulfide 2-Chloroethanol Chloromethane Cresol(s) Cycolhexanol
Cyclohexanone Dimethylformamide Dimethylsulfoxide 1,4-Dioxane 2-Ethoxyethanol Ethylbenzene Ethyl formate Ethylene glycol n-Hexane
2-Hexanone Methanol 2-Methoxyethanol Methyl formate Nitrobenzene Nitrotoluene(s) Phenol iso-Propyl benzene n-Propanol (from ACGIH1)
1,1,2,2-Tetrachloroethane Tetrachloroethene Tetrachloromethane Toluene Toluidine(s) 1,1,2-Trichloroethane Trichloromethane Xylene(s)
20.1.1.2 Metabolism, distribution, excretion Specific toxicity of solvents is directly related to their metabolism which is predominantly catalyzed by cytochrome P-450 mixed-function oxidases in the liver or other tissues. Relevant examples of specific metabolism are toxic epoxides of benzene (hemopoietic toxicity), n-hexane 2,5-hexanedione (peripheral neurotoxic effects), metabolites of ethylene glycol ethers (reproductive toxicity), and unidentified metabolites from trichloroethylene (renal-toxic effects).13 It should be emphasized that only the metabolites of these solvents are associated with toxic effects. Other relevant metabolic pathways result in detoxified substances, such as biotransformation processes in the liver − conjugation with glycine, glucuronic acid and sulphuric acid (e.g., via hydroxylation of toluene) or biotransformation by hydrolysis, oxidation and conjugation (e.g., glycol ethers). It should be noted that metabolism processes vary according to the following conditions:14 • Species, sex, age, genetics, e.g., variability in enzymatic factors such as polymorphisms (cytochrome systems) or tissue repair mechanisms15 • Life style − diet, smoking, drug consumption, physical activity • Saturation. Massive concentrations of solvents result in saturation of metabolic pathways. This is important with regard to detoxification • Induction of enzymes. Specific induction of enzyme systems by chemicals (solvents as well as other chemicals such as drugs) may consequently provoke an increase or decrease of solvent toxicity • Interactions may be involved in enhancing or reducing toxicity of solvents. For example Bloch et al.16 showed that in cases of alcohol abuse an increase in the toxic effects of benzene and other lipophilic petroleum derivatives occurs. Also, it has been shown that benzene inhibits the metabolism of toluene.17 Solvents can be excreted via various pathways: • Exhalation (unchanged) • Urine tract and biliary tract (unchanged or metabolites, e.g. water-soluble conjugates) 20.1.1.3 Modeling of toxicokinetics and modifying factors The complexity of toxicokinetic processes of solvents can be described in models, e.g., predicting exposure situations and distribution phenomena in the human body and quantifying these processes (e.g. dose-effect response relationships). This applies especially to simula-
1318
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg
tion of physiological and physicochemical parameters18 or to assessing low exposures to complex chemical mixtures.19 20.1.2 TOXICOLOGY 20.1.2.1 General effects General effects of solvents concern primarily acute exposures to high solvent concentrations. Despite some variations of symptoms, the resulting effects on the central nervous system (CNS) are rather stereotypical.20 Several solvents have depressant or narcotic effects, and hence, some solvents are used as anesthetics.21 The main acute health hazards result from the narcotic effects. Their intensity is proportional to the solvent concentrations in brain tissue and is caused by the solvents themselves (physical and chemical interactions with neural membranes, nerve cells or neurotransmitters of the CNS). General CNS dysfunctions after solvent exposure, are initially euphoria and disinhibition, higher exposures result in pre-narcotic symptoms such as dizziness, euphoria, disorientation and confusion, nausea, headache, vomiting, ataxia, paresthesia, increased salivation and tachycardia.22,23 The symptoms are rapidly reversible when the solvents are removed. In addition to the non-specific acute narcotic effects of solvents mentioned above, alterations of behavioral, cognitive and psychomotoric functions are typically found after short-term exposure to solvent levels close to the TLV. Overexposure leads to convulsions, coma and death. Typical changes are paresthesias, visual and auditory deficits, cognitive deficits (short-term and long-term memory loss), confusion, disorientation, affective deficits (nervousness, irritability, depression, apathy, compulsive behavior) and motor deficits (weakness in extremities, incoordination, fatigue, tremor).24,25 It is difficult to develop useful methods and models for testing these behavioral effects of solvents but for this purpose tests of attention and reaction, cognitive tests and other test systems are used.26,27 Acute CNS dysfunction diseases can show mild (organic affective syndrome), moderate or severe (acute toxic encephalopathy) symptoms.28,29 Unspecific irritations of skin and mucosa membrane structures can be caused by solvents. Various solvents are significant occupational irritants, e.g., solvents which cause irritant contact dermatitis.30 Intact skin structures can be destroyed by solvents which dissolve grease and fat. Typically, the dermatitis is characterized by dryness, scaling and fissuring and is usually located on the hands. It is often caused by handling solvent-contaminated products or by cleaning procedures.31,32 Unspecific irritation of mucous membranes is often caused by solvent vapors, e.g., irritation of the eyes and various sections of the airways. 20.1.2.2 Specific non-immunological effects Table 20.1.2 summarizes the main specific effects of solvents:33-47 • Hepatotoxicity • Nephrotoxicity • Reproductive toxicity • Hemopoietic toxicity • Neurotoxicity • Ocular toxicity
20.1 Toxicokinetics, toxicodynamics, and toxicology
1319
Table 20.1.2. Examples for specific effects of selected solvents Organ-system
Solvents
Symptoms
Liver
halogenated hydrocarbons (e.g., carbon tetrachloride, tetrachloroethane, chloroform), acute (necrosis, steatosis) and chronic (cirethanol, 1,1,1-trichloroethane, trichloroeth- rhosis) hepatotoxic symptoms ylene, bromobenzene, dimethylformamide
Kidney
acute tubular necrosis, glomerular and tubuhalogenated hydrocarbons (e.g., carbon tetlar dysfunctions (e.g., albuminuria, rachloride), toluene, dioxane, diethylene proteinuria), glomerulonephritis, note: modglycol, ethylene glycol, glycol ethers, conjuification of solvent effects caused by renal gates of trichloroethylene dysfunctions possible
Reproductive system
carbon disulfide, benzene, glycol ethers, disturbance of menstrual cycle; reduced nitrobenzene sperm counts, embryotoxic effects
Hemopoietic system
benzene metabolites (e.g., benzoquinone, marrow depression, myelotoxic effects hydroquinone)
Nerval system
n-hexane, ethanol, styrene, tetrachloroethylene
peripheral neuropathy (especially distal axons, axon swelling and degeneration, loss of sensibility, muscular atrophy, loss of tendon reflexes)
Eye
methanol
impaired vision
Note: the data shown come predominantly from data of occupational exposure.
20.1.2.3 Immunological effects Various solvents have well-known allergic potentials. Allergic symptoms of the respiratory tract (rhinitis, tracheitis, bronchitis, asthma), allergic contact dermatitis and conjunctivitis can be provoked by solvents. The allergic effects of solvents can also contribute to other diseases such as MCS, autoimmune diseases. Nowadays, solvents or by-products with allergic potential occur mainly at workplaces and, to a lesser degree, in consumer products. According to EG regulations, solvent ingredients of some consumer products, e.g., cosmetic products, must be labeled. It is often difficult to detect the causative solvent allergen (allergens which cause cross allergies, secondary products of solvents such as oxygenated terpenes, unknown allergens). Various specific test systems are available for carrying out individual test diagnoses: e.g., chamber tests,48 skin tests such as patch-tests49 and special applications of biological monitoring. Solvent-induced allergies can occur at a variety of working sites, e.g., in shoe factories,50 in electronic industries,51 in synthetic chemical industries,52 in metal industries53 or in perfume and potter industries (oil of turpentine and other solvents).54 Similar occurrence of solvents can be found in consumer products, e.g., in nail polishes (e.g., toluene).55 Allergic solvent substances are listed in various catalogues and databases.1,2,49 Examples of allergic solvents are terpene products with high sensitivity potential, which can cause positive test reactions (patch-test) or even allergic diseases (contact sensitization and dermatitis). Allergic dermatitis can even be provoked by d-limonene in the air.56 Terpenes and terpenoid substances are found especially in “natural products”, e.g., cosmetic products, foods, and plants (oilseed rape).57,58
1320
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg
Allergic potential of solvent products depends on the typical solvent structure. For example, in glycol ethers their allergic potential is proportional to the charge of interacting molecules.59 Allergic effects can also be associated with other skin conditions caused by solvents such as irritations. Multiple areas of skin damage, including solvent allergies, can change the skin structure and provoke severe skin disease.60 In addition to other substances (pesticides, food additives, dust, smoke, etc.), allergic effects of solvents are discussed as an initial cause of MCS.61 Organic solvents are associated with human autoimmune diseases, but defined pathomechanisms of these solvents have not yet been detected (role of solvents in the initiation or progression of autoimmune diseases).62 20.1.2.4 Toxic effects of solvents on other organisms In addition to humans, microorganisms animals and plants are also exposed to solvents. The interaction between organisms and solvents are often specific. For example, the reactions elicited by certain solvents depend on the species and abilities of the particular organism affected. Hydrophobic organic solvents, in particular, are toxic to living organisms, primarily because they disrupt cell membrane structure and mechanisms. Some living organisms especially certain bacterial species, are able to adapt to these solvents by invoking mechanisms such as accelerating repair processes (through changes in the rate of phospholipid biosynthesis), reduction of the diffusion rate of the solvent and active reduction of the intracellular concentration of the solvent. More information and examples are shown in Chapter 14.4.2. 20.1.2.5 Carcinogenicity The term carcinogenicity is used for toxicants that are able to induce malignant neoplasms. Carcinogens can be effective at different stages of the carcinogenic process, e. g., initiation, promotion and progression. They may interact with other noxes and thereby enhance tumor development. Interactive carcinogenesis can be described as co- and syn-carcinogenesis. A co-carcinogen is defined as a non-carcinogenic compound that is able to enhance tumor development induced by a given carcinogen. In syn-carcinogenesis two or more carcinogens, each occurring in small amounts that are usually not sufficient to induce a tumor in a specific target organ, may interact to lead to tumor formation in that organ. As with all carcinogens the carcinogenic potency of solvents has been assessed by short-term in vitro tests, e. g., Ames assay, by long-term tumor induction experiments in animals and - especially important for the evaluation of the carcinogenic action in humans prospective and retrospective epidemiological studies, for solvent exposure mainly in work places. From this data it is generally not possible to evaluate the carcinogenic action of solvent mixtures, which occur in the majority of exposure situations. It is also important to note, that for a number of reasons, e. g., very long latency period of tumor generation, accumulation of single hits in the target cells, significance of repair mechanisms it is not possible to define TLVs for carcinogens. In accordance with the evidence available, different classes for chemical carcinogens have been developed by health authority organizations.1,2,34-36 Examples of the classification of carcinogenic solvents are presented in Table 20.1.3.
20.1 Toxicokinetics, toxicodynamics, and toxicology
1321
Table 20.1.3. Carcinogenicity - Survey of selected solvents Solvent
Category*
Organ-System MAK
EG
ACGIH
IARC
NTP
Benzene
hemopoietic system
1
K1
A1
1
K
Bromomethane
upper gastrointestinal, tract and respiratory, tract (animals)
3
K3
n.c.**
3
n.l.**
Carbon tetrachloride
lymphatic system, liver (mice, rats), mamma (rats), suprarenal gland (mice)
3
K3
A2
2B
R
Epichlorohydrin
lung, CNS, forestomach (rats), nasal cavity, skin (mice)
2
K2
A3
2A
R
Chloroethane
uterus (mice)
3
K3
n.l.
3
n.l.
Cyclohexanone
suprarenal gland (rats)
3
n.c.
A4
3
n.l.
1,2-Dibromoethane
forestomach (mice), lung (mice, rats), nasal cavity, peritoneum, mamma, connective tissue (rats)
2
K2
A3
2A
R
1,2-Dichloroethane
brain, lymphatic and hemopoietic system, stomach, pancreas; lung, mamma, stomach (mice, rats), lymphatic system (mice)
2
K2
A4
2B
R
Dichloromethane
liver, lung (mice, rats), mamma (rats), lymphosarcomas (mice)
3
K3
A3
2B
R
3
K3
A4
3
n.l.
n.c.
n.c.
A4
3
n.l.
1,2-Dichloropropane liver (mice), mamma (rats) Dimethylformamide
testes
1,4-Dioxane
liver (rats, guinea pigs), biliary tract (guinea pigs), mamma , peritoneum (rats), nasal cavity (mice)
4
n.i.* *
A3
2B
R
1,2-Epoxypropane
mamma, upper respiratory tract, thyroid gland (mice, rats)
2
K2
A3
n.l.
n.l.
Hexamethyl phosphoramide
nasal cavity, lung (rats)
n.l.
n.i.
A3
2B
R
2-Nitropropane
liver (rats)
2
n.l.
A3
2B
R
Nitrobenzene
lung, thyroid gland, mamma (mice), liver, kidney, uterus (rats)
3
K3
A3
2B
n.l.
2- Nitrotoluene
epididymis (rats)
2
K2
n.c.,BEI**
3
n.l.
Phenol
lymphatic system, hemopoietic system suprarenal gland, thyroid gland, skin (mice, rats)
3
n.c.
n.c.,BEI
3
n.l.
Tetrachloroethane
liver (mice)
3
K3
A3
3
n.l.
Tetrachloroethylene
oesophagus, kidney, hemopoietic system, lymphatic system; liver (mice), hemopoietic system (rats)
3
K3
A3
2A
R
1322
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg
Solvent
Category*
Organ-System MAK
EG
ACGIH
IARC
NTP
Tetrachloromethane
stomach, liver, kidney, thyroid gland (rats, mice)
3
K3
A2
n.l.
n.l.
o-Toluidine
mamma, skin, bladder, liver, spleen, peritoneum, connective tissue (rats), vessels (mice)
2
n.i.
A3
n.l.
R
1,1,2-Trichlorethane
liver, suprarenal gland (mice)
3
K3
A4
3
n.l.
Trichloroethylene
kidney; liver, biliary tract, kidney, lung, cervix, testes, lymphatic system (rats, mice)
1
K3
A5
2A
n.l.
Chloroform
stomach, liver, kidney, thyroid gland (mice, rats)
4
K3
A3
n.l.
R
1,2,3-Trichloropropane
oral mucosa (mice, rats), uterus (mice), liver, pancreas, forestomach, kidney, mamma (rats)
2
n.i.
A3
2A
R
*Categories MAK (German regulations)2 1: substances that cause cancer in humans and can be assumed to make a significant contribution to cancer risk. Epidemiological studies provide adequate evidence of a positive correlation between the exposure of humans and the occurrence of cancer. Limited epidemiological data can be substantiated by evidence that the substance causes cancer by a mode of action that is relevant to humans. 2: substances that are considered to be carcinogenic for humans because sufficient data from long-term animal studies or limited evidence from animal studies substantiated by evidence from epidemiological studies indicate that they can make a significant contribution to cancer risk. Limited data from animal studies can be supported by evidence that the substance causes cancer by a mode of action that is relevant to humans and by results of in vitro tests and short-term animal studies. 3: substances that cause concern that they could be carcinogenic for humans but cannot be assessed conclusively because of lack of data. In vitro tests or animal studies have yielded evidence in one of the other categories. The classification in Category 3 is provisional. Further studies are required before a final decision can be made. A MAK value can be established provided no genotoxic effects have been detected. 4: substances with carcinogenic potential for which genotoxicity plays no or at most a minor role. No significant contribution to human cancer risk is expected provided the MAK value is observed. The classification is supported especially by evidence that increases in cellular proliferation or changes in cellular differentiation are important in the mode of action. To characterize the cancer risk, the manifold mechanisms contributing to carcinogenesis and their characteristic dose-time-response relationships are taken into consideration. 5: substances with carcinogenic and genotoxic potential, the potency of which is considered to be so low that, provided the MAK value is observed, no significant contribution to human cancer risk is to be expected. The classification is supported by information on the mode of action, dose-dependence and toxicokinetic data pertinent to species comparison. EG65 K1: confirmed human carcinogen K2: compounds which should be considered as carcinogen K3: compounds with possible carcinogenic evidence ACGIH1 A1: confirmed human carcinogen A2: suspected human carcinogen A3: confirmed animal carcinogen with unknown relevance to humans A4: not classifiable as a human carcinogen A5: not suspected as a human carcinogen
20.1 Toxicokinetics, toxicodynamics, and toxicology
1323
IARC34-36 1: carcinogenic to humans 2A: probably carcinogenic to humans 2B: possibly carcinogenic to humans 3: not classifiable as to its carcinogenicity to humans NTP66 K: Known to be a Human Carcinogen R: Reasonably Anticipated to be a Human Carcinogen (RAHC) **Notes: italic: cancer in humans n.c.: not classified as carcinogenic n.i.: no information available n.l.: not listed BEI: not classified as carcinogenic but biological monitoring is recommended
20.1.2.6 Risk assessment For risk assessment of solvent exposure, and in addition to factors for general risk assessment (age, gender, race, diet, physical activity, stress, physical noxes, etc.) it is important to consider: • Occupational exposure (high doses) and environmental exposure (low doses) to solvents separately. • The effect of exposure time, e. g., life long environmental low exposure or occupational intermittent high exposure. • Exposure assessment (generally the most neglected aspect in risk assessment). This involves extensive ambient monitoring over a long period of time. Only a small amount of data on biological monitoring of solvents and/or metabolites (representing the “effective” dose) is available. • The high volatility of solvents, e. g., VOCs and the fast biotransformation rate (in the environment and within the human body) for most of the solvents. • Complex mixtures and numerous sources of environmental exposure. • Especially for environmental solvent exposure: High-to-low-dose extrapolation for evaluation of adverse health effects may be misleading. • Confounding factors, e.g., smoking and alcohol consumption, as adverse health effects which may dominate in cases of low solvent exposure. • Risk in this context is defined in terms of the probability as occurrence of a particular adverse health effect, e. g. 1 in 106. • Finally, as in general risk assessment, definition of a risk level that is acceptable. 20.1.3 CONCLUSIONS • For solvent exposure at workplaces considerable amount of evidence for adverse health effects has been gathered. • In this regard, specific and carcinogenic effects in particular have been discussed (see Table 20.1.2 and 20.1.3). • For environmental solvent exposure only a few examples of adverse health effects have been documented. • It is rather unlikely that potentially toxic environmental solvent exposures, e. g., benzene or halogenated hydrocarbons, can be prevented in the near future.
1324
Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg
• Many suspicions, but only a small amount of scientific data demonstrate a correlation between “environmental diseases”, e. g., sick building syndrome and solvent exposure. • It has been hypothesized that - as a rule - exposure to mixtures of solvents at low non-toxic doses of the individual constituent represents no danger to health.63 • There exists overwhelming evidence of adverse health effects caused by accepted environmental noxes such as tobacco smoke and the consumption of alcoholic beverages. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
35
ACGIH, TLV´s and BEI´s, ACGIH, Cincinnati, 1998. DFG, MAK- und BAT-Werte-Liste, VCH, Weinheim, 1999. I. Astrand, Scand. J. Work Environ. Health, 1, 199, 1975. I. Astrand in Occupational Health Hazards of Solvents, A. England, K. Ringen, M.A. Mehlman, Eds., Princeton, NJ, 1986, pp. 141-142. K.H. Cohr, in Safety and Health Aspects of Organic Solvents, V. Riihimäki, U. Ulfvarson, Eds., Alan R. Liss, N.Y., 1986, pp. 45-60. J.J.G. Opdam, Br. J. Ind. Med., 46, 831 (1989). M.K. Bahl, J. Soc. Cosmet. Chem., 36, 287 (1985). M. Bird, Ann. Occup. Hyg., 24, 235 (1981). J. Angerer, E. Lichterbeck, J. Bergerow, S. Jekel, G. Lehnert, Int. Arch. Occup. Environ. Health, 62, 123 (1990). I. Brooke, J. Cocker, I. Delic, M. Payne, K. Jones, N.C. Gregg, D. Dyne, Ann Occup. Hyg., 42, 531 (1998). G. Johanson, Toxicol. Lett., 43, 5 (1988). S. Kezic, K. Mahieu, A.C. Monster, F.A. de Wolff, Occup. Environ. Med., 54, 38 (1997). E.A. Lock, Crit. Rev. Toxicol., 19, 23 (1988). A. Lof, G. Johanson, Crit. Rev. Toxicol., 28, 571 (1998). H.M. Mehendale, Toxicology, 105, 251 (1995). P. Bloch, A. Kulig, M. Paradowski, T. Wybrzak-Wrobel, Pol. J. Occup. Med., 3, 69 (1990). O. Inoue, K. Seiji, T. Watanabe, M. Kasahara, H. Nakatsuka, S.N. Yin, G.L. Li, S.X. Cai, C. Jin, M. Ikeda, Int. Arch. Occup. Environ. Health, 60, 15 (1988). V. Fiserova-Bergerova, Scand. J. Work Environ. Health, 11, 7 (1985). S. Haddad, K. Krishnan, Environ. Health Perspect., 106, 1377 (1998). W.K. Anger in Neurobehavioral Toxicology, Z. Annau, Ed., John Hopkins University Press, Baltimore, MD, 1986, pp. 331-347. A. Laine, V. Riihimäki in Safety and Health Aspects of Organic Solvents, V. Riihimäki, U. Ulfvarson, Eds., Alan R. Liss, N.Y., 1986, pp. 123-126. E. Browning, Toxicity and Metabolism of Industrial Solvents, Elsevier Publishing Co., N. Y., 1965. R.E. Gosselin, R.P. Smith, H.E. Hodge, Clinical Toxicology of Commercial Products, Williams and Wilkins, Baltimore, 1984. E.L. Baker, Ann. Rev. Public Health, 9, 233 (1988). P. Grasso, M. Sharratt, D. M., Davies, D. Irvine, Food Chem. Toxicol., 22, 819 (1984). R.B. Dick, Neurotoxicol. Teratol., 10, 39 (1988). W.K. Anger, Neurotoxicology, 11, 627 (1990). P. Arlien-Soborg, L. Hansen, O. Ladefoged, L. Simonsen, Neurotoxicol. Teratol., 14, 81 (1992). WHO, Organic solvents and the central nervous system, WHO European Office Copenhagen, (1985). J.F.Fowler, Dermatology, 10, 216 (1998). K.E. Andersen in Safety and Health Aspects of Organic Solvents, V. Riihimäki, U. Ulfvarson, Eds., Alan R. Liss, N. Y., 1986, pp. 133-138, 1986. C.G.T. Mathias, Occup. Med. State of the Art Rev., 1, 205 (1986). M. Hodgson, A.E. Heyl, D.H. Van Thiel, Arch. Intern. Med., 149, 1793 (1989): IARC, IARC Monographs on the evaluation of carcinogenic risks to humans. Some organic solvents, resin monomers and related compounds, pigments and occupational exposures in paint manufacture and painting, WHO, 47, IARC, Lyon, 1989. IARC, IARC Monographs on the evaluation of carcinogenic risks to humans. Dry cleaning, some chlorinated solvents and other industrial chemicals, WHO, 63, IARC, Lyon, 1995.
20.1 Toxicokinetics, toxicodynamics, and toxicology
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66
1325
IARC, IARC Monographs on the evaluation of carcinogenic risks to humans. Re-evaluation of some organic chemicals, hydrazine and hydrogen peroxide. WHO, 71, IARC, Lyon, 1999. DFG, Gesundheitsschädliche Arbeitsstoffe. Toxikologisch-arbeitsmedizinische Begründungen von MAK-Werten, VCH, Weinheim, 1999. R.R. Lauwerys, A. Bernard, C. Viau, J.P. Buchet, Scand. J. Work Environ. Health, 11, 83 (1985). E.A. Lock, Crit. Rev. Toxicol., 19, 23 (1988). N.A. Nelson, T.G. Robins, F.K. Port, Am. J. Nephrol., 10, 10 (1990). H.J. Mason, A.J. Stevenson, G.M. Bell, Ren. Fail., 21, 413 (1999). O. Ladefoged, H.R. Lam, G. Ostergaard, E.V. Hansen, U. Hass, S.P. Lund, L. Simonsen, Neurotoxicology, 19, 721 (1998). A.M. Seppalainen, Crit. Rev. Toxicol., 18, 245 (1988). P.S. Spencer, H.H. Schaumburg, Scand. J. Work Environ. Health, 11, 53 (1985). L.H. Welch, S.M. Schrader, T.W. Turner, M.R. Cullen, Am. J. Ind. Med., 14, 509 (1988). I.J Yu, J.Y. Lee, Y.H. Chung, K.J. Kim, J.H. Han, G.Y. Cha, W.G. Chung, Y.M. Cha, J.D. Park, Y.M. Lee, Y.H. Moon, Toxicol. Letters, 109, 11 (1999). W.G. Chung, I.J. Yu, C.S. Park, K.H. Lee, H.K. Roh, Y.N. Cha, Toxicol. Letters, 104, 143 (1999). J.C.Selner, Regul. Toxicol. Pharmacol., 24, 87 (1996). K.E. Andersen, S.C. Rastogi, L. Carlsen, Acta Derm. Venereol., 76, 136 (1996). G. Mancuso, M. Reggiani, R.M. Berdodini, Contact Dermatitis, 34, 17 (1996). H.H. Tau, M. Tsu Li-Chan, C.L. Goh, Am. J. Contact. Dermat., 8, 210 (1997). T. Chida, T. Uehata, Sangyo Igaku, 29, 358 (1987). P.J. Coenraads, S.C. Foo, W.O. Phoon, K.C. Lun, Contact Dermatitis, 12, 155 (1985). J.T. Lear, A.H. Heagerty, B.B. Tan, A.G. Smith, J.S. English, Contact Dermatitis, 35, 169 (1996). E.L. Sainio, K. Engstrom, M.L. HenriksEckerman, L. Kanerva, Contact Dermatitis, 37, 155 (1997). A.T. Karlberg, A. DoomsGoossens, A., Contact Dermatitis, 36, 201 (1997). D.M. Rubel, S. Freeman, I.A. Southwell, Australas J. Dermatol., 39, 244 (1998). M. McEwan, W.H. McFarlane-Smith, Clin. Exp. Allergy, 28, 332 (1998). G. Angelini, L. Rigano, C. Foti, M. Grandolfo, G.A. Vena, D. Bonamonte, L. Soleo, A.A. Scorpiniti, A.A., Contact Dermatitis, 35, 11 (1996). J. van de Walter, S.A. Jimenez, M.E. Gershwin, Int. Rev. Immunol., 12, 201 (1995). D. Eis, Allergologie, 22, 538 (1999). J.J. Powell, J. Van-de-Water, M.E. Gershwin, Environ. Health Perspect., 197, 667 (1999). F.R. Cassee, Crit. Rev. Toxicol., 28, 73 (1998). Römpp, Lexikon Chemie, J. Falbe, M. Regitz, Eds., Thieme, Stuttgart (1999). GISBAU, WINGIS, Bau-Berufsgenossenschaften (Professional Associations of the Building Industry in Germany), 1999. U. S. Department of Health and Human Services, National Toxicology Program, The 8th Report on Carcinogens, 1998.
1326
Jenni A Ogden
20.2 COGNITIVE AND PSYCHOSOCIAL OUTCOME OF CHRONIC OCCUPATIONAL SOLVENT NEUROTOXICITY Jenni A Ogden Department of Psychology, University of Auckland, Auckland, New Zealand
20.2.1 INTRODUCTION Many organic solvents used in industry are neurotoxic, and may lead to a range of largely irreversible cognitive and psychological or psychiatric impairments in workers who are exposed over long periods of time, or who have had a peak exposure (an episode in which they were briefly exposed to a larger than normal level of solvent). The most vulnerable workers are those who work in the spray painting, boat building, printing, textile, plastic, agricultural and pharmaceutical industries. Often self-employed workers or those in small businesses are more at risk because the safety measures they take are not as closely monitored, and peer pressure to use safety equipment even when it is unwieldy, restrictive or expensive, is unlikely to be as strong as in large workshops. In addition they may be less well educated regarding the neurotoxic effects of the solvents they work with. The great majority of workers diagnosed with OSN are men, presumably because men make up the bulk of the workforce in trades and industries that use neurotoxic solvents. The chronic, and often slow and insidious effects of occupational solvent neurotoxicity (OSN) include psychological and psychiatric symptoms, impairments in cognitive functioning, and negative psychosocial consequences. The Scandinavian countries are the research leaders in this field, and in recent years health professionals and industries in the United States and other major industrialized countries have become increasingly aware of the debilitating symptoms that can affect workers exposed to neurotoxins over a long time.1 There have been allegations that OSN is often over-diagnosed by health professionals who are zealous believers, and that a significant number of workers who complain of OSN symptoms are malingering in the hope of obtaining financial compensation.2 While these allegations almost certainly have some credibility, especially in countries such as the USA, where civil litigation has resulted in large settlements and the existence of OSN is now enshrined in legal precedent,2 there is ample evidence that the OSN syndrome does exist and is a major health problem for workers in industries that utilize neurotoxic solvents. A number of research studies establishing the existence of OSN have been conducted in countries where there is only limited, if any, financial gain to be made from diagnosing OSN, including Hong Kong3 and New Zealand.4 One of the primary difficulties researchers and health professionals face when trying to ensure that the symptoms they are observing are indeed the result of OSN, lies in the fact that the neurological damage resulting from chronic neurotoxin exposure tends to be diffuse, or may, for example, involve a neurotransmitter imbalance. It is therefore unlikely to be evident on a Computerized Tomograph (CT) or Magnetic Resonance Image (MRI) of the brain. A neurological examination is rarely helpful,5 and in many cases the psychological and cognitive impairments are the only clear indicators of neurotoxicity. A neuropsychological assessment which utilizes a range of tests to assess cognitive abilities including attention, concentration, psychomotor speed, memory and visuospatial skills, along with a psychological interview or questionnaire assessing depression, irritability, mo-
20.2 Cognitive and psychosocial outcome
1327
tivation and fatigue, thus plays a major role in diagnosing chronic OSN.6 The World Health Organization (WHO) and the Nordic and New Zealand Governments all require that a neuropsychological assessment be used in the diagnosis of solvent neurotoxicity.7-9 Many victims of OSN do not realize that their chronic fatigue, irritability, poor memory and other problems may be associated with the solvents in their workplace, and by the time they seek help from their doctor, psychologist or marriage guidance counsellor, the OSN symptoms are likely to be compounded and masked by other work and relationship problems (themselves possibly a consequence of the OSN symptoms).6 Identification of OSN as the primary cause of the problems is therefore even more difficult, and proving cause and effect usually impossible. That OSN is a significant cause of the person’s problems, can, however, often be established beyond reasonable doubt, provided that some guidelines are followed. The individual must clearly have been exposed to neurotoxins over a long period (usually set, rather arbitrarily, at 10 years or more of occupational exposure), or have suffered a peak exposure. Other major contributors to neurological impairment should be excluded (e.g., significant traumatic brain injury, or alcohol addiction), there should be no evidence of malingering, and the pattern of cognitive impairments and psychological symptoms should be typical of OSN. 20.2.2 ACUTE SYMPTOMS OF SOLVENT NEUROTOXICITY Neurotoxic solvent exposure can result in some workers experiencing nausea, vomiting, loss of appetite, severe headaches, confusion, light-headedness and dermatitis. The solvent may be detectable on their breath and skin for hours and even days after they have left the solvent environment. Most of these symptoms resolve when they stop working with solvents but return when they come into contact with solvents again. Workers who suffer these acute symptoms do not necessarily go on to develop the chronic syndrome of OSN, perhaps in many cases because they are so disabled by the acute symptoms they stop working before irreversible damage occurs. Some workers who suffer acute symptoms do remain in the work environment, sometimes because of financial necessity, or because they do not realize the solvents are the cause of their problems.10 Some workers who develop a chronic OSN syndrome have suffered from acute symptoms, but others have not. The reason for these individual differences is not clear. 20.2.3 CATEGORIZATION OF OSN The 1985 International Solvent Workshop11 proposed three types of OSN, as follows: • Type 1 OSN: Characterized by subjective complaints of fatigue, irritability, depression and episodes of anxiety. No cognitive impairments are demonstrable on neuropsychological testing, and the psychological symptoms resolve on removal from the solvents. This is also known as the organic affective syndrome, or the neurasthenic syndrome. • Type 2 OSN: A more severe and chronic form than Type 1 in which many of the symptoms and cognitive impairments are thought to be irreversible when the worker is removed from the solvent environment. It is also known as mild toxic encephalopathy. Type 2 has been divided further into two sub-types based on psychological symptoms (Type 2A) and cognitive impairments (Type 2B). Type 2A sufferers have a range of symptoms which may include sustained personality and mood disturbances, fatigue, poor impulse control and poor motivation. Type 2B symptoms include poor concentration, impairments of new verbal and visual
1328
Jenni A Ogden
learning and memory, psychomotor slowing, and in more severe cases, executive (or frontal-lobe) impairments. These can include impoverished verbal fluency, difficulties with abstract thinking, and impairments in the ability to make plans and organize tasks logically. These cognitive symptoms must be demonstrable on neuropsychological tests following a solvent-free period. There is some research which indicates that this separation of Type 2 OSN into psychological and cognitive impairment profiles is largely unrealistic, as most workers with Type 2 OSN have symptoms of both types.10,12 Type 2 OSN is the primary focus of this section given its largely irreversible nature and its frequency in the workplace. • Type 3 OSN: This is the most severe form of OSN and signals an irreversible dementia with severe impairment across most cognitive and emotional domains. It is also known as severe toxic encephalopathy, and is fortunately rare in occupational situations. It is more likely to occur in long-term recreational solvent abusers. 20.2.4 ASSESSMENT OF OSN There have been a few studies reporting specific symptoms caused by a specific solvent. The widely used industrial solvent trichloroethylene (TCE), has, for example, been reported to result in severe agitated depression, sometimes accompanied by violent behaviors towards self and others.13 Toluene and TCE can cause peripheral neuropathy, and TCE can damage the trigeminal or fifth cranial nerve, resulting in a loss of sensation to the face, mouth and teeth.1 It is, however, rare to be able to pinpoint a specific solvent as the cause of specific cognitive or psychological symptoms, and most research on occupational solvent neurotoxicity has been carried out on workers exposed to a mixture of solvents. A core neuropsychological battery has been developed by the WHO/Nordic Council,8 and most other formal and informal batteries developed for the assessment of OSN include a similar range of tests, as these are the tests most sensitive to the common neuropsychological impairments of OSN.9,14,15,16 Specific tests used in these batteries will not be listed here, as neuropsychologists qualified to administer, score, and interpret these tests can find specialist information in texts written on OSN assessment.1 The assessment of OSN may be initiated if a worker receives a poor score on a screening workplace questionnaire designed to assess the frequency of self-reported problems such as irritability and poor memory.12 In other cases the worker comes to the attention of a health professional because of interpersonal or memory problems which concern the worker, family, or work colleagues. In New Zealand, in 1993 the Occupational Safety and Health Service (OSH) of the Government Department of Labour, established a panel of experts to develop national guidelines for the diagnosis of OSN.4,9 Workers who are diagnosed as suffering from OSN are registered as part of the Notifiable Occupational Disease System. Other panels provide a similar function for other occupational diseases such as asthma and asbestos-related disorders. Following is a description of the procedures for diagnosing OSN that the New Zealand panel has developed and tested since 1993.4,9 Individuals, industries, industrial health workers, or general practitioners can notify a possible case of OSN to the panel. Occupational hygienists then attempt to measure the types and levels of solvents the worker has been potentially exposed to throughout his or her working life. This is easier if the worker is currently in the solvent environment, but estimates only can be made of solvent levels in previous workplaces, and of the workplace and worker’s appropriate use of protective equipment over the years. If there is reason to suspect that the worker has been exposed to neurotoxic solvents for 10 years or more, or has suf-
20.2 Cognitive and psychosocial outcome
1329
fered peak exposures, the occupational physician will examine and interview the worker (and where possible a close family member) and make an initial assessment regarding the worker’s symptoms. It is not uncommon at this interview stage for the worker, often a middle-aged tradesman not accustomed to talking about his cognitive or emotional problems, to break down in tears. Most health professionals experienced in assessing OSN are in no doubt that it is a real syndrome with devastating consequences for the worker and family.6 If the symptom complex generally fits with that typical of OSN, the symptoms are significant enough to be causing the worker or his family concern, and other possible causes have been explored and considered to be unlikely as the primary cause of the problems, the worker will proceed to a neuropsychological assessment. Whenever possible, this should take place following two or more weeks away from solvents. This is again a somewhat arbitrary time period, arrived at in an attempt to find a balance between the real time it takes for any acute effects of a mixture and range of solvents to resolve, and the amount of time (usually unpaid) an undiagnosed worker is willing or able to take away from his workplace. The assessment usually commences with a psychological assessment, which may include both an interview and standard questionnaires on mood, fatigue levels, motivation, memory problems in daily life and so on. Often, with the worker’s permission, information is also obtained from family members and work colleagues. Not only does this allow an assessment of the problems the worker is experiencing at work and at home, but also gives the neuropsychologist some idea of the time course of these problems. Other possible confounding psychosocial factors are checked out at this point. Whilst factors such as a high use of alcohol, or a series of minor head injuries whilst playing sport 10 years previously, or a recent marriage breakup, may not negate the possibility of the worker being diagnosed as suffering from OSN, clearly these factors must be taken into account in making the diagnosis and the confidence that can be placed in that diagnosis, as well as when designing an intervention or rehabilitation program for the worker. Having ascertained that the worker’s exposure levels and psychological and subjective cognitive symptoms (e.g., complaints of memory problems) meet the criteria for possible OSN, a battery of carefully chosen neuropsychological tests is then given. This is often scheduled for a later session, given the distress that the worker may have expressed during the interview, and the high fatigue levels that are a common consequence of OSN. This battery should include one or more tests which can, along with education and occupational history, provide an estimate of the worker’s cognitive ability level prior to working with solvents. Also included should be some tests which one would not expect to be impaired by solvents, such as well-established vocabulary (meanings of words). Tests which are included because of their sensitivity to OSN symptoms include tests of concentration and attention, new verbal and visuospatial learning and memorizing (old, well-established memories are rarely impaired), reaction time, psychomotor speed, and planning, organizational and abstraction abilities. If the pattern of spared and impaired psychological and neuropsychological test results is typical of OSN, and other factors can be ruled out as the primary cause of this profile, the worker will be diagnosed as having OSN.6,10 This pattern analysis provides one way of guarding against malingering, as the worker does not know which tests he or she should remain unimpaired on and which are commonly impaired following OSN. In addition, on many tests, it is very difficult or impossible for the malingerer to perform in a way that is consistent with true organic impairment, even if he or she has been coached on how to per-
1330
Jenni A Ogden
form poorly on the tests. For example, if an individual was unable to remember any new visual stimuli (an extremely rare condition), when given a memory test where the worker is shown 50 photographs of unknown faces, and is then shown fifty pairs of faces and must choose from each pair the face which he or she has previously seen, he or she should obtain a score of approximately 50% (chance level) correct. If the score was considerably worse than that, malingering or exaggerating might reasonably be suspected. Tests which measure reaction or response times for increasingly complex tasks are also difficult to malinger successfully on as humans are not good at estimating response times in milliseconds, or even seconds. A diagnosis of Type 2 OSN is based on score deficits (measured by the number of Standard Deviations (SD) below the worker’s estimated premorbid ability level) on those tests commonly impaired by OSN. At least three neuropsychological test scores must fall more than 1 SD below the scores expected for that worker to be categorized as mild Type 2 OSN, three test scores below 2 SDs for moderate Type 2 OSN, and three or more test scores more than 3 SDs below the expected levels for moderate-severe Type 2 OSN.4 The presence and severity of typical psychological symptoms are also taken into account, and in clear cases in which either psychological or cognitive symptoms are very dominant, this information informs a decision regarding Type 2A or Type 2B OSN. Whilst psychological symptoms are the reason most workers come to the attention of health professionals, because of the difficulty of measuring the severity of these symptoms and of attributing them to a neurological syndrome, only workers who demonstrate neuropsychological impairments on testing are positively diagnosed with OSN. The New Zealand experience has, however, demonstrated that the vast majority of workers with significant solvent exposure histories and severe psychological problems do demonstrate neuropsychological impairments, and vice versa.10 20.2.5 DO THE SYMPTOMS OF TYPE 2 OSN RESOLVE? Occasionally after an extended period away from solvents (perhaps 6 months to a year), the worker’s psychological symptoms resolve, and on re-testing it is found that his or her neuropsychological impairments have also resolved. In these cases the classification is changed to Type 1 OSN (resolved). A recent New Zealand study re-assessed 21 men with confirmed cases of OSN 6 to 41 (mean 27) months after ceasing exposure.17 An exposure score was calculated for each worker by using the formula AxBxC, where A = years of solvent exposure, B = a weighting for the occupational group (where boat builders, spray painters and floorlayers had the highest weighting of 3), and C = a weighting reflecting the lack of safety precautions taken by the worker relative to other workers in the same job. Neuropsychological and psychological symptoms at the initial and follow-up assessments were categorized as mild, moderate or moderate-severe (using the system described above) by a neuropsychologist blind to the men’s initial diagnosis or exposure history. Twelve men (57%) showed no improvement (or in one case a slight worsening) on cognitive and psychological assessment. Seven men showed some improvement on cognitive tests (but not to “normal” levels), only three of whom also improved on psychological assessment. A further two men showed an improvement in psychological functioning only. Men given a more severe OSN diagnosis at their initial assessment were more likely to improve than men with milder symptoms at the time of their first assessment. Possible explanations for this include the likelihood that some of the more severe symptoms on initial assessment were exacerbated by the lingering effects of acute solvent exposure, or that those with mild OSN were
20.2 Cognitive and psychosocial outcome
1331
misdiagnosed and their “symptoms” were due to some other cause or were “normal” for them, or that there were psychosocial difficulties present at the first assessment which exacerbated the organic symptoms and resolved with rehabilitation. The disturbing message is, however, that the symptoms of Type 2 OSN are often persistent, and in these cases probably permanent. Even in those individuals where improvement occurs, their symptoms rarely resolve completely. There was no association between improvement on neuropsychological tests and either time between the two assessments or total time away from solvents. There was no correlation between the exposure score and severity at diagnosis or extent of recovery, and there was no association between a past history of peak exposures and either severity at initial diagnosis or change on neuropsychological assessment. A recent review18 of studies looking at whether the degree of impairment is related to the dose severity concludes that although several studies have demonstrated significant dose response relationships, there are disturbing inconsistencies, with some studies showing no relationship,19,20 and one study showing a dose response relationship in painters with levels of exposure considerably lower than the negative studies.21 Methodological problems and differences and different research populations probably account for these inconsistent findings, and more research is clearly required. 20.2.6 INDIVIDUAL DIFFERENCES IN SUSCEPTIBILITY TO OSN One possible reason for the inconsistent findings both across and within studies examining the relationship between exposure levels and OSN symptoms, may be that individuals have different susceptibilities to solvents. It is not uncommon to diagnose one worker with moderate Type 2 OSN, yet find no symptoms or serious complaints whatsoever in his workmate who has worked by his side in the same spray painting workshop for twenty years. On closer assessment it may be discovered that the affected worker sustained a number of minor head injuries in his younger football-playing days, or has smoked a marijuana joint every weekend for the past 20 years. Subclinical neuronal damage caused by previous insults, or even by normal aging, may make an individual more susceptible to OSN. Another possibility is that some people are biologically, and even genetically more susceptible to solvents. In this sense, OSN can be likened to the post-concussional syndrome following a mild to moderate traumatic brain injury.6 Not only are the psychological and neuropsychological symptoms very similar, but for reasons which cannot be explained simply by lifestyle differences or malingering, individuals appear to differ widely regarding their susceptibility to developing a post-concussional syndrome. In illustration of this, a recent study reports varying outcomes from apparently equivalent head injuries in a group of athletes.22 20.2.7 PSYCHOSOCIAL CONSEQUENCES OF OSN, AND REHABILITATION The common psychological and physical symptoms of OSN of fatigue, irritability, depression, sometimes aggression and violence, headaches, and hypersensitivity to noise and alcohol, along with memory difficulties, poor concentration, poor motivation, and slowed thinking, are a recipe for disaster in interpersonal relationships. Thus it is not uncommon for workers to be diagnosed and treated first for a psychiatric disorder (especially clinical depression) and for their marriages to break up, before they are even suspected of having OSN.6,10 Once OSN is diagnosed, the prospect of losing their job is a grim one for most victims, most of whom are tradesmen in middle age or older who may have difficulty obtaining
1332
Jenni A Ogden
or even training for another occupation, especially given their memory, motivation, and concentration problems. Rehabilitation programmes6,10 begin with psychoeducation for the worker and his family about the effects of solvents and the importance of protecting himself from exposure in the future. Family members can be taught strategies to reduce the stress on the victim, such as encouraging him to have a rest in the afternoon, and limit his alcohol intake, and by helping him avoid noisy environments such as parties and the family room in the early evening when children are irritable and hungry. Counseling and therapy for the victim and family can be helpful in assisting them to vent their anger at the unfairness of their situation, grieve for their lifestyle and cognitive abilities lost, and come to terms with a “different” person (whose memory may be permanently impaired, and concentration span and motivation lowered). Financial and practical assistance is more often than not of extreme importance, as it is difficult to find the motivation to work on one’s psychological and family problems when one is worried about feeding and clothing the children. Antidepressant or anti-anxiety medications may be of assistance in severe cases of mood disorder. In some cases both the neurological damage and the psychological overlay can result in aggressive and violent behaviors not typical of the worker in his younger days. In these cases it is important to first attend to the safety of family members, and then to try and involve the worker in anger management programs, or other therapy with the goal of helping him understand how to control his aggressive or violent behaviors. Similarly, alcohol may be a problem given that it seems likely that neurotoxic solvents damage the pre-frontal lobes, thus resulting in a heightened susceptibility to intoxication. A rehabilitation program aimed at reducing alcohol intake will be important in this case. Vocational counseling and training are important not only to guide the worker towards a new occupation where solvents are preferably absent, or where protection from solvent exposure is good, but it is also important for the victim’s self-esteem and mood. Unfortunately, in many countries where unemployment is high, the prospects of finding a satisfying new career in middle-age are bleak. The task for the rehabilitation therapist in these sad cases is to encourage the worker to take up new hobbies and recreational activities, to spend more quality time with family and friends, and to try and live on a sickness benefit or unemployment benefit without losing self-respect. REFERENCES 1 2 3 4. 5 6 7
8 9 10
D.E.Hartman, Neuropsychological Toxicology: Identification And Assessment of Human Neurotoxic Syndromes. 2nd Ed. Plenum Press, New York, 1995. P.R.Lees-Haley, and C.W.Williams, J.Clin.Psychol., 53, 699-712 (1997). T.P.Ng, S.G.Ong, W.K.Lam, and G.M.Jones, Arch.Environ, Health, 12, 661-664 (1990). E.W.Dryson, and J.A.Ogden, N.Z. Med.J., 111, 425-427 (1998). J.Juntunen in Neurobehavioral Methods In Occupational Health, R.Gilioli, M.G.Cassitto, and V.Foa, Eds., Pergamon Press, Oxford, 1983, pp. 3-10 J.A.Ogden, Fractured Minds. A Case-Study Approach To Clinical Neuropsychology. Oxford University Press, New York, 1996, pp. 174-184; 199-213. World Health Organization and Commission of the European Communities, Environmental Health Document 6: Neurobehavioral Methods In Occupational and Environmental Health: Symposium Report. WHO Regional Office for Europe and Commission of the European Communities, Copenhagen, 1985. World Health Organization, Nordic Council of Ministers, Organic Solvents And The Central Nervous System, EH5, WHO, Copenhagen, 1985. E.W.Dryson, and J.A.Ogden, Chronic Organic Solvent Neurotoxicty: Diagnostic Criteria. Department of Labour, Wellington, 1992. J.A.Ogden, N.Z. J. Psychol., 22, 82-93 (1993).
20.3 Pregnancy outcome following solvent exposure
11 12 13 14 15 16 17 18 19 20 21 22
1333
E.L.Baker, and A.M.Seppalainen, Neurotoxicology, 7, 43-56 (1986). T.L.Pauling, and J.A.Ogden, Int. J. Occup. Environ. Health, 2, 286-293 (1996). R.F.White, R.G.Feldman, and P.H.Travers, Clin. Neuropharm., 13, 392-412 (1990). H.Hanninen in Neurobehavioral Methods In Occupational Health, R.Gilioli, M.G.Cassitto, and V.Foa, Eds., Pergamon Press, Oxford, 1983, pp. 123-129 L.A.Morrow, C.M.Ryan, M.J.Hodgson, and N.Robin, J Nerv. Ment. Dis., 179, 540-545, (1991). C.M.Ryan, L.A.Morrow, E.J. Bromet, J. Clin. Exp. Neuropsychol., 9, 665-679, (1987). E.W.Dryson, and J.A.Ogden, Organic solvent induced chronic toxic encephalopathy: Extent of recovery and associated factors following cessation of exposure. Submitted. S.Mikkelsen, Environ. Res., 73, 101-112, (1997). J.Hooisma, H.Hanninen, H.H.Emmen, and B.M.Kulig, Neurotoxicol. Teratol., 15, 397-406, (1993). A.Spurgeon, D.C.Glass, I.A.Calvert, M.Cunningham-Hill, and J.M.Harrington, J. Occup. Environ. Med., 51, 626-630, (1994). M.L.Bleecker, K.I.Bolla, J.Agnew, B.S.Schwartz, and D.P.Ford, Am. J. Ind. Med., 19, 715-728, (1991). S.N.Macciocchi, J.T.Barth, and L.M.Littlefield, Clin. Sports Med., 17, 27-36, (1998).
20.3 PREGNANCY OUTCOME FOLLOWING MATERNAL ORGANIC SOLVENT EXPOSURE Kristen I. McMartin and Gideon Koren The Motherisk Program, Division of Clinical Pharmacology and Toxicology, Hospital for Sick Children, Toronto, Canada
20.3.1 INTRODUCTION Organic solvents are a structurally diverse group of low molecular weight liquids that are able to dissolve other organic substances.1 Chemicals in the solvent class include aliphatic hydrocarbons, aromatic hydrocarbons, halogenated hydrocarbons, aliphatic alcohols, glycols, and glycol ethers. Fuels are a mixture of various hydrocarbons. They are generally ubiquitous in industrialized society, both at work and at the home. They may be encountered as individual agents or in complex mixtures such as gasoline. Incidental exposures may include vapors from gasoline, lighter fluid, spot removers, aerosol sprays and paints. These short duration and low level exposures may often go undetected. More serious exposures occur mainly in the industrial or laboratory settings during manufacturing and processing operations such as dry cleaning, regular working with paint removers, thinners, floor and tile cleaners, glue and as laboratory reagents. Gasoline sniffing or glue sniffing, albeit not occurring in the occupational setting, is another source of exposure to organic solvents during pregnancy. Counseling pregnant women who are occupationally exposed to numerous chemicals (mostly organic solvents) is difficult because it is hard to estimate the predominant chemicals and their by-products. Even after identifying the more toxic agents, it is still difficult to assess the circumstances of exposure as for many chemicals one can measure neither airborne nor blood levels. Smelling the odor of organic solvents is not indicative of a significant exposure as the olfactory nerve can detect levels as low as several parts per million which are not necessarily associated with toxicity. As an example, the odor threshold of toluene is 0.8 parts per million whereas the TLV-TWA (threshold limit value-time weighted average) is 50 parts per million. In addition, reproductive information on many individual solvents is at best sparse, either limited to animal studies or nonexistent.
1334
Kristen I. McMartin, Gideon Koren
Many organic solvents are teratogenic and embryotoxic in laboratory animals depending on the specific solvent, dose, route of administration and particular animal species.1 The various malformations described include hydrocephaly, exencephaly, skeletal defects, cardiovascular abnormalities and blood changes. Also, some studies suggest poor fetal development and neurodevelopmental deficits. In a portion of these studies exposure levels were high enough to induce maternal toxicity. Organic solvents are a diverse, complex group and because exposure usually involves more than one agent and different circumstances, adequate human epidemiological studies are difficult to interpret. Many studies are subject to recall and response bias and are not always controlled for other risk factors such as age, smoking, ethanol, and concurrent drug ingestion. It is hard to prove or quantify the suspicion that organic solvents are a reproductive hazard. One may even expect that a ubiquitous exposure to solvents would by chance alone be associated with an increase in birth defects or spontaneous abortions, which may differ from one study to another. While fetal toxicity is biologically sensible in cases of intoxicated mothers, evidence of fetal damage from levels that are not toxic to the mother is scanty, inconsistent or missing. This chapter will review the reproductive toxicology of organic solvents with particular focus on exposure during pregnancy. Firstly, examples of animal studies with regard to three organic solvents will be discussed. This will be followed by information obtained from human studies including: a meta-analysis of pregnancy outcome following maternal organic solvent exposure; results from the first prospective study by the Motherisk Program at the Hospital for Sick Children on gestational exposure during pregnancy; and finally, a proactive approach for the evaluation of fetal safety in chemical industries. 20.3.2 ANIMAL STUDIES There are numerous experimental studies that examine the reproductive effects of organic solvents in animals. The reproductive effects of maternal organic solvent exposure will be summarized using three organic solvents as examples. The solvents discussed will be benzene, toluene and tetrachloroethylene. Benzene Watanabe and Yoshida2 were the first to claim teratogenic effects of benzene after administration during organogenesis only. Groups of 15 mice were given single subcutaneous injections of 3 ml benzene/kg on one of days 11-15 of pregnancy. This dose caused leukopenia lasting 24-48 hours but had no effect on body weight in the dams. Litter size ranged form a average of 6.5-8.5 in the 4 treatment groups. Malformations were seen in most treated groups; cleft palate occurred in 5.5% of fetuses exposed on day 13 and in 1.0% of fetuses exposed on day 14 and agnathia or micrognathia was seen on 0.9%, 2.4% and 1.0% of fetuses exposed on days 11, 13 and 14 respectively. Extra 14th ribs were seen in 10-16% of fetuses in all treated groups. Fetuses from 5 dams treated on day 15 had no malformations but 24% had extra 14th ribs. In the absence of any control data it is not known if these represent significant increases in malformations and anomaly rates. Extra 14th ribs for example, can be a common skeletal variant in some strains of mice and rats.9 Matsumoto et al.3 have given groups of 8-11 mice subcutaneous injections of 0, 2, or 4 ml of benzene/kg on days 8 and 9 or 12 and 13 of pregnancy. Fetuses were examined externally and for skeletal defects only; internal soft tissues were not examined. They claim that fetal weight was significantly decreased in both groups given 4 ml/kg and placental weight significantly reduced in those given 4 ml/kg on days 12 and 13 of pregnancy. However, re-
20.3 Pregnancy outcome following solvent exposure
1335
working of the data shows p values of >0.4 in all cases.9 Sporadic malformations (cleft palate and open eye) did not differ significantly between treated and control groups, neither did the incidence of dead or resorbed embryos and fetuses. A small degree of retarded ossification was seen in fetuses from dams given 4 ml/kg. Nawrot and Staples4 investigated the effects of oral administration by gavage of 0.3, 0.5 or 1.0 ml/kg on days 6-15 of pregnancy or 1.0 ml/kg on days 12-15 of pregnancy in the mouse. After dosing on days 6-15, 0.5 and 1.0 ml/kg caused some maternal mortality and embryolethality. Fetal weight was significantly reduced at all 3 dose levels but no increase in malformations was seen. There were similar findings after dosing on days 12-15 except that resorptions occurred later in gestation. The study is reported in abstract only and no further details are given. Murray et al.5 exposed groups of 35-37 mice to 0 or 500 ppm benzene for 7 hr/day on days 6-15 of pregnancy. Acceptable teratological methods were used.9 There was no evidence of maternal toxicity. There were no effects on implants/dam, live fetuses/dam, resorptions/dam or malformation rates. Fetal body weight was significantly reduced and delayed ossification significantly increased in fetuses from the benzene group. Iwanaga et al.6 demonstrated an increased postnatal susceptibility to benzene toxicity in mice exposed prenatally to benzene by injection of the dams with 4 ml benzene/kg on day 9 or 12 of gestation. At 10 weeks of age the offspring were injected with 5 daily doses of 0.1 ml benzene/kg and the effects on erythrocytes, leukocytes, body weight, thymus and spleen were more marked than in non-prenatally exposed controls. There have been several inhalational studies on benzene in the rat. In an unpublished study summarized by Murray et al.,5 teratogenic effects were observed at 500 ppm when rats were exposed to 0, 10, 50 or 500 ppm benzene for 7 hr/day on days 6-15 of pregnancy and a low incidence of exencephaly, kinked ribs and abnormal ossification of the forepaws was noted at 500 ppm. In another unpublished study quoted by Murray et al.5 no teratogenicity but increased embryoloethality was seen after exposure to 10 or 40 ppm for 6 hours/day on days 6-15 of pregnancy in the rat. Hudak and Ungvary7 exposed groups of 19-26 rats to 0 or 313 ppm benzene for 24 hours/day on days 9-14 of pregnancy. Acceptable teratological methods were used.9 There was no maternal mortality but maternal weight gain was significantly reduced. There were no significant effects on live fetuses/dam, resorbed or dead fetuses/dam or malformation rate. Mean fetal weight was significantly reduced and retarded ossification, abnormal fusion of sternebrae and extra ribs were all significantly increased in the benzene-exposed group. Green et al.8 exposed groups of 14-18 rats to 100, 300 or 2200 ppm benzene for 6 hours/day on days 6-15 of pregnancy, each benzene-exposed group having a concurrent 0 ppm control group. Maternal weight gain was significantly reduced in the 2200 ppm group, but not at lower exposure levels. There were no significant effects on implants/dam, live fetuses/dam, resorptions/dam or malformation rates. There was a significant 10% reduction in fetal weight in the 2200 ppm benzene group and skeletal anomalies were sporadically increased in benzene-exposed groups (missing sternebrae at 100 ppm, delayed ossification of sternebrae in female offspring only at 300 ppm and 2200 ppm and missing sternebrae at 2200 ppm). The authors suggest the higher number of affected female fetuses is in accordance with other observations on the increased susceptibility of females to benzene toxic-
1336
Kristen I. McMartin, Gideon Koren
ity.9 In addition, they observed a non-significant low incidence of hemorrhages in all 3 benzene-exposed groups which were not seen in control fetuses. In conclusion, embryolethal and teratogenic effects are not seen even at maternally toxic doses but significant fetotoxicity in terms of reduced body weight sometimes accompanied by increases in skeletal variants and delayed ossification is seen at doses which are not necessarily toxic to the dam. The absence of any such effects in a large number of adequately conducted studies reported in full suggests these observations may be of no biological significance. The role that benzene-induced maternal anemia may play in any adverse effects on the offspring is not known.9 Toluene Euler10 exposed mice to a mixture of toluene and trichloroethylene similar to that which has been used in the soling of shoes. The mixture was composed of 32 ppm (120 mg/m3) toluene and 64 ppm (340 mg/m3) trichloroethylene, equivalent to inhaling 157 mg/kg toluene and 406 mg/kg trichloroethylene in the mice. They inhaled the mixture for 10 days before mating or during part or the whole of pregnancy. Differences were noted between treated and control groups in pregnancy rates, length of pregnancy, damaged embryos, birth weights and neonatal mortality but the direction and magnitude of these differences is not stated. No groups were exposed to toluene alone. Nawrot and Staples4 gave mice 0.3, 0.5, or 1.0 ml toluene/kg orally by gavage on days 6-15 of pregnancy or 1.0 ml/kg on days 12-15 of pregnancy. There was no maternal toxicity except a decrease in maternal weight gain in those dosed on days 12-15. There was a significant increase in embrylolethality at all 3 dose levels and a significant reduction in fetal weight in the 0.5 and 1.0 ml/kg groups after dosing on days 6-15. Those dosed with 1.0 ml/kg on days 6-15 had a significant increase in numbers of fetuses with cleft palate which was not simply due to general growth retardation. Treatment on days 12-15 only had no adverse effects on the offspring. The study is reported in abstract only and no further details are given. Teratological investigations on inhaled toluene in mice and rats have been carried out by Hudak et al.7 Mice were exposed to 0, 133 or 399 ppm (500 or 1500 mg/m3) toluene for 24 hr/day on days 6-13 of pregnancy. In the high dose group all 15 exposed dams died within the first 24 hr of exposure. No maternal deaths occurred in the 11 mice exposed to 133 ppm and there were no effects on implants/dam, live fetuses/dam, dead and resorbed fetuses/dam, malformations or anomaly rates, but fetal weight was significantly reduced by 10% in comparison with controls. It is not stated whether 133 ppm had any effect on maternal weight gain.7 In conclusion, similar to benzene, toluene does not appear to be teratogenic. It is fetotoxic, causing a reduction in fetal weight in mice and rats and retarded ossification and some increase in skeletal anomalies in rats at doses that are below those toxic to the dam as well as at toxic doses.9 Embryolethality has also been seen with inhalation of very high concentrations lethal to some of the dams or following oral administration of non-toxic doses.9 Tetrachloroethylene Schwetz et al.11 exposed rats and mice to 300 ppm tetrachloroethylene for 7 h/day on days 6-15 of pregnancy. The dams were killed just before term and the fetuses examined by acceptable teratological methods but results are given on a per litter basis only. The number of treated animals in each case was 17 and the number of controls (air exposed) 30 for both rat and mouse studies.
20.3 Pregnancy outcome following solvent exposure
1337
Effects of tetrachloroethylene on the dams varied between species.11 In the mouse relative liver weight was significantly increased and the absolute liver weight increased but not significantly and with no effect on maternal body weight. In the rat there was a non-significant decrease in absolute and relative liver weights and a significant 4-5% decrease in mean body weight. Food consumption was unaffected. Effects on the embryo and fetus also differed.11 In the mouse there was no effect on implantation sites, live fetuses or resorption rates but mean fetal weight was significantly reduced, 59% of litters containing runts (weight less than 3 standard deviations below the mean) compared with 38% of control litters. Whereas in the rat, resorption rate was significantly increased from 4% in controls to 9% in the exposed group, while fetal body was unaffected (mean slightly higher than controls). In the mouse, examination for anomalies revealed an increase in delayed ossification of the skull bones (significant) and of the sternebrae (nonsignificant) as might be expected from the fetal weight data. There were also significant increases in the incidence of split sterenbrae and subcutaneous edema. No gross malformations were found. In the rat, gross malformations (short tail) were reported but the incidence did not differ significantly from that in controls. There were no other significant differences in soft tissue or skeletal abnormalities.11 The results of this study are difficult to assess, partly because no indication of the numbers of fetuses affected within affected litters is given and partly because of the uncertain nature of the “subcutaneous edema” reported.9,11 Exposure to tetrachloroethylene and the concurrent controls were part of a large study on four different solvents. The incidence of subcutaneous edema in the mouse ranged from 8-59% of litters affected which seems very high and while the incidence in the tetrachloroethylene group was highest at 59%, it was as high as 45% in nonconcurrent controls (27% in concurrent controls).11 In the rat, the incidence of this particular anomaly also varied enormously between groups from 0% (tetrachloroethylene group) to 28% (trichloroethylene group).11 It is therefore important to know how strict were the criteria for designation of “subcutaneous edema” and in particular whether the designation was made before or after fixing, subcutaneous edema being a common fixative artifact.9 However, the retardation of growth and ossification and the increased incidence of split sternebrae in fetal mice exposed to tetrachloroethylene were clear effects and in the absence of any effect on maternal body weight, suggest that tetrachloroethylene has some maternal hepatotoxicity but has no effect in the rat where there is no hepatotoxicity at 300 ppm.11 The results of a behavioral teratology study in the rat by Nelson et al. have been reported.12 Rats were exposed to 0 or 900 ppm tetrachloroethylene for 7 hours/day on days 7-13 or 14-20 of pregnancy (9-16 rats per group). The dams were affected by this level, showing reduced food consumption and lower weight gain during exposure but histopathological examination of the maternal liver and kidney in dams sacrificed on day 21 of pregnancy revealed no abnormalities.12 Postnatally, offspring were tested for olfaction, neuromuscular ability, exploratory and circadian activity, aversive and appetitive learning.12 There was evidence of impaired neuromuscular ability.12 Offspring from dams exposed on days 7-13 were poorer than controls in ascent of a wire mesh screen during the second week of life and were poorer than controls on a rotorod test on one of the 3 days tested in the fourth week of life.12 Offspring from dams exposed on days 14-20 performed less well in ascent of a wire mesh screen.
1338
Kristen I. McMartin, Gideon Koren
However, the latter group were consistently superior to controls on the rotorod later in development.12 Both exposed groups were generally more active in open field tests than controls but only those exposed on days 14-20 of gestation differed significantly from controls.12 Biochemical analyses of whole brain neurotransmitter levels showed no effects in newborns but significant reductions in acetylcholine levels at 21 days of age in both exposed groups of offspring and reduced dopamine levels at 21 days of age in those from dams exposed on days 7-13.12 There were no significant differences between exposed and control groups on any other of the tests.12 Exposure of offspring to 100 ppm on days 14-20 of gestation showed no significant differences from controls on any of the above behavioral tests.12 It was not stated whether neurotransmitter levels were measured in this low-dose group.9,12 In view of these results, suggesting some fetotoxicity in the mouse but not the rat at 300 ppm and postnatal effects in the rat at 900 ppm but not 100 ppm, there is a need for further studies at low levels between 900 and 100 ppm to establish a more accurate no-effect-level.9 20.3.3 PREGNANCY OUTCOME FOLLOWING MATERNAL ORGANIC SOLVENT EXPOSURE: A META-ANALYSIS OF EPIDEMIOLOGIC STUDIES [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Introduction Evidence of fetal damage or demise from organic solvent levels that are not toxic to the pregnant woman is inconsistent in the medical literature. A mathematical method has been previously developed and utilized to help overcome bias and arrive at a single overall value that describes the exposure-outcome relationship; namely, meta-analysis.15 The risk for major malformations and spontaneous abortion from maternal inhalational organic solvent exposure during pregnancy is summarized using meta-analysis.31 Besides being more objective than the traditional methods of literature review, it has the ability to pool research results from various studies thereby increasing the statistical strength/power of the analysis. This is especially useful in epidemiologic studies, such as cohort studies or case control studies since very often large numbers of subjects are required in order for any problem to be significantly addressed. This is particularly true for teratogenic studies where the frequencies of malformation are often very low. Methods A literature search was conducted to collect studies for the meta-analysis. Using Medline, Toxline and Dissertation Abstracts databases spanning 1966-1994, literature was identified concerning the problem in question. In addition, external colleagues were consulted (regarding unpublished studies) whose area of interest is in occupational exposure and reproductive toxicology. All references from the extracted papers and case reports were investigated. Standard textbooks containing summaries of teratogenicity data were consulted for further undetected references. Inclusion criteria consisted of human studies of any language which were 1) case control or cohort study in design; 2) included maternal inhalational, occupational, organic solvent exposure; 3) had an outcome of major malformation and/or spontaneous abortion; and 4) included first trimester pregnancy exposure. Exclusion criteria consisted of animal studies, non-inhalational exposure, case reports, letters, editorials, review articles and studies
20.3 Pregnancy outcome following solvent exposure
1339
that did not permit extraction of data. For subgroup analysis, we also identified and analyzed cohort and case-control studies specifically involving solvent exposure. Major malformations were defined as malformations which were either potentially life threatening or a major cosmetic defect.13 Spontaneous abortion was defined as the spontaneous termination of pregnancy before 20 weeks gestation based upon the date of the first day of the last normal menses.14 To obtain an estimate of the risk ratio for major malformations and spontaneous abortion in exposed versus unexposed infants, an overall summary odds ratio (ORs) was calculated according to the protocol established by Einarson et al.15 Additionally, homogeneity of the included studies, power analysis and the extent of publication bias were also examined as described by Einarson et al.15 Results and discussion The literature search yielded 559 articles. Of these, 549 in total were rejected for various reasons. The types of papers rejected were: animal studies (298), case reports/series (28), review articles (58), editorials (13), duplicate articles (10), not relevant (62), malformation not specified (29), spontaneous abortion not defined (31), unable to extract data (4), no indication of timing of exposure (16). Five papers were included into the major malformation analysis (Table 20.3.1) and 5 papers were included into the spontaneous abortion analysis (Table 20.3.2). Table 20.3.1. Studies of teratogenicity of organic solvents meeting criteria for meta-analysis [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Authors Axelsson et al.
Study type
Data collection
16
C
R
“serious malformations”
17
CC
R
cardiac malformations
CC
R
CNS, oral clefts, musculoskeletal, cardiac defects
CC
R
“major malformations”
C
R
“major malformations”
Tikkanen et al.
Holmberg et al. Cordier et al. 20
Lemasters
19
18
Malformation described
CC=Case control; C=Cohort; R=Retrospective
A. Malformations In total 5 studies describing results from organic solvent exposure were identified (Table 20.3.3). The summary odds ratio obtained was 1.64 (95% CI: 1.16 - 2.30) which indicates that maternal inhalational occupational exposure to organic solvents is associated with an increased risk for major malformations. The test for homogeneity yielded a chi square of 2.98 (df=4, p=0.56). When studies were analyzed separately according to study type, the chi square value from the test for homogeneity of effect for cohort studies was 0.52 (df=1, p=0.47) and for case control studies it was 0.01 (df=2, p=0.99). Their combinability remains justified on the basis of the lack of finding heterogeneity among the results. Meta-analysis of both the cohort studies and case-control studies produced similar results, i.e., they demonstrate a statistically significant relationship between organic solvent exposure in the first trimester of pregnancy and fetal malformation. The summary odds ratio
1340
Kristen I. McMartin, Gideon Koren
for cohort studies was 1.73 (95% CI: 0.74 - 4.08) and 1.62 (95% CI: 1.12 - 2.35) for case-control studies. Table 20.3.2. Studies of spontaneous abortion of organic solvents meeting criteria for meta-analysis. [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Authors
Study type
Data collection
21
CC
R
22
C
R
C
P
C
R
C
P
Windham et al.
Lipscomb et al. Shenker et al. Pinney
23
24
Eskenazi et al.
25
CC=Case control, C=Cohort, R=Retrospective, P=Prospective
Table 20.3.3. Results of studies comparing outcomes of fetuses exposed or not exposed to organic solvents. [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Reference
Congenital Defect
Exposure
Yes
No
Total
organic solvents
yes no total
3 4 7
489 492 981
492 496 988
organic solvents
yes no total
23 546 569
26 1026 1052
49 1572 1621
organic solvents
yes no total
11 1464 1475
7 1438 1475
18 2902 2950
organic solvents
yes no total
29 234 263
22 285 307
51 519 570
styrene Lemasters20
yes no total
4 13 17
68 822 890
72 835 907
TOTAL
yes no total
70 2261 2331
612 4100 4712
682 6354 7036
Axelsson et al.16
Tikkanen et al.17
Holmberg et al.18
Cordier et al.19
20.3 Pregnancy outcome following solvent exposure
1341
In this meta-analysis, major malformations were defined as “potentially life threatening or a major cosmetic defect”.13 In the general population there is a 1-3% baseline risk for major malformations. Estimate incidence via cohort studies indicated 2 studies with a total of 7 malformations in 564 exposures or 1.2% rate of malformations which falls within the baseline risk for major malformations. Publication bias is the tendency for statistically significant studies to be submitted and accepted for publication in preference to studies that do not produce statistical significance.15 This may be the case for solvent exposure and major malformations. Determining the extent of possible publication bias (file drawer analysis) is not unlike power analysis for nonsignificant results. Each provides some quantitative measure of the magnitude of the findings with respect to disproving them and requires judgment for interpretation. In order to perform a file drawer analysis effect sizes must be calculated from the summary statistic. Effect sizes represent the magnitude of the relationship between two variables. Unlike statistical significance, which is directly related to sample size, an effect size may be thought of as significance without the influence of sample size. In other words, effect size represents the “true” impact of an intervention. Cohen has determined that an effect size d=0.2 is considered small, 0.5 is medium and 0.8 is large.15 The result from this file drawer analysis indicates that one would have to obtain 2 articles with a small effect size (d=0.001) to bring the study’s overall effect size (d=0.071) to a smaller effect size of 0.05. One of the acceptable studies achieved such a small effect size. The smallest effect size was d=0.000682.16 It would therefore seem probable to have some studies stored away in file drawers with very small effect sizes (lack of statistical significance). Unfortunately, no statistical test yet exists to precisely determine such a probability and one must therefore exercise judgment. There are some considerations to bear in mind when interpreting results of this meta-analysis: 1. Environmental exposure in pregnancy is seldom an isolated phenomenon, therefore, analysis of human teratogenicity data may require stratification for a number of factors depending on the intended focus of the analysis. 2. Organic solvents belong to many classes of chemicals. Not all of the studies have examined the exact same groups of solvents in terms of both extent and range of solvents as well as frequency and duration of exposure. 3. The malformations listed in each of the papers seems to reflect a diverse range of anomalies. One might expect to notice a particular trend in malformations between studies, however, this does not appear to be the case. Certain factors should be kept in mind when evaluating the results such that a number of studies were case control in design. Certain factors inherent in this study design may affect the interpretation of their results, including recall of events during pregnancy, selection of samples based on volunteer reporting and a change in the knowledge over time regarding factors considered to significantly affect the fetus. Mothers of malformed children may understandably report exposure more often than mothers of healthy children. The recall of the exact name of the chemical, amount of exposure, starting and stopping date of exposure are also difficult to establish retrospectively. Recall may be affected by the method of questioning; when asked open ended questions, women may not recall details as well as when questioned with respect to specific chemical exposure. As a result, there could be systematic bias toward reporting exposure.
1342
Kristen I. McMartin, Gideon Koren
It is important to consider the criteria or “proof” for human teratogenicity as established by Shepard:26 1. Proven exposure to agent at critical time(s) in prenatal development. One of the inclusion criteria for this meta-analysis, with malformations as the outcome of exposure, was first trimester exposure to organic solvents. 2. Consistent findings by two or more epidemiologic studies of high quality including: control of confounding factors, sufficient numbers, exclusion of positive and negative bias factors, prospective studies if possible, and studies with a relative risk of six or more. When this happens it is unlikely that methodological problems or systematic biases can influence the results of the studies conducted in different contexts and different study designs. The studies included in this meta-analysis usually controlled for such items as geographical location and date of birth, however, other potential confounding factors such as maternal age, alcohol, and smoking that could lead to subsequent problems in outcome presentation were not consistently reported. In addition, this meta-analysis included studies that were contained within large databases spanning many years. The majority of information about occupational exposure in general during pregnancy originates from Scandinavia, namely, the Institute of Occupational Health in Helsinki. For example, Finland monitors spontaneous abortions through the spontaneous abortion registry. The registry contains all information about women who were hospitalized with spontaneous abortions covering approximately 90% of all spontaneous abortions in Finland. Finland also monitors births via the Finnish Register of Congenital Malformations. All new mothers in Finland are interviewed during their first prenatal visit, at 3 months post-delivery, at Maternity Care Centers located in every province throughout Finland. When scanning the literature, there are no studies that prospectively examine occupational exposure to organic solvents during pregnancy and pregnancy outcome with regard to malformations. The studies are retrospective, either case-control or cohort in design. In contrast, however, there are a number of studies that prospectively examine occupational exposure during pregnancy and pregnancy outcome with regard to spontaneous abortion. In all the studies there was an attempt to ascertain the occupational exposure by an industrial hygienist who blindly assessed the group exposure information. In addition, the individual studies included in the meta-analysis did not obtain an odds ratio or relative risk of 6.0 or more with a significant 95% confidence interval. The larger the value of the relative risk, the less likely the association is to be spurious. If the association between a teratogen is weak and the relative risk small (i.e., range 1.1-2.0), it is possible to think that the association is indeed due to unknown confounding factors and not to the teratogen under study. However, weak associations may be due to misclassification of exposure or disease. They may also indicate an overall low risk but the presence of a special subgroup at risk of teratogenesis within the exposed group. 3. Careful delineation of the clinical cases. A specific defect or syndrome, if present, is very helpful. If the teratogen is associated only to one or a few specific birth defects, the possibility of a spurious association becomes smaller. In this meta-analysis, the malformations were variable with no specific trend apparent. 4. Rare environmental exposure associated with rare defect. 5. Teratogenicity in experimental animals important but not essential. 6. The association should make biologic sense.
20.3 Pregnancy outcome following solvent exposure
1343
When a chemical or any other environmental factor caused a malformation in the experimental animals and/or the biological mechanism is understood, the observation of an association in humans becomes more plausible. Although the statistical association must be present before any relationship can be said to exist, only biological plausible associations can result in “biological significance”. The mechanisms by which many solvents exert their toxicity are unclear and may vary from one solvent to another. Halogenated hydrocarbons such as carbon tetrachloride may generate free radicals.27 Simple aromatic compounds such as benzene may disrupt polyribosomes, whereas some solvents are thought to affect lipid membranes and to penetrate tissues such as the brain.27 In 1979 a syndrome of anomalies (hypertonia, scaphocephaly, mental retardation and other CNS effects) was suggested in two children in a small American Indian community where gasoline sniffing and alcohol abuse are common.28 Four other children had similar abnormalities, however, in these cases it was impossible to verify gasoline sniffing. Also, it is unclear what was the contribution of the lead in the gasoline or the alcohol abuse in producing these abnormalities. It is important to remember that the mothers in many of these cases showed signs of solvent toxicity indicating heavy exposure. This is not the case in most occupational exposures during pregnancy. While fetal toxicity is biologically sensible in cases of intoxicated mothers, the evidence of fetal damage from levels that are not toxic to the mother is scanty and inconsistent. 7. Proof in an experimental system that the agent acts in an unaltered state. 8. Important information for prevention. Several lists of criteria for human teratogenicity have included the dose (or concentration) response relationship.1 Although a dose response may be considered essential in establishing teratogenicity in animals it is extremely uncommon to have sufficient data in human studies. Another criterion which is comforting to have but not very often fulfilled is biologic plausibility for the cause. Shepard states that at present there is no biologically plausible explanation for thalidomide embryopathy and that at least one half of all human teratogens do not fit this criterion.26 B. Spontaneous abortion Estimates for clinically recognized spontaneous abortions as a proportion of all pregnancies vary markedly. In ten descriptive studies reviewed by Axelsson,29 the proportion of spontaneous abortions varied from 9% to 15% in different populations. The variation depended not only on the characteristics of the population but on the methods used in the study, i.e., the selection of the study population, the source of pregnancy data, the definition of spontaneous abortion, the occurrence of induced abortions and their inclusion or otherwise in the data. The weaknesses of the studies using interviews or questionnaires pertain to the possibility of differential recognition and recall (or reporting) of spontaneous abortions and of differential response. Both exposure and the outcome of pregnancy may influence the willingness of subjects to respond to a study. One advantage of interview data is that it is more likely to provide information on early spontaneous abortion than medical records. However, the validity of information on early abortion which may be difficult to distinguish from a skipped or delayed menstruation has been suspect. Spontaneous abortions which have come to medical attention are probably better defined than self-reported abortions. The feasibility of using medical records as a source of data depends on the pattern of use of medical facilities in the community and the coverage and correctness of the records.
1344
Kristen I. McMartin, Gideon Koren
Of concern is the potential selection bias due to differing patterns of use of medical services. The primary determinant for seeking medical care is probably gestational age so that earlier abortions are less likely to be medically recorded than later abortions.29 The advantage of data on medically diagnosed spontaneous abortions, compared to interview data is that the former are independent of an individuals own definition, recognition and reporting. In total, 5 papers describing results from organic solvent exposure were identified (Table 20.3.4). The summary odds ratio obtained was 1.25 (95% CI: 0.99 - 1.58). The test for homogeneity yielded a chi square=4.88 (df=4, p=0.300). When studies were analyzed separately according to study type, the chi-square value for homogeneity of effect for cohort studies was 4.20 (df=3, p=0.241). Meta-analysis of both cohort and case-control studies produced similar results, i.e., they do not demonstrate a statistically significant relationship between organic solvent exposure in pregnancy and spontaneous abortion. The summary odds ratio for cohort studies was 1.39 (95% CI: 0.95 - 2.04) and 1.17 (95% CI: 0.87 - 1.58) for case control studies. Their combinability seems justified on the basis of the lack of finding heterogeneity among the results. Table 20.3.4. Results of studies comparing outcomes of fetuses exposed or not exposed to organic solvents. [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Reference
Spontaneous Abortion
Exposure
yes
no
total
any solvent product
yes no total
89 272 361
160 575 735
249 847 1096
organic solvent
yes no total
10 87 97
39 854 893
49 941 990
organic solvents
yes no total
12 16 28
8 21 29
20 37 57
organic solvents
yes no total
35 25 60
228 166 394
263 191 454
organic solvents Eskenazi et al.25
yes no total
4 7 11
97 194 291
101 201 302
TOTAL
yes no total
150 407 557
532 1810 2342
682 2217 2899
Windham et al.21
Lipscomb et al.22
Schenker et al.23
Pinney24
The overall ORs of 1.25 indicates that maternal inhalational occupational exposure to organic solvents is associated with a tendency towards a small increased risk for spontaneous abortion. The addition of one study of similar effect size would have rendered this trend statistically significant.
20.3 Pregnancy outcome following solvent exposure
1345
Traditionally, a power analysis would be conducted to determine the number of subjects or in this situation the number of “studies” that need to be added to produce a significant result. In order to perform a power analysis effect sizes must be calculated from the summary statistic. The result from this power analysis indicates that one would have to obtain 2 studies with a medium effect size (0.5) to bring this study’s overall effect size (d=0.095) to a small effect size of 0.2. Similarly, 5 articles with an effect size of d=0.3 are needed to bring the study’s overall effect size to 0.2. The largest effect size in the spontaneous abortion analysis was d=0.2. None of the acceptable studies achieved such a large effect size as 0.5. It may be improbable because one would expect that such results would undoubtedly have been published. Unfortunately, no statistical test yet exists to precisely determine such a probability and one must therefore exercise judgment. This meta-analysis addresses the use of organic solvents in pregnancy. Organic solvent is a very broad term that includes many classes of chemicals. There may still exist rates of abortion higher than the value reported with certain groups of solvents. However, a detailed analysis of classes of solvents is in order to incriminate a particular solvent. Not all of the studies have examined the same groups of solvents in terms of both extent and range of solvents as well as frequency and duration of exposure. Hence it would be very difficult to obtain any clear estimate of risk for a given solvent given the limited number of studies available. Conclusion The meta-analysis examining organic solvent use in pregnancy did not appear to find a positive association between organic solvent exposure and spontaneous abortions (ORs = 1.25, confidence interval 0.99 - 1.58). The results from the meta-analysis examining organic solvent use in the first trimester of pregnancy and major malformations indicate that solvents are associated with an increased risk for major malformations (ORs = 1.64, confidence interval 1.16 - 2.30). Because of the potential implications of this review to a large number of women of reproductive age occupationally exposed to organic solvents, it is important to verify this cumulative risk estimate by a prospective study. Similarly, it is prudent to minimize women’s exposure to organic solvents by ensuring appropriate ventilation systems and protective equipment. Meta-analysis can be a key element for improving individual research efforts and their reporting in the literature. This is particularly important with regard to an estimate of dose in occupational studies as better reporting of the quantification of solvent exposure is needed in the reproductive toxicology literature. 20.3.4 PREGNANCY OUTCOME FOLLOWING GESTATIONAL EXPOSURE TO ORGANIC SOLVENTS: A PROSPECTIVE CONTROLLED STUDY [Adapted, by permission, from S. Khattak, G. K-Moghtader, K. McMartin, M. Barrera, D. Kennedy and G. Koren, JAMA., 281, 1106 (1999) Copyright 1999, American Medical Association] The Motherisk Program at the Hospital for Sick Children was the first to prospectively evaluate pregnancy and fetal outcome following maternal occupational exposure to organic solvents with malformations being the primary outcome of interest.30 Methods The study group consisted of all pregnant women occupationally exposed to organic solvents and counseled between 1987-1996 by the Motherisk Program at the Hospital for Sick Children. Details concerning the time of exposure to organic solvents were recorded for de-
1346
Kristen I. McMartin, Gideon Koren
termination of temporal relationship between exposure and conception. The details on chemical exposure were recorded, including occupation, type of protective equipment used, and other safety features, including ventilation fans. Adverse effects were defined as those known to be caused by organic solvents (e.g., irritation of the eyes or respiratory system, breathing difficulty, headache). Temporal relationship to exposure was investigated to separate these symptoms from those associated with pregnancy. One hundred twenty-five pregnant women who were exposed occupationally to organic solvents and seen during the first trimester between 1987and 1996. Each pregnant woman who was exposed to organic solvents was matched to a pregnant woman who was exposed to a nonteratogenic agent on age (+/- 4 years), gravidity (+/- 1) and smoking and drinking status. The primary outcome of interest was major malformations. A major malformation was defined as any anomaly that has an adverse effect on either the function or the social acceptability of the child. The expected rate of major malformations is between 1% to 3%. Results and discussion Significantly more major malformations occurred among fetuses of women exposed to organic solvents than controls (13 vs 1; relative risk, 13.0; 95% confidence interval, 1.8-99.5). Twelve malformations occurred among the 75 women who had symptoms temporally associated with their exposure, while none occurred among 43 asymptomatic exposed women (p<0.001). (One malformation occurred in a woman for whom such information was missing.) More of these exposed women had previous miscarriage while working with organic solvents than controls (54/117 [46.2%] vs 24/125 [19.2%]; p<0.001). However, exposed women who had a previous miscarriage had rates of major malformation that were similar to exposed women who had no previous miscarriage. The Motherisk protocol allowed us to record in a systematic manner all exposure data and other maternal and paternal medical details at the time of exposure during the first trimester of pregnancy and to follow up pregnancy outcomes prospectively in this cohort. The control group was assessed in an identical manner. This prospective study confirms the results of our recent meta-analysis.31 Women occupationally exposed to organic solvents had a 13 fold risk of major malformations as well as an increased risk for miscarriages in previous pregnancies while working with organic solvents. Moreover, women reporting symptoms associated with organic solvents during early pregnancy had a significantly higher risk of major malformations than those who were asymptomatic suggesting a dose-response relationship. Other factors, for example, type of solvent, might have accounted for the presence of symptoms in some women. Although some human teratogens have been shown to cause a homogeneous pattern of malformation(s), in other cases no specific syndrome has been described.32 No homogenous pattern of malformations is obvious from the prospective study. However, organic solvents although traditionally clustered together, are a diverse group of compounds that should not be expected to cause similar patterns of reproductive toxic effects. Although more prospective studies will be needed to confirm our results, it is prudent to minimize women’s exposure to organic solvents during pregnancy. This is most important during the first trimester of pregnancy.
20.3 Pregnancy outcome following solvent exposure
1347
20.3.5 A PROACTIVE APPROACH FOR THE EVALUATION OF FETAL SAFETY IN CHEMICAL INDUSTRIES [Adapted, by permission from K.I. McMartin and G. Koren, Teratology, 60, 130 (1999) Copyright 1999 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Introduction Women, their families and employers are concerned about potential fetal risks that may be associated with occupational exposure to chemicals. To be able to assess such risks in a particular plant, one has to quantify local exposure and contrast it with evidence-based literature data. There are, however, numerous obstacles that prevent such risk assessment from being routinely performed. In the reproductive literature there are few studies that actually quantify exposure levels. In the instance where authors attempt to quantify or stratify exposure, the exposure frequencies and the exposure doses are inconsistent between studies. For many chemicals one can measure neither airborne nor blood levels. Smelling the odor of organic solvents is not indicative of a significant exposure as the olfactory nerve can detect levels lower than several parts per billion, which are not necessarily associated with toxicity. Odor thresholds for some solvents are far below several parts per million (ppm). Examples of some odor thresholds33 include carbon disulfide (0.001 ppm vs. TLV-TWA (skin) [Threshold Limit Value-Time Weighted Average] 10 ppm), acetaldehyde (0.03 ppm vs. TLV-TWA 25 ppm), and ethyl mercaptan (2x10-5 ppm vs. TLV-TWA 0.5 ppm).34 In the workplace, exposure is usually to several chemicals that may change between working days or even within a single day. The amounts of chemicals absorbed are often unknown, and the circumstances of exposure may vary from workplace to workplace or even within the same operation. Typically, investigations into fetal safety are induced by single or clusters of specific malformations, or by symptoms in exposed women. We recently reported a proactive consultation process where, for a selected chemical compound to which women working in the Products and Chemicals Divisions at Imperial Oil Limited (IOL) may be exposed, actual exposure data were contrasted with literature values and a risk assessment was constructed.35 Methods An agent inventory list was used to analyze the component (the name of material or agent), exposure group, the number of employees within an exposure group, and the routine rating factor for routine work. Exposure group is defined as a group of employees who have similar exposures to chemical, physical, and/or biological agents when: 1) holding different jobs but working continuously in the same area (e.g., process workers), or 2) holding unique jobs in an area or moving frequently between areas (e.g., maintenance workers). The routine rating factor for routine work (work which is part of the normal repetitive duty for an exposure group) is defined as follows: Rating Factor (RF) Definition 0 No reasonable chance for exposure 1-5 Minimal, exposure not expected to exceed 10% of the occupational exposure limit (OEL) 6-9 Some daily routine exposures may be expected between 10% and 50% of the OEL 10-15 Some daily routine exposures may exceed 50% of the OEL
1348
Kristen I. McMartin, Gideon Koren
The rating factor (RF) can be assessed using industrial hygiene professional judgment or monitored data. NRRF is the non-routine rating factor for non-routine work defined as job task or activities which are done seasonally, occasionally or cyclical. The definitions listed for RRF apply. For each component a listing was created with respect to individual chemicals, including rating factors, for female exposure in the Products and Chemicals Divisions. In addition, a literature search was performed for each chemical that incorporated female occupational exposure during pregnancy with human teratogenicity and spontaneous abortion as pregnancy outcomes. Teratogenicity and spontaneous abortion were chosen as the outcomes of interest as they represent the majority of endpoints examined in studies focusing on female occupational exposure during pregnancy. Most of the selected female reproductive toxicology studies examined explicitly stated chemical exposure levels: either as parts per million, stratifying as to number of days of exposure, or as estimates of the percentage of the threshold limit values. Medline, Toxline, and Dissertation Abstracts databases were utilized to search for all research papers published in any language from 1966 to 1996. In total, 559 studies were obtained from the literature search. Of these, only 21 studies explicitly stated some sort of exposure level for the various chemicals. These chemical exposure levels in the literature and subsequent pregnancy outcomes were compared to IOL chemical exposure indices. The following is an example of one of the many chemical exposures encountered, namely exposure to toluene. For other compounds, Table 20.3.5 contrasts values in the literature with IOL indices of chemical exposure. Table 20.3.5. Examples of IOL compound exposure indices contrasted to literature values. [Adapted, by permission from K.I. McMartin and G. Koren, Teratology, 60, 130 (1999) Copyright 1999 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Chemical
Reference 42
Literature Exposure Levels
IOL Exposure Levels “no reasonable chance for exposure to minimal exposure not expected to exceed 10% OEL” TLV-TWA: 2 ppm
Aniline
Posluzhnyi
“low exposure area”
Benzene
Mukhametova and Vozovaya43
“no reasonable chance for expo“within or lower than the maxi- sure to some daily exposures may mum permissible levels” be expected between 10-50% OEL” TLV-TWA:10 ppm
Chloroform
Taskinen et al.44 >once a week
“no reasonable chance for exposure to minimal exposure not expected to exceed 10% OEL” TLV-TWA: 10 ppm
>once a week
“no reasonable chance for exposure to minimal exposure not expected to exceed 10% OEL” TLV-TWA: 50 ppm
44
Taskinen et al. Dichloromethane
Windham et al.21
20.3 Pregnancy outcome following solvent exposure
Chemical
Reference
Literature Exposure Levels
1349
IOL Exposure Levels
Styrene
Saamanen Harkonen46
70-100 ppm 20-300 ppm
“no reasonable chance for exposure to minimal exposure not expected to exceed 10% OEL” TLV-TWA: 50 ppm, TLV-STEL: 100 ppm
Toluene
Euler10 Syrovadko36 Ng et al.39
298 ppm 13-120 ppm 50-150 ppm
“no chance for exposure to some daily exposure exceeding 50% of the OEL” TLV-TWA: 50 ppm
45
IOL: Imperial Oil Limited, OEL: Occupational Exposure Limit
Results and discussion Six studies were found that quantified toluene concentrations. The countries that reported these observations included Germany, Russia, Finland and Singapore. In general, IOL toluene levels are lower than those reported in the literature. A few case reports of malformations in association with toluene exposure have appeared. Euler10 reported 2 cases of multiple malformations where the anomalies were similar in children born to women who worked in shoemaking and were exposed to a soling solution containing toluene and trichloroethylene. The average concentration of toluene in the air was 298 ppm (1.12 mg/l) and of trichloroethylene 230 ppm (1.22 mg/l). No further details of these cases were given. Toutant and Lippmann28 reported a single case of adverse pregnancy outcome in a woman addicted to solvents (primarily toluene). The woman, aged 20 years, had a 14-year history of daily heavy solvent abuse. On admission to the hospital, she had ataxia, tremors, mild diffuse sensory deficits, short-term memory loss, blunted affect, and poor intellectual functioning compatible with severe solvent and/or alcohol abuse. The male child born at term was microcephalic with a flat nasal bridge, hypoplastic mandible, short palpebral fissures, mildly low-set ears, pronounced sacral dimple, sloping forehead and incoordination of arm movements with unusual angulation of the left shoulder and elbow. There was a poor sucking reflex and movements were jerky at 2-4 days of age, although this improved spontaneously. The authors of this report point out the similarities between this case and fetal alcohol syndrome and suggest that there may be an analogous “fetal solvent syndrome” or that excessive solvent intake may enhance the toxicity of alcohol. Syrovadko36 studied the outcome of pregnancy in a substantial number of women exposed to toluene. Toluene exposure averaged 55 ppm (range 13-120 ppm). The factory had its own maternity section where the women had their deliveries. Records of labor and newborns were examined for 133 women in contact with toluene and for 201 controls from the factory offices. There was no detectable effect on fertility. In the exposed group, records showed a mean pregnancy rate of 3.2/worker compared with 2.6/worker in the control group. There were no significant differences between exposed and control groups in the mortality or adverse effects on the newborn. In the Finnish study of Holmberg37 on central nervous system defects in children born to mothers exposed to organic solvents during pregnancy, 3 of the cases were exposed to toluene, or toluene in combination with other solvents. In one case with hydranencephaly and death 24 days after birth, there was exposure to toluene, xylene, white spirit and methyl ethyl ketone from rubber products manufacture. The second case had multiple abnormali-
1350
Kristen I. McMartin, Gideon Koren
ties of hydrocephalus, agenesis of the corpus callosum, pulmonary hypoplasia and diaphragmatic hernia, and died 2 hours after birth. In the second case, the mother was exposed to toluene while manufacturing metal products. The third case had lumbar meningomyelocele and survived. The mother was exposed to toluene and white spirit. Toluene air concentrations were not stated. A case-referent study concerning selected exposures during pregnancy among mothers of children born with oral clefts was conducted in Finland.38 The study covered the initial 3.5 years’ material and was a more detailed extension of earlier retrospective studies concerning environmental factors in the causation of oral clefts, using cases accumulated from the Finnish Register of Congenital Malformations. More case mothers (14) than referent mothers (3) had been exposed to organic solvents during the first trimester of pregnancy. The mothers were considered “substantially” exposed if their estimated continuous exposure had been at least one-third of the current TLV concentration or if the estimated peak exposure had reached the TLV concentration, e.g., during home painting in confined spaces. Various solvents included: lacquer petrol, xylene, toluene, acetates, alcohols, denatured alcohol, methyl ethyl ketone, dichloromethane, turpentine, styrene, and aromatic solvent naphtha (C4-C14 aromatics). Ng et al.39 examined the risk of spontaneous abortion in workers exposed to toluene. Rates of spontaneous abortions were determined using a questionnaire administered by personal interview to 55 married women with 105 pregnancies. The women were employed in an audio speaker factory and were exposed to high concentrations of toluene (mean 88 ppm, range 50-150 ppm). These rates of spontaneous abortion were compared with those among 31 women (68 pregnancies) who worked in other departments in the same factory and had little or no exposure to toluene (0-25 ppm) as well as with a community control group of women who underwent routine antenatal and postnatal care at public maternal health clinics. Significantly higher rates for spontaneous abortions were noted in the group with higher exposure to toluene (12.4 per 100 pregnancies) compared with those in the internal control group (2.9 per 100 pregnancies) and in the external control group (4.5 per 100 pregnancies). Among the exposed women, significant differences were also noted in the rates of spontaneous abortion before employment (2.9 per 100 pregnancies) and after employment in the factory (12.6 per 100 pregnancies). Tikkanen et al.17 performed a study to explore for possible associations between occupational factors and cardiovascular malformations. Information on the parents of 160 infants with cardiovascular malformations and 160 control parents were studied. The mother was considered “substantially” exposed to “organic solvents” if the estimated continuous exposure was at least one third of the ACGIH threshold limit value concentration or the estimated short term exposure reached the TLV concentration (while painting kitchen walls). Organic solvents were categorized as 1) “hydrocarbons”, 2) “alcohols” and 3) “miscellaneous”. Hygiene assessments of exposures were classified as i) “any exposure intensity” (at any period in pregnancy and in the first trimester only) and ii) “substantial exposure intensity” (at any period in pregnancy and in the first trimester only). Of the 320 mothers, 41 case and 40 control mothers reported an exposure to organic solvents.16 The hygiene assessment indicated some solvent exposure in 27 case and 25 control mothers. Twenty-one case and 16 control mothers had been exposed in the first trimester. Of these, substantial exposure to hydrocarbons occurred for 6 case and 2 control
20.3 Pregnancy outcome following solvent exposure
1351
mothers; one case and one control mother to toluene at work and five cases and one control mother to lacquer petrol while painting indoors at home for 1 to 2 days. Lindbohm et al.40 investigated the association between medically diagnosed spontaneous abortions and occupational exposure to organic solvents (case-control design). The study population was composed of women who were biologically monitored for solvents. The workers were classified into exposure categories on the basis of work description and the use of solvents as reported in the questionnaires and on measurements of biological exposure. Three exposure levels were distinguished: high, low, and none. The level of exposure was assessed on the basis of the reported frequency of solvent use and the available information on typical levels of exposure in that particular job, as based on industrial hygiene knowledge. The feasibility of biological monitoring data for classification of exposure was limited because the solvent measurements describe only short-term exposure (from 2 hours to a few days) and only 5% of the workers had been measured during the first trimester of pregnancy. Therefore, the exposure classification was based mainly on the work task description and reported solvent usage. Exposure was defined as “high” if the worker handled the solvents daily or 1-4 days a week and the level of exposure was high according to biological exposure measurements or industrial hygiene measurements available at the Institute of Occupational Health. Exposure was defined as “low” if the worker handled solvents 1-4 days a week and the level of exposure according to the measurements of the Institute was low or if the worker handled solvents less than once a week. Otherwise, the level of exposure was defined as “none”. After classification, the work tasks and the related exposures were listed by the level of exposure which was checked by an independent, experienced industrial hygienist. The final population for the analysis was restricted to the matched case-control sets who confirmed their pregnancy and reported in detail their occupational exposures during early pregnancy (73 cases and 167 controls). The odds ratios for tetrachloroethylene and aliphatic hydrocarbons, adjusted for potentially confounding factors, increased with the level of exposure.40 For toluene the reverse was the case. Aliphatic hydrocarbons had not been biologically monitored, but industrial hygiene measurements had been performed by the Institute of Occupational Health in two printing houses which contributed subjects to this study. In two of four measurements, the concentrations of white spirit in air exceeded, during the cleaning of the printing machine, the Finnish Threshold Limit Value (150 ppm). All the printers included in this study reported that their work included cleaning of the machine. The association of tetrachloroethylene, toluene and aliphatic hydrocarbons with spontaneous abortions was also examined by detailed records of occupational task.40 The odds ratio of spontaneous abortion for aliphatic hydrocarbons was increased among graphic workers [5.2 (1.3-20.8)] and painters [2.4 (0.5-13.0)] but not among other workers. However, in the latter group the proportion of highly exposed workers was only 30%, whereas it was 69% in the two former groups. The odds ratio was increased also among toluene-exposed shoe workers [odds ratio 9.3 (1.0-84.7)] and dry cleaners exposed to tetrachloroethylene [odds ratio 2.7 (0.7-11.2)]. The results of the study by Lindbohm et al.40 support the hypothesis of a positive association between spontaneous abortion and exposure to organic solvents during pregnancy and suggest that exposure, especially to aliphatic hydrocarbons, increases the risk of abortion. The highest risk for aliphatic hydrocarbons was found among graphic workers who
1352
Kristen I. McMartin, Gideon Koren
were employed as offset printing workers or printing trade workers. They used the solvents for cleaning the printing machines and as diluent for printing ink. In cleaning the machines, exposure to mixtures of nonaromatic mineral oil distillates with 0-15% aromatic compounds may reach a high level for a short period.40 The workers were also exposed among other things to toluene, 1,1,1-trichloroethane, thinner, and xylene. Although the data suggest that the findings are due to aliphatic hydrocarbons, combined solvent effects cannot be excluded because of the multiple exposures to different solvents.40 The mean measured level of blood toluene among the shoe workers was slightly higher (0.51 mmol/L, 13 morning samples) than the mean among the other toluene-exposed workers (0.38 mmol/L, 10 morning samples).40 The shoe workers also reported use of toluene more frequently than the other toluene-exposed workers. Industrial hygiene measurements had been performed in three of the five work places of the shoe workers. The concentration of toluene in air varied from 1 ppm to 33 ppm. Other solvents detected were acetone and hexane. In two of the three shoe factories from which industrial hygiene measurements were available, relatively high levels of hexane (33-56 ppm) were measured. Hexane, being an aliphatic compound, may have contributed to the excess of spontaneous abortions.40 Comparison with IOL levels The routine rating factor and non-routine rating factor from the Products and Chemical Divisions range from 00 to 11 indicating no reasonable chance for exposure to some daily exposures exceeding 50% of the OEL. The TLV-TWA for toluene is 50 ppm.34 The Euler case reports documented air concentrations of 298 ppm for toluene and 230 ppm for trichloroethylene.10 Both of these air concentrations exceed current standards, but no further details of these cases were given. Syrovadko reported a toluene exposure of 55 ppm (range 13-120 ppm), again, exceeding current standards.36 Holmberg et al.38 and Tikkanen17 considered workers “substantially” exposed if their estimated continuous exposure had been at least one-third of the current TLV concentration or if the estimated peak exposure had reached the TLV concentration. Similarly, Ng38 described high concentrations of toluene (mean 88, range 50-150 ppm) exceeding current standards. All these exposure levels for toluene exceed the current threshold limit value. IOL toluene exposure levels are considerably lower than any value reported in the literature. Lindbohm et al.,40 for two of four air measurements, reported concentrations of white spirit exceeded the Finnish Threshold Limit Value (150 ppm) during the cleaning of the printing machine. Industrial hygiene measurements were performed in three of the five work places of the shoe workers. The concentration of toluene in air varied from 1 ppm to 33 ppm. Other solvents detected were acetone and hexane. In two of the three shoe factories from which industrial hygiene measurements were available, relatively high levels of hexane (33-56 ppm) were noted. The routine rating factor and non-routine rating factor from the Products and Chemicals Divisions for hexane isomers range from 00 to 05 indicating no reasonable chance for exposure or minimal exposure not expected to exceed 10% of the occupational exposure limit (OEL). The routine rating factor and non-routine rating factor from the Products and Chemicals Divisions for n-hexane range from 00 to 07 indicating no reasonable chance for exposure or some daily exposures between 10% and 50% of the OEL. The TLV-TWA of n-hexane is 50 ppm or 176 mg/m3.34 The TLV-TWA of other hexane isomers is 500 ppm or 1760 mg/m3 and the TLV-STEL is 1000 ppm or 3500 mg/m3.34 In comparison with the pre-
20.3 Pregnancy outcome following solvent exposure
1353
vious hexane levels reported in the literature, IOL hexane exposure levels are substantially lower. In mice levels of inhalation exposure to toluene have included 100 to 2,000 ppm at various times during gestation as well as at various durations of exposure (6-24 hours/day).41 Growth and skeletal retardation were noted at lower levels (133 ppm and 266 ppm, respectively) when such exposures were of a 12-24 hour duration for at least half of the gestation period.41 Human levels of inhaled toluene exposure that would be comparable would be those obtained by chronic abusers (5,000-12,000 ppm). The only noted malformations were an increase in the frequency of 14th ribs, which was noted at 1,000 ppm on days 1-17 of gestation for 6 hours/day. As Wilkins-Haug41 notes this has been the highest exposure studied in the mouse model and is comparable to the inhaled toluene exposure which produces euphoria in humans (500 ppm). In 1991 we were approached by the medical department of Imperial Oil Limited to develop a proactive approach of risk evaluation of their female workers. The paradigm developed and used by us could be extrapolated to any other chemical operation. Its advantage is in its proactive nature, which aims at informing workers and preventing potential fetal risks, while also preventing unjustified fears which may lead women to quit their jobs or, in extreme cases, even consider termination of otherwise wanted pregnancies. Upon comparing the occupational literature that presented any quantifiable chemical exposure dose or estimate of dose for any chemical with the IOL routine rating factors in the Products and Chemicals Divisions, we could conclude that IOL chemical exposure levels overall were lower than those reported in the literature. Of utmost importance is the need in published occupational reports for at least some industrial hygiene documentation, namely improved reporting of a quantifiable chemical exposure dose (for example, as implemented and currently utilized by IOL) and ideally a standard and consistent way of reporting this in the occupational literature. 20.3.6 OVERALL CONCLUSION The Motherisk program is an information and consultation service for women, their families and health professionals on the safety/risk of exposure to drugs, chemicals, radiation and infection during pregnancy and lactation. Chemical exposure in the workplace is a common source of concern among our patients and health professionals. Occupational exposure to organic solvents during pregnancy is associated with an increased risk of major fetal malformations. This risk appears to be increased among women who report symptoms associated with organic solvent exposure. Although more prospective studies will be needed to confirm our results, it is prudent to minimize women’s exposure to organic solvents during pregnancy. This is most important during the first trimester of pregnancy. Moreover, symptomatic exposure appears to confer an unacceptable level of fetal exposure and should be avoided by appropriate protection and ventilation. Health care professionals who counsel families of reproductive age should inform their patients that some types of employment may influence reproductive outcomes. Of utmost importance is the need in published occupational reports for some industrial hygiene documentation. Specifically, improved reporting of a quantifiable chemical exposure dose (for example, as implemented and currently utilized by IOL) and ideally a standard and consistent way of reporting this in the occupational literature pertaining to human reproductive toxicology.
1354
Kristen I. McMartin, Gideon Koren
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
J. Schardein, in Chemically Induced Birth Defects, Marcel Dekker, New York, 1985, pp. 645-658. G. Watanabe and S. Yoshida, Acta. Medica. Biol. Niigata., 12, 285 (1970). W. Masumoto, S. Ijima and H. Katsunuma, Congenital Anomalies, 15, 47 (1975). P.S. Nawrot and R.E. Staples, Teratology, 19, 41A (1979). F.J. Murray, J.A. John, L.W. Rampy, R.A. Kuna and B.A. Schwetz, Ind. Hyg. Ass. J., 40, 993 (1979). R. Iwanaga, T. Suzuki and A. Koizumi, Jap. J. Hyg., 25, 438 (1970). A. Hudak and G. Ungvary, Toxicology, 11, 55 (1978). J.D. Green, B.K.J. Leong and S. Laskin, Toxicol. Appl. Pharmacol., 46, 9 (1978). S.M. Barlow and F.M. Sullivan, in Reproductive Hazards of Industrial Chemicals, Academic Press, London, 1982. H.H. Euler, Arch. Gynakol., 204, 258 (1967). B.A. Schwetz, B.M.J. Leong and B.J. Gehring, Toxicol. Appl. Pharmacol., 32, 84 (1975). B.K. Nelson, B.J. Taylor, J.V. Setzer and R.W. Hornung, J. Environ. Pathol. Toxicol., 3, 233 (1980). O. Heinonen, in Birth Defects and Drugs in Pregnancy, PSG Publishing, Littleton, MA, 1977, pp. 65-81. F.G. Cunningham, P.C. McDonald and N. Gant in Williams Obstetrics, Appleton and Lange, Norwalk, Connecticut, 1989, pp. 489-509. T.R. Einarson, J.S. Leeder and G. Koren, Drug. Intell. Clin. Pharm., 22, 813 (1988). G. Axelsson, C. Liutz and R. Rylander. Br. J. Ind. Med., 41, 305 (1984). J. Tikkanen and O. Heinonen, Am. J. Ind. Med., 14, 1 (1988). P.C. Holmberg, K. Kurppa, R. Riala, K. Rantala and E. Kuosma, Prog. Clin. Biol. Res., 220, 179 (1986). S. Cordier, M.C. Ha, S. Ayme and J. Goujard, Scand. J. Work Environ. Health., 18, 11 (1992). G.K. Lemasters, An epidemiological study of pregnant workers in the reinforced plastics industry assessing outcomes associated with live births, University of Cincinnati, Cincinnati, 1983. G.C. Windham, D. Shusterman, S.H. Swan, L. Fenster and B. Eskenazi, Am. J. Ind. Med., 20, 241 (1991). J.A. Lipscomb, L. Fenster, M. Wrensch, D. Shusterman and S. Swan, J. Occup. Med., 33, 597, (1991). M.B. Schenker, E.B. Gold, J.J. Beaumont, B. Eskenazi, S.K. Hammond, B.L. Lasley, S.A. McCurdy, S.J. Samuels, C.L. Saiki and S.H. Swan, Final report to the Semiconductor Industry Association. Epidemiologic study of reproductive and other health effects among workers employed in the manufacture of semiconductors, University of California at Davis, 1992. S.M. Pinney, An epidemiological study of spontaneous abortions and stillbirths on semiconductor employees, University of Cincinnati, Cincinnati, 1990. B. Eskenazi, M.B. Bracken, T.R. Holford and J. Crady, Am. J. Ind. Med., 14, 177, (1988). T.H. Shepard, Teratology, 50, 97, (1994). Y. Bentur in Maternal Fetal Toxicology, G. Koren, Ed., Marcel Dekker, New York, 1994, pp. 425-445. C. Toutant, and S. Lippmann, Lancet, 1, 1356, (1979). G. Axelson, R. Rylander, Int. J. Epidemiol., 13, 94, (1984). S. Khattak, G. K-Moghtader, K. McMartin, M. Barrera, D. Kennedy and G. Koren, JAMA., 281, 1106 (1999). K.I. McMartin, M. Liau, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998). G. Koren, A. Pastuszak and S. Ito, N. Engl. J. Med., 338, 1128 (1998). M.J. Ellenhorn and D.G. Barceloux, Medical Toxicology: Diagnosis and Treatment of Human Poisoning. Elsevier Science Publishing Company Inc., New York, 1988, pp. 1412-1413. Anonymous, Threshold limit values for chemical substances and physical agents and biological exposures, American Conference of Governmental Industrial Hygienists, Cincinnati, 1993, pp. 12-35. K.I. McMartin and G. Koren, Teratology, 60, 130 (1999). O.N. Syrovadko, Gig Tr Prof Zabol., 21, 15 (1977). P.C. Holmberg, Lancet, 2, 177, (1979). P.C. Holmberg, S. Hernberg, K. Kurppa, K. Rantala, and R. Riala, Int. Arch. Occup. Environ. Health, 50, 371 (1982). T.P. Ng, S.C. Foo, and T. Yoong, Br. J. Ind. Med., 49, 804, (1992). M.L. Lindbohm, H. Taskinen, M. Sallmen, and K. Hemminki, Am. J. Ind. Med., 17, 447, (1990). L. Wilkins-Haug, Teratology, 55, 145, (1997). P.A. Podluzhnyi, Gig. Sanit., 1, 44, (1979). G.M. Mukhametova, and M.A. Vozovaya, Gig. Tr. Prof. Zabol., 16, 6 (1972). H. Taskinen, M.L. Lindbohm, and K. Hemminki, Br. J. Ind. Med., 43, 199 (1986). A. Saamanen, Styreeni, Institute of Occupational Health, Helsinki, 1991. H. Harkonen, S. Tola, M.L. Korkala, and S. Hernberg, Ann. Acad. Med. Singapore, 13, 404 (1984).
20.4 Industrial solvents and kidney disease
1355
20.4 INDUSTRIAL SOLVENTS AND KIDNEY DISEASE Nachman Brautbar University of Southern California, School of Medicine, Department of Medicine, Los Angeles, CA, USA
20.4.1 INTRODUCTION Industrial solvents are used extensively in the industry, as well as modern living. The principle class of components are the chlorinated and non-chlorinated hydrocarbons. The various types of commonly used hydrocarbons are presented in Figure 20.4.1.
Figure 20.4.1. General classification of hydrocarbons based on general structure.[Adapted, by permission, from A.T. Roy, Nephron, 58, 385, 1991.]
Solvents are absorbed into the human body through several routes including 1) inhalation through the lungs, 2) absorption through the skin, 3) ingestion (in rare cases). The main route of absorption is commonly pulmonary, the lung, and this depends on several factors including the frequency of breathing, diffusion of solvent vapors across the alveolar membrane, partial pressure of solvent vapor in inspired air and blood, and solubility of the solvent in blood as the result of to air partition coefficient, and blood flow through the lungs.1,2,3 Once in the circulation, 25% of the cardiac output which is about 1200 cc of blood per minute passes through the kidneys. Therefore it is no surprise that with this amount of blood passing through the kidney and carrying solvents (from either industrial inhalation, skin absorption, and on rare occasions ingestion) the effects of solvents on the kidney has become a practical clinical one. Since inhaled hydrocarbons are readily absorbed into the blood stream and become lipophilic and readily pass across the lipid membranes. In addition to reaching the kidney, the solvents reach the brain (as does the most ancient solvent, alcohol) and enter the blood brain barrier in high concentration. Skin absorption is the second most important route for solvent entry into the body and at times is much more significant than inhalation. The reason is that absorption of organic solvent vapors by inhalation at the threshold limit value is insignificant and is less than 2% of the amount absorbed via inhalation under the same exposure conditions.3 In contrast, solvents may be absorbed through the skin in significant amounts even at below the threshold limit value.3 Factors that effect the skin absorption of solvents include the composition of
1356
Nachman Brautbar
the skin, whether the skin is healthy or not (there is increased absorption if the skin has reduced cellular membrane), and the lipid solubility of the solvent. As far as the gastrointestinal tract, commonly this is not a significant route of absorption. Solvents absorbed via the gastrointestinal tract are removed immediately by the liver through the first-pass metabolism. If the amount of solvents and quantity of solvents ingested is increased significantly and exceeds the capacity of the liver to metabolize the solvents, then the gastrointestinal tract route will become significant.4,5,6 The distribution of an organic solvent in the human body depends upon its partial pressure in the arterial blood and the solubility of the solvent in the tissue, as well as the blood flow rate through the tissue.7 Data on tissue distribution of various solvents are limited at best. The metabolism of solvents depend on the solvent. Alcohols are metabolized via alcohol dehydrogenase, whereas other organic solvents are mainly metabolized by the cytochrome P-450-dependent enzymes. These enzymes may be found in the liver, kidneys, lungs, gastrointestinal tract, gonads, adrenal cortex, and other body organ tissues. The metabolism of solvents has been described extensively,8,9 and the reader is referred to those writings. The metabolites of the organic solvents are eliminated via the kidneys through urine excretion and to some extent, by exhalation of the unchanged original solvent. Commonly the parent solvent is eliminated by the kidneys and this amounts to less than 1%. The metabolites are the main source of excretion of the metabolized parent solvent. In the last several decades, there have been several studies in experimental animals, case reports in humans, case studies in humans, and epidemiological studies in humans on the effects of solvents on the kidney, both acutely and chronically. The scope of this chapter is the clinical chronic effects of solvents on the kidney (chronic nephrotoxicology). 20.4.2 EXPERIMENTAL ANIMAL STUDIES The toxic effects of organic solvents on the kidneys has been studied in several experimental species, especially mice and rats. Damage to the kidney has been shown in these experimental animals in the form of acute damage to various parts of the nephron, especially the tubules. This has usually been described as tubular degeneration with regenerative epithelium, deposits of mineral crystals and of intralobular proteins, and interstitial inflammation.8,10-15 Several studies have shown glomerular damage in experimental animal16,17 and have suggested that long-term solvent exposure alters the immune system and leads to the glomerulopathy with mesangial IgA deposits. While the exact mechanism is not known and various mechanisms have been postulated, it is reasonable to accept a mechanistic approach which takes into account genetic, environmental, susceptibility (such as pre-existing diseases including hypertensive kidney disease and diabetes), direct tubular toxicity, permeability changes and immunosuppression. 20.4.3 CASE REPORTS The earlier documentation of chronic renal disease and hydrocarbon exposure consists of case reports, and this data was summarized by Churchill et al.18 describing Goodpasture’s syndrome in 15 adults, epimembranous glomerulonephritis in 5 adults, and subacute proliferative glomerulonephritis in one adult. The hydrocarbon exposures were for solvents in 12 patients, gasoline in 4, gasoline-based paint in 3, jet fuel, mineral turpentine and un-
20.4 Industrial solvents and kidney disease
1357
specified in 1 case report. These case reports were previously summarized by us in a previous publication19 and are represented in the following table. Table 20.4.1. Case series report of glomerulonephritis and hydrocarbon exposure. [Data from reference number 19] Investigator
n
20
Diagnosis
Agent
2
Goodpasture’s
Gasoline
1
Goodpasture’s
Gasoline
1
Goodpasture’s
Gasoline-based paint spray
Beirne and Brennan23
5 1
Goodpasture’s RPGN
Degreasing and paint Solvents and jet fuel
D’Apice, et al.24
2
Goodpasture’s
Gasoline mineral turpentine
Kleinknecht, et al.25
2
anti-GBM nephritis
Organic solvent vapors
1
anti-GBM nephritis
Stoddard solvent
1
subacute GN
Paint solvent
4
epimembranous GN
Solvents
1
epimembranous GN
?
Sperace
Heale, et al.
21
Klavis and Drommer
Daniell, et al.
22
26
Von Scheele, et al. Ehrenreicht, et al. Cagnoli, et al.
27
28
29
GBM = Glomerular basement membrane; RPGN = rapidly progressive glomerulonephritis; GN= glomerulonephritis
While these studies represent case reports, they suggest an association between exposure and the development of chronic glomerular disease. 20.4.4 CASE CONTROL STUDIES Several case-control studies have examined the role of organic solvent exposure in a population of patients with glomerulonephritis. A total of 14 case control studies examining human exposure to solvents and glomerulonephritis have been conducted and are documented here in Table 20.4.2.30 Table 20.4.2. Glomerulonephritis and organic solvents: Case-control studies summarized. [Data from reference number 30] Investigator Lagrue, et al. Bell, et al.
31,32
33
Ravnskov, et al. Ravnskov, et al.
35
Porro, et al.
4.9*, 5.2* increased
34
36
Increased risk factor
3.9* 2.8*
Investigator Nuyts, et al.
39
Zimmerman, et al. 41
Ravnskov
Finn, et al.
Increased risk factor 1.1*
40
increased increased
42
3.6*, 3.2* 43
3.9*
Van der Laan
1.1
Yaqoob, et al.37
aliphatic 15.5, halogenated 5.3, aromatic-oxygenated 2.0
Harrison, et al.44
Yaqoob, et al.38
increased
*P<0.05, statistically significant
8.9*
1358
Nachman Brautbar
The study by Lagure, et al,31,32 showed significantly increased risk of solvent related glomerulonephritis of 4.9. That this increased risk of glomerulonephritis follows a dose-response relationship was shown in the study of the populations examined by Ravnskov, et al.,34,35 Bell, et al.,33 Porro, et al.,36 Yaqoob, et al.,37,38 Nuyts, et al.,39 and demonstrates: 1) temporal relationship between exposure to solvents and the development of kidney disease, 2) a dose-response relationship, strongly showing the causal link between solvent exposure and glomerulonephritis. The study by Nuyts, et al.,39 examined a large population of 272 patients with chronic renal failure and assessed several occupational exposures, among those were hydrocarbons. The increased risk of chronic kidney disease in the form of renal failure in patients exposed to solvents was 5.45. The study of Askergren et al.45 looked into kidney functions in patients exposed to various organic solvents, specifically excretion of red blood cells in the urine in 101 patients exposed to solvents as compared to 39 non-exposed controls. Those who were exposed to organic solvents significantly excreted more cells than the ones who were not exposed. These studies showed the role for organic solvent exposure in the development of damage to the glomerules since excretion of red blood cells represents damage to the glomerules rather than tubules. That exposure to solvents is associated with glomerular damage rather than tubular damage fits with the various case reports and case-control studies and further suggest a plausible causal connection between exposure to industrial solvents and glomerular damage leading later on to chronic glomerulonephritis. The study by Bell et al.33 studied 50 patients who had organic solvent exposure and biopsy-proven proliferative glomerulonephritis. They have shown that none of these patients had evidence of any other systemic disease or preexisting infection, and compared those with 100 control subjects matched for age, sex and social class. This study is important since exposure assessment was done and showed significantly greater exposure scores in patients with glomerulonephritis compared to the control subjects. Furthermore, the degree of exposure was significantly higher in those patients who have more severe glomerulonephritis than those who have less severe glomerulonephritis, further indicative of a dose response relationship. This is a study which demonstrates significant statistical association, as well as dose response relationship between solvent exposure and kidney damage in the form of glomerular lesion and end-stage glomerulonephritis, ranging from mild to chronic severe glomerulonephritis. The study by Daniell et al.26 evaluated the risk of developing glomerular lesion associated with hydrocarbon exposure and showed a does-response relationship and variations in disease severity in relation to the exposure intensity. They showed an increase risk of developing glomerular nephritis, ranging from 2.8 to 8.9 fold increase as compared to the non-exposed population. There was clear temporal relationship between the exposure, absence of any other causes, a dose-response relationship which further validated the observations of Bell et al.33 and conclude that intense or long-term exposure (low-level but long-term or short-term and high levels) to commonly used industrial solvents played a causal role in the development of glomerular damage and chronic glomerulonephritis. In a comprehensive study, Yaqoob et al.46 performed a population study which looked into 3 groups of healthy men working in 3 different areas of a major car manufacturing plant. They have studied 3 groups, Group 1 included 112 paint sprayers exposed to a paint-based mixture of hydrocarbons, Group 2 which was composed of 101 transmission shop workers with exposure to petroleum-based mineral oils, and Group 3 which was comprised of 92 automated press operators with minimal background exposure to lubricating
20.4 Industrial solvents and kidney disease
1359
oils and who acted as internal controls. The 3 groups studied were comparable in age, duration of employment, duration of hydrocarbon exposure, and other factors. The cumulative exposure to hydrocarbons was evaluated. The hydrocarbon exposure scores were significantly higher in Groups 1 and 2, as compared to Group 3 (which served as an internal control and epidemiologically is a good working population control group, since this method takes into account the healthy worker). The principal hydrocarbons used throughout the period of time of the study were toluene, xylene, and n-butyl alcohol in paints and various petroleum fractions in the mineral oils. The study evaluated markers of kidney dysfunction in the subjects chronically exposed to hydrocarbons at the described work site. The authors concluded that paint exposure in the long-term is associated with renal impairment and micro-proteinuria without elevation in serum creatinine (which indicates that the kidney functions from a creatinine clearance point of view are still intact, and are less sensitive as a biological marker of glomerular damage) is a feature of workers chronically exposed to petroleum based mineral oils. The investigators also reported significant urinary excretion of protein which also indicated early glomerular damage in susceptible individuals. The authors concluded from these studies that chronic hydrocarbon exposure can be associated with renal impairment. They further concluded that the significance of the early markers of renal damage can predict progressive deterioration in renal functions. These data indicate that chronic hydrocarbon exposure may be associated with early and sub-clinical renal dysfunction leading to a chronic glomerulonephritis. Porro et al.36 performed a case referent study and they looked into a group of 60 patients with chronic glomerulonephritis established by biopsy, with no evidence of any other systemic diseases, and was compared to 120 control subjects who were not exposed to solvent vapors. Exposure assessment was based on scores from questionnaires. Exposure was significantly higher in the case group studies than in the reference control group for both total and occupational solvent exposure. They further found that the odds ratio of chronic glomerulonephritis for patient’s occupationally exposed to solvents was 3.9 and using a logistic regression model and they showed a dose-response effect of occupational exposure to solvents and glomerulonephritis. Histological studies of the 60 patients with chronic glomerulonephritis ruled out other systemic disease and demonstrated the whole-spectrum of glomerular diseases, the most common one is IgA nephropathy. When the sub-group of patients with IgA nephropathy and their matched controls were separately examined, the cases appeared to be significantly more exposed than the patients with other non-glomerular diseases such as kidney stones. Based on their findings, the investigators concluded that their results are in agreement with the hypothesis that the onset of glomerulonephritis could be related to a non-acute exposure to solvents even of light intensity. The work of De Broe et al.47 looked into occupational renal diseases and solvent exposure. They have concluded that the relation between hydrocarbon exposure and glomerulonephritis seems to be well-defined from an epidemiological point of view. They further show, in a case-control study of a group of patients with diabetic nephropathy, that hydrocarbon exposure was found in 39% of the patients with that particular form of kidney disease. They find that this was in agreement with the findings of Yaqoob et al.37 who found higher levels of hydrocarbon exposure in patients with incipient and overt diabetic nephropathy than in diabetic patients without clinical evidence of nephropathy. These data indicate a particular sensitivity of patients with diabetic kidney toward the damaging effects of the hydrocarbons. The findings of these investigators are agreement with the study of
1360
Nachman Brautbar
Goyer,48 who showed that existing renal diseases, particularly hypertensive and diabetic nephropathies, are clear risk factors predisposing to abnormal accumulation and excess blood levels of any nephrotoxic drugs and chemicals, as well as solvents. Indeed this observation makes a lot of scientific and clinical sense, since it is known that the ability of the kidney to excrete the breakdown metabolites of various materials including industrial solvents is reduced with any incremental reduction of kidney function, and there would certainly be more accumulation of these breakdown products, as well as the parental solvents in the kidney tissue, and as such, it makes sense that these individuals with underlying kidney disease such as hypertensive kidney disease, diabetic kidney disease, or interstitial kidney disease which may not yet be clinically overt, are at a significantly increased risk of developing chronic kidney disease as a result of the documented damaging effects of solvents on the kidney. 20.4.5. EPIDEMIOLOGICAL ASSESSMENT The epidemiological diagnostic criteria for most cases of end-stage kidney disease is deficient since no etiologic information is available in the majority of the cases. Fewer than 10% of the end-stage renal disease cases are characterized etiologically.49 Clinically, many patients are classified histologically such as glomerulonephritis, but little effort is made to look for toxic factors. Indeed, the majority of the clinicians seeing patients with end-stage kidney disease are not trained to look into occupational, environmental, or toxicological issues and end-stage renal disease. Many patients are listed as having hypertensive end-stage kidney disease and are presumed therefore to be “idiopathic” in origin, however, these cases may very well be the result of other industrial and/or environmental factors, among them, solvent exposure. Many of the problems in the epidemiological analysis is the result of a great reserve capacity of the kidney that can function relatively adequately despite slowly progressive damage. End-stage kidney disease is typically not diagnosed until considerable kidney damage has already occurred at the time when the patient seeks clinical attention. Furthermore, kidney biopsy and post-mortem examination, almost always find small kidneys, inadequate to help in the histopathological assessment, and therefore the etiology is either missed or is misclassified as “idiopathic” or “unknown”. Indeed the study by Stengel. et al.50 looked at organic solvent exposure and the risk of IgA nephropathy. These investigators have shown that the risk of IgA nephropathy is highest among the most exposed group to oxygenated solvents. The study by Yaqoob et al.37 showed an increased risk factor of 15.5 for development of glomerulonephritis in patients exposed to aliphatic hydrocarbons and a risk factor of 5.3 in patients exposed to halogenated hydrocarbons. These epidemiological data further supports observations made in the case reports, case studies and experimental animal studies. The epidemiological studies by Steenland et al.51 had evaluated the risks and causes of end-stage kidney disease and concluded that regular exposure to industrial solvents played a significant role in the development of chronic end-stage kidney disease. Based on the current literature from experimental animal studies, case reports, case-control studies, and epidemiological studies, one can conclude that the studies show: 1. Biological plausibility. 2. A temporal relationship between exposure to industrial solvents and the development of chronic kidney disease (glomerulonephritis). 3. A dose-response relationship. 4. Consistency of association.
20.4 Industrial solvents and kidney disease
1361
5. Statistical association in the majority of the studies. These criteria fulfill the Bradford-Hill criteria,52 and establish the basic criteria required for causation. 20.4.6. MECHANISM Immune-mediated mechanisms play a major role in the pathogenesis of glomerular disease, in general. In the vast majority of the cases, antigen-antibody reaction and immune complexes form in the kidney, mainly around the glomerular capillary wall and mesangium. Cellular antigens, both endogenous such as DNA and tumor antigens, as well as exogenous such as viral antigen hepatitis B and C, drugs, and bacteria have been shown to be causative factors in human glomerular immune-mediated diseases. The most common pathological process described in association with solvent exposure and chronic glomerular nephritis has been that of IgA nephropathy, Good Pasture’s syndrome, and proliferative glomerulonephritis. Unlike acute renal failure caused by hydrocarbons, where the renal damage is secondary to the nephrotoxins and mainly cause damage of the proximal tubule acute renal failure, the glomerular chronic renal failure, appears to be immunologically mediated. Among others, genetic factors may be involved in the pathogenesis of hydrocarbon induced nephropathy. It has been suggested that the propensity to develop this autoimmune disease depends on a combination of a genetic component and predilection, and environmental component.24 Individuals susceptible to glomerular or tubular injury by hydrocarbons may develop chronic kidney disease through three possible mechanisms. The first mechanisms is direct tubular toxicity which is commonly the cause of acute renal failure. While it is true that the initial injury of acute renal failure is directed toward the tubule of the nephron, glomerulonephritis may be the result of an autoimmune reaction to the tubulotoxins.53,54 The second mechanism mainly involves immunosuppression. Ravnskov,53 in a review of the pathogenesis of hydrocarbon associated glomerulonephritis, suggested that hydrocarbons are immunosuppressives and this effect is noted in several locations in the immunological cascade. This includes leukocyte mobility and phagocytosis suppression such as shown in the benzene effects in mice.55 This suppression of the normal immune response by hydrocarbons may play a role in the pathogenesis of immune-mediated glomerular lesions. The third mechanism involves alteration in membrane permeability. Good Pasture’s syndrome is mediated by antibodies reactive with the glomerular basement membrane and alveolar basement membranes. Antibodies in experimental models can usually bind to alveolar basement membranes in vitro by indirect immunofluorescence. Experimental studies suggested that hydrocarbons alter the permeability of pulmonary capillaries, thereby allowing anti-glomerular basement membranes to bind to the alveolar basement membranes.56 This etiology is further supported by the observation that differential sensitivity to exposures due to genetic factors since DR3 and DR4 antigens are more frequent in patients with toxic nephritis than in the general population.57,58 The study by Zimmerman et al.40 have shown that in 6 of 8 patients with Good Pasture’s syndrome had extensive occupational exposure to solvents ranging from 4 months to 10 years. The results of this study suggested that interaction between the inhaled hydrocarbons and the lung and kidney basement membranes could induce autoantibodies to these membranes. Goyer48 suggested an autoimmune mechanisms responsible for glomerular lesions following chronic exposure to solvents. Based on case studies and case reports, it is proposed that chronic exposure to low levels of solvents in susceptible individuals induces an initial cell injury sufficient to damage cell membranes and to
1362
Nachman Brautbar
provide the antigen triggering the immune response, accelerating a cascade of a reaction ending with glomerulonephritis. REFERENCES 1 2 3 4 5 6 7 8 9 10 11
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
Morrison, RR, and Boyd RN, Organic Chemistry, 5th Edition, Morrison RT, Boyd RN, Eds, Allyn & Bacon, Boston (1987). Domask, WG, Renal Effects Of Petroleum Hydrocarbons, Mehlman, MA, Hemstreet GP III, Thorpe JJ, Weaver NK, Eds, Princeton Scientific Publishers, Princeton, 1-25. Pederen, LM., Pharmacol Toxicol, 3, 1-38 (1987). Ervin ME, Clinical Management of Poisoning and Drug Overdose, Addad LM, Winchester JF, Eds, Saunders, Philadelphia, 771 (1983). Wolfsdorf J, J Pediatr, 88, 1037 (1976). Janssen S, Van der Geest S, Meijer S, and Uges DRA, Intensive Care Med, 14, 238-240 (1988). Smith TC, and Wollman H, The Pharmacological Basis of Therapeutics, Goodman A, Gilman L, Eds, MacMillan, New York, 260-275 (1985). Clayton GD, and Clayton EE (Eds), Patty’s Industrial Hygiene and Toxicology, Wiley, New York, 26 (1981) and 2C (1982). World Health Organization and Nordic Council of Ministers, Organic Solvents and the Central Nervous System, Environmental Health 5, Copenhagen and Oslo, 1-135 (1985). Browning E, Toxicity & Metabolism of Industrial Solvents, Elsevier Publishing Company, Amsterdam-London-New York (1965). Mehlman MA, Hemstreet GP III, Thorpe JJ, and Weaver NK, Series: Advances in Modern Environmental Toxicology: Volume VII- Renal Effects of Petroleum Hydrocarbons, Princeton Scientific Publishers, Princeton (1984). Carpenter CP, Geary DL Jr Myers, et al, Toxicol Appl Pharmacol, 41, 251-260 (1977). Gibson JE, and Bus JS, Ann NY Acad Sci, 534, 481-485 (1988). Thomas FB, and Halder CA, Holdsworth CE, Cockrell By, Renal Heterogeneity and Target Cell Toxicity, Back PH, Lock EA, Eds, Chichester, 477-480 (1985). Halder CA, Van Gorp GS, Hatoum NS, and Warne TM, Am Ind Hyg Assoc J, 47, 164-172 (1986). Coppor, et al, American Journal of Kidney Disease, 12, 420-424 (1988). Emancipator SN, Kidney International, 38, 1216-1229 (1990). Churchill DN, Fina A, and Gault MH, Nephron, 33, 169-172 (1983). Roy AT, Brautbar N, and Lee DBN, Nephron, 58, 385-392 (1991) Sperace GA, Am Rev Resp In Dis, 88, 330-337 (1963). Hele WF, Matthisson Am, and Niall JF, Med J Aust, 355-357 (1969). Klavis G, and Drommer W, Arch Toxicol, 26, 40-50 (1970). Beirne GJ, and Brennan JT, Arch Environ Health, 25, 365-369 (1972). D’Apice AJF, et al, Ann Intern Med, 88, 61-62 (1980). Kleinknecht D, et al, Arch Intern Med, 140, 230-232 (1980). Daniell WE, Couser WG, and Rosenstock L, JAMA, 259, 2280-2283 (1988). Von Scheele C, Althoff P, Kempni V, and Schelin H, Acta Med Scan, 200, 427-429 (1976). Ehrenreicht T, Yunis SL, and Churg J, Environ Res, 24, 35-45 (1977). Cagnoli L, et al, Lancet, 1068-1069 (1980). Brautbar N and Barnett A, Environmental Epidemiology and Toxicology, 2, 163-166 (1999). Lagrue G, et al, J Urol Nephrol, 4-5, 323-329 (1977). Lagrue G, et al, Nouv Press med, 6, 3609-3613 (1977). Bell GM, et al, Nephron, 40, 161-165 (1985). Ravnskov U, Forsbergg B, and Skerfving S, Acta Med Scand, 205, 575-579 (1979). Ravnskov U, Lundstrom S, and Norden A, Lancet, 1214-1216 (1983). Porro A, et al, Br J Ind Med, 49, 738-742 (1992). Yaqoob M, Bell GM, Percy D, and Finn R, Q J Med, 301, 409-418 (1992). Yaqoob M, et al, Diabet Med, 11, 789-793 (1994). Nuyts GD, et al, Lancet, 346, 7-11 (1994). Zimmerman SW, Groehler K, and Beirne GJ, Lancet, 199-201 (1975). Ravnskov U, Acta Med Scand, 203, 351-356 (1978). Finn R, Fennerty RG, and Ahmad R, Clin Nephrol, 14, 173-175 (1980). Van der Laan G, Int Arch Occup Envion Health, 47, 1-8 (1980).
20.5 Lymphohematopoietic study of workers exposed to benzene
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
1363
Harrison DJ, Thompson D, and MacDonald MK, J Clin Pathol, 39, 167-171 (1986). Askergren A, Acta Med Scand, 210, 103-106 (1981). Yaqoob M, et al, Q J Med, 86, 165-174 (1993). De Broe ME, D’Haese PC, Nuyts GD, and Elseviers MM, Curr Opin Nephrol Hypertens, 5, 114-121 (1996). Goyer RA, Med Clin of North America, 74, 2, 377-389 (March 1990). Landrigan P, et al, Arch Environ Health, 39, 3, 225-230 (1984). Stengels B, et al, Int J Epidemiol, 24, 427-434 (1995). Steenland NK, Thun MJ, Ferguson BA, and Port FK, AJPH, 80, 2, 153-156 (1990). Hill AB, The Environment and Disease: Association or Causation? President’s Address. Proc Royal Soc Med, 9, 295-300 (1965). Ravnskov U, Clin Nephrol, 23, 294-298 (1985). Tubbs RR, et al, Am J Clin Pathol, 77, 409 (1982). Horiguchi S, Okada H, and Horiuchi K, Osaka City Med J, 18, 1 (1972). Yamamoto T, and Wilson CB, Am J Pathol, 126, 497-505 (1987). Emery P, et al, J Rheumatol, 11, 626-632 (1984). Batchelor JR, et al, Lancet, i, 1107-1109 (1984).
20.5 LYMPHOHEMATOPOIETIC STUDY OF WORKERS EXPOSED TO BENZENE INCLUDING MULTIPLE MYELOMA, LYMPHOMA AND CHRONIC LYMPHATIC LEUKEMIA Nachman Brautbar University of Southern California, School of Medicine, Department of Medicine, Los Angeles, CA, USA
20.5.1 INTRODUCTION Benzene, is an aromatic hydrocarbon and historically has been produced during the process of coal tar distillation and coke production, while today benzene is produced mainly by the petrochemical industry. Based on the National Institute of Occupational Safety and Health (NIOSH), in the United States, it has been estimated that in 1976 two million Americans were exposed occupationally to benzene.1 Worldwide production of benzene is approximately 15 million tons2 and the production in the United States is estimated to be increasing at least 3% annually,3 approaching 6 million tons of benzene produced in the United States in 1990 and 6.36 million tons produced in 1993.2 Benzene has been described as a clear, colorless, non-corrosive, and flammable liquid with a strong odor. Benzene is used as an excellent solvent and degreasing agent, and as a basic aromatic unit in the synthetic process of other chemicals.4 Exposure to benzene in the occupational setting most commonly occurs in the chemical, printing, rubber, paint, and petroleum industry. Among other sources of exposure to benzene, non-occupational exposures in the form of cigarette smoking and exposure to gasoline and its vapors during fueling motor vehicles.5 20.5.2 ROUTES OF EXPOSURE The major route of exposure to benzene is inhalation through the lungs of benzene vapors, however skin absorption of benzene has been shown to be significant depending on the circumstances of the exposure such as time of contact between the benzene and the skin.1,6 Benzene absorption, as other solvents, through the skin is enhanced if the skin condition is altered by either disease or loss of skin or cracking of the skin.4 The dermal absorption of benzene deserves much more attention than previously described in various texts of toxicol-
1364
Nachman Brautbar
ogy and occupational medicine, as well as environmental and industrial health. Dermal absorption of benzene in workers who use either toluene containing benzene or other solvents containing benzene, is a significant factor in calculating dosimetry and absorption of benzene and can be calculated utilizing standard accepted methodology. A recent study by Dr. Brenner et al.6 described chronic myelogenous leukemia due to skin absorption of benzene as a contaminant of other solvents. The investigators in that study6 concluded that the total benzene absorbed dose via skin and inhalation was equivalent to an accumulated vapor exposure of 196.4 + 42 ppm-years. Dermal exposure accounted for 97% of the total absorbed dose of benzene. Inhalation of benzene from occupational, smoking and ambient non-occupational sources accounted for only 3% of the benzene dose. The authors presented the reports of dermal absorption of benzene in the following table. Table 20.5.1. Summary of benzene dose expressed as equivalent ppm-years. [Adapted, by permission, from D. Brenner, Eur. J. Oncol., 3(4), 399-405, 1998.] Solvent
Dermal
Occupational inhalation
Cigarette smoke
Ambient inhalation
Total
Toluene, MEK, Acetone
170.4
19.2
0.05*
0.29
189.9
1998 Case 1 Mineral spirits
41.1
17.8
0.1*
0.23
59.3
1998 Case 2 Refinery process streams
19.0
4.6
1.4
0.4
196.4
Case 1997
*Second hand cigarette smoke
Dose, ppm years
Therefore, workers who are exposed to solvents containing benzene should be evaluated for skin absorption dosimetry, in addition to other sources such as inhalation, to address the range of levels of exposure. Therefore, dermal absorption of bencummulative inhalation dose zene, especially in connection with benzene cummulative dermal dose as a byproduct in other solvents, is of extreme importance in dosimetry analysis. 200 Once benzene reaches the blood, it is metabolized mainly in the liver. The metabolic products are excreted in the urine 150 within 48 hours from absorption. Several metabolites have been found in the urine after benzene exposure, among those are phe100 nol, quinone, hydroquinone, and muconic acid.7 The liver utilizes the cytochrome P450 and oxidization system for the metab50 olism of benzene.8,9 Specific cellular toxic effects from benzene have been described and those in0 clude, among others, the central nervous 63-66 67-72 73-78 79-83 84-91 92-93 system (doses of over 100 ppm), liver, kidTime period ney, skin, immunological, and carcinoFigure 20.5.1. Cumulative Dose of benzene over 30 genic. The various toxicological effects of years. [Adapted, by permission, from D. Brenner, Eur. J. benzene will not be discussed in this chapter Oncol., 3(4), 399-405, 1998.]
20.5 Lymphohematopoietic study of workers exposed to benzene
1365
since the scope of this chapter is the hematopoietic effects, the reader is referred to other sources.3 20.5.3 HEMATOPOIETIC EFFECTS OF BENZENE Benzene is a proven human carcinogen. The toxicity of benzene has been known since the 19th Century when aplastic anemia was first reported.4,10 Indeed the causal link between benzene and bone marrow toxicity in the form of hematoxicity and bone marrow suppression was described already in 1897.11 In 1928, Delore et al. described leukemia as a result of benzene exposure.12 In 1932 Lignac13 reported lymphoblastoma in association with benzene exposure. Several studies have reported the association between exposure to benzene and hematopoietic toxicity and leukemia.14,15 Acute myeloid leukemia has been the most frequent form of leukemia found to be related to benzene exposure. Other forms of leukemia have been described in association with benzene exposure, such as erythroleukemia, thrombocytopenia, acute myeloid leukemia, myelodysplastic syndrome, acute lymphoblastic leukemia, chronic lymphocytic leukemia, and Hodgkin’s and non-Hodgkin’s lymphoma. As a result of the high toxicity of benzene the American Petroleum Institute in their paper on benzene exposure in 1948 have concluded that the only safe level of exposure to benzene is no exposure at all.16 The language utilized was as follows, “In as much as the body develops no tolerance to benzene, and as is there is a wide variation in individual susceptibility, it is generally considered that the only absolutely safe concentration for benzene is zero.” The following hematological conditions have been described in association with benzene.17-40 1. Acute myelogenous leukemia. 2. Erythroleukemia. 3. Aplastic anemia. 4. Acute monocytic leukemia. 5. Chronic myelogenous leukemia. 6. Myelofibrosis and myeloid metaplasia. 7. Thrombocythemia. 8. Acute lymphoblastic leukemia. 9. Chronic lymphocytic leukemia. 10. Lymphomas and related disorders. 11. Multiple myeloma 12. Myelodysplastic syndrome. 20.5.4 CARCINOGENIC EFFECTS OF BENZENE Several well conducted epidemiological scientific studies and data have provided the epidemiological basis for benzene as a hematopoietic and lymphopoietic cancer. In his paper entitled “Benzene Health Effects”, Mehlman described a wide range of the hematotoxicity of benzene.41 Nilsson et al.42 described leukemia, lymphoma and multiple myeloma in seamen exposed to benzene in tankers. In this study, an increased incidence of lymphatic and hematopoietic malignancies was described and, while it is true that the cargo vapors from gasoline and other light petroleum products and chemicals have been studied, benzene exposure during loading, unloading and tank cleaning operations was concluded to be the likely source of the carcinogenic exposure. Rinsky et al.43 described various hematological malignancies in their study of benzene exposure and showed that the overall standardized
1366
Nachman Brautbar
mortality ratio for leukemia and multiple myeloma were increased significantly. The investigators of this study concluded that there is a quantitative association between benzene exposure and development of leukemia. Wong44 evaluated a mortality study of chemical workers occupationally exposed to benzene and found a significantly increased risk for lymphohematopoietic malignancies. Linet et al.45 studied hematopoietic malignancies and related disorders among benzene exposed workers in China and showed a wide spectrum of hematopoietic malignancies. Song-Nian Yin et al.46 in a cohort study of cancer among benzene exposed workers in China, studied workers employed in a variety of occupations and showed a statistically significant increased deaths among benzene exposed subjects for leukemia, malignant lymphoma, neoplastic diseases of the blood, and other malignancies. The rates were significantly elevated for the incidence of lymphohematopoietic malignancies risk ratio of 2.6, malignant lymphoma risk ratio of 3.5, and acute leukemia risk ratio 2.6. A significant excess risk was also found for aplastic anemia and myelodysplastic syndrome. These investigators concluded that employment in benzene exposure occupations is associated with a wide spectrum of myelogenous and lymphatic malignant diseases and related disorders of the hematopoietic lymphatic system. Hayes et al.47 in one of the largest epidemiological studies on benzene exposure, showed a wide spectrum of hematological neoplasms and their related disorders in humans. The risk for these conditions is elevated at average benzene exposure levels of less than 10 ppm. These investigators further concluded that the pattern of benzene exposure appears to be important in determining the risk of developing specific diseases. Wong48 studied a cohort of 7,676 male chemical workers from seven plants who were occupationally exposed continuously or intermittently to benzene for at least 6 months, and compared them to a group of male chemical workers from the same plant who had been employed for at least 6 months during the same period but were never occupationally exposed to benzene and showed a significantly increased risk of lymphohematopoietic malignancies. In experimental animals benzene has been shown to be associated, in rats, with cancers of zymbal gland, oral cavity, nasal cavities, skin, forestomach, mammary gland, Harderian gland, preputial glands, ovary, uterus, angiosarcoma of liver, hemolymphoreticular neoplasia, lung cancers and leukemia.49 The ability of benzene metabolites to effect lymphocytic growth and function in vitro, is shown to correlate with the oxidation capacity and concentration of the metabolites at the target site. Benzene also effects macrophages, as well as lymphocytes.36,50,51 Kalf and Smith52,53 have shown that benzene exposure reduces the ability of marrow stromal cells to support normal stem cell differentiation. From these experimental animal studies, and in vitro studies a wide range of bone marrow effects of benzene metabolites is shown. It has been concluded that the hematotoxicity of benzene depends on the breakdown metabolites and can effect the stem cell at any point in time, for instance myeloid, erythroid, macrophages, lymphocytic stem cells, and therefore benzene has been named as a pleural potential stem cell toxicant.54 In addition to carcinogenic effects, animal studies have shown the effects of benzene exposure on the immune system. Reid et al.55 showed a significant decrease in splenic cell proliferation in mice exposed to benzene for 14 days. Experimental animal studies also reported reduced circulating white blood cells, as well as changes in spleen morphology and weight in various experimental animal studies.54 These experimental animal studies further support the observation from 1913 by Winternits and Hirschfelder56 that rabbits exposed to
20.5 Lymphohematopoietic study of workers exposed to benzene
1367
benzene showed an increased susceptibility to pneumonia and tuberculosis. The experimental animal data and the epidemiological studies clearly show that 1) benzene is a carcinogen for the lymphohematopoietic system, 2) benzene has a direct effect on the immune system, 3) benzene has a direct effect on the early development of the blood cells, and 4) benzene is a pluripotent hematological carcinogen. 20.5.5 RISK ASSESSMENT ESTIMATES The United States EPA has used several databases in their estimates for benzene exposure and risk. (Environmental Protection Agency, 5.0 Benzene, 5.1. Chemical and Physical Properties, EPA, 1988) The data utilized by the EPA to assess the risk included the study by Rinsky et al. in 198157 where the duration of exposure was at least 24 years and exposure levels are between 10 to 100 ppm (8 hour TWA) with a statistically significant increase incidence of leukemia. The study of Ott et al.57 in 1978 showed levels of anywhere from 2 to 25 ppm (8 hour TWA) with increased incidence of leukemia; and Wong et al.57 1983, where the exposure was at least 6 months, levels were from 1 ppm to 50 ppm, and there was a statistically significant increase in the incidence of leukemia, lymphatic and hematopoietic cancers. The International Agency for Research on Cancer (IARC) has classified benzene as a Group 1 carcinogen. 58 A Group 1 carcinogen is defined as an agent that is carcinogenic to humans. This classification is based on sufficient evidence for carcinogenicity in humans. IARC based this conclusion on the fact that numerous case reports and follow-up studies have suggested a relationship between exposure to benzene and the occurrence of various types of leukemia. In addition, IARC considers the evidence for carcinogenicity to animals to be sufficient. No unit risk was determined by IARC for benzene. The regulatory agencies in the estimated risk of increased benzene related cancers rely mainly on the study by Rinsky, 1987, which concluded that the mean annual cumulative exposure level of less than 1 ppm accumulated over 40 years working life-time would not be associated with increased death from leukemia. This epidemiological study showed an exponential decrease in the risk of death from leukemia which could be achieved by lowering occupational exposure to benzene. According to the model derived in this study, a worker occupationally exposed to benzene at an average exposure level of 10 ppm for 40 years would have an increased risk of death form leukemia 154.5. If the average exposure was lowered to 1 ppm, that excess risk would decrease to 1.7. At 0.1 ppm times 40 years cumulative exposure the risk be virtually equivalent to background risk,26 Infante et al.59 have shown a relative risk of 5.6 with an estimated cohort exposure of 10 to 100 ppm over 8.5 years average, and Vigliani60 showed a relative risk of 20 estimated cohort exposure to 200 to 500 ppm over 9 years average, and Aksoy showed a relative risk of 25 with an estimated cohort exposure of 150 to 210 ppm over 8.7 years average.20,40 These studies clearly show that the risk of developing lymphohematopoietic cancers is significant, and that benzene is carcinogenic from an epidemiological point of view at very low levels of 0.1 ppm. 20.5.6 LEVELS OF EXPOSURE Regulatory levels of exposure to various chemicals, among them solvents and benzene, have been a subject for constant pressure from industry manufacturers on one hand, regulatory agencies, health care, and patients on the other hand. The most common question asked is “Is there a safe level of exposure to benzene?” The answer to that question has been given by the American Petroleum Institute in their paper on benzene, 1948,16 and their statement
1368
Nachman Brautbar
Figure 20.5.2. Extrapolations of levels of exposure to benzene. [Adapted, by permission, from R.A. Rinsky, New England J. Med., 1987.]
Figure 20.5.3. Extrapolations of levels of exposure to benzene. [Adapted, by permission, from R.A. Rinsky, New England J. Med., 1987.]
that the only safe level of exposure to benzene is no exposure at all. Rightly so, the paper based that opinion on the fact that the body develops no tolerance to benzene and there is wide variation in individual susceptibility and therefore the only absolutely safe concentration for benzene is zero. This approach has been confirmed from a point of view of epidemiological studies and experimental animal studies showing benzene to cause cancer in experimental animals, and case reports and epidemiological studies in humans. Based on epidemiological studies and extrapolation from experimental animal studies, IARC’s position is that a linear regression line should be applied for the dose response crossing the zero point for low level exposure of benzene. The EPA concurred that at low levels of exposure (since no epidemiological studies are available at low levels of exposure) linear dose response is indicated. The modalities of exponential dose response relationship for low levels
20.5 Lymphohematopoietic study of workers exposed to benzene
1369
is not applicable here based on the most recent EPA and IARC positions.57,58 This linear, non-threshold model assumes that every increment of dose is accompanied by a commensurate increment in the excess cancer risk. The use of this toxicological model allows extrapolation of risks from relatively high dose levels, where cancer responses can be measured, to relatively low dose levels, where such risks are too small to be measure directly through epidemiological studies.61 Figures 20.5.2 and 20.5.3 demonstrate the extrapolation from high levels of exposure to low levels of exposure utilizing the linear modality. Indeed, in its most recent publication the U.S. EPA, 10/14/98,62 further supports that approach and the panel members who evaluated the data felt that for the leukemogenic effects of benzene the linear model is consistent with the spirit of the proposed cancer risk assessment guidelines. 20.5.7 CELL TYPES: HEMATOLYMPHOPROLIFERATIVE EFFECTS OF BENZENE The hematopoietic cell type toxicity from benzene have been described in animal data, case reports, case studies, animal data, and epidemiological data. Essentially, Wong in his OSHA testimony63 concludes that for the continuously exposed group the lymphohematopoietic cancer risk ratio was 3.2 with a statistical significance of p<0.05, the risk ratio for non-Hodgkin’s and other lymphopoietic cancer, i.e., all lymphopoietic cancers minus Hodgkin’s disease, for the continuously exposed group was 3.77. The data demonstrated a statistical significant dose response relationship between cumulative exposure to benzene and mortality from all lymphopoietic cancer combined with leukemia. Wong further stated that it would be appropriate to combine lymphoma and leukemia in some of the analyses. This approach has also been recommended recently by other investigators64 and therefore the agency conducting the hearing felt that the analysis based on the revised grouping of lymphopoietic cancers with Hodgkin’s disease separated out was appropriate.63 In the documents submitted to the OSHA hearing, Wong concluded that a dose response relationship for all lymphatic and hematopoietic cancers has been demonstrated. Based on the case reports, case studies, and epidemiological studies, the sub-classification to cell types is indicated from a medical point of view to treat various hematopoietic diseases and cancers with the various appropriate treatments per each type of cell injured, however the data from the experimental data and clinical analysis of benzene cases clearly show that benzene causes damage to the stem cell and therefore it is a pluripotent toxin, causing a wide range of lymphohematopoietic malignancies. 20.5.8 EPIDEMIOLOGICAL STUDIES Wong,48 studied a cohort of 7,676 male chemical workers from seven plants who were occupationally exposed continuously or intermittently to benzene for at least 6 months, and compared them to a group of male chemical workers from the same plant who had been employed for at least 6 months during the same period but were never occupationally exposed to benzene. When the group with no occupational exposure was used for direct comparison, the continuously exposed group experienced a relative risk from lymphohematopoietic cancer of 3.2 with a statistical significance of p<0.05. That paper further concluded that the medical problems are replete with reports documenting the transition from certain lymphomas and multiple myelomas to leukemias. It was concluded that the transitions or progressions from lymphoma to leukemia are further complicated by the historical changes in nomenclature and in diagnostic overlap between the 2 disorders. It was felt,
1370
Nachman Brautbar
based on the work of others, that the major clone in chronic myelocytic leukemia affected cells capable of becoming lymphocyte, granulocyte, and erythrocyte differentiations leading to the conclusion that transformation events occur at an early multipotent stem cell level. Nilsson et al.42 investigated Swedish seamen, 20-64 years of age, who had been exposed to cargo vapors for at least 1 month on chemical or product tankers, had an increased risk of lymphatic and hematopoietic malignancies odds ratio of 2.6 with 95% confidence interval, with a significant exposure response relation. The odds ratio was increased for non-Hodgkin’s lymphoma at 3.3 with 95% confidence interval and was statistically significant. Rinsky et al.43 studied a cohort of 11,065 white men with at least 1 ppm per day of cumulative exposure to benzene. They have demonstrated that there was a statistical significant increase in death from all lymphatic and hematopoietic neoplasms, 15 observed versus 6.6 expected standard mortality ratio which is 227, 95% confidence interval, further demonstrating that benzene is toxic to all cell types. Hayes et al.47 studied a cohort of 74,821 benzene exposed and 35,805 unexposed workers from 1972 until 1987 in 12 cities in China. By and large this is the largest and most significant cohort of benzene workers studied and published. The investigators found that 1) benzene exposure is associated with a spectrum of hematological neoplasms, 2) workers with 10 or more years of benzene exposure had a risk ratio of developing non-Hodgkin’s lymphoma of 4.2 with 95% confidence interval, and the development of this neoplasm was linked most strongly to exposure that had occurred at least 10 years before the diagnosis, and 3) the risk for the combination of acute non-lymphocytic leukemia and related myelodysplastic syndromes was significantly increased among those with more recent benzene exposure. These studies confirm the previous studies proving that the damage from benzene is to all cell type. Linet et al.45 studied hematopoietic malignancies and related disorders among benzene exposed workers in China and showed a wide range of hematopoietic malignancies. Yin et al.46 examined a large cohort of benzene workers and concluded that benzene exposed workers have a statistically significant excess death due to leukemia, risk ratio of 2.3 with 95% confidence interval; malignant lymphoma, risk ratio of 4.5 with 95% confidence interval; and non-neoplastic diseases of the blood. In summary, these epidemiological studies published in the peer-reviewed scientific literature and relied on by scientific and governmental agencies clearly show 1) significant statistical association between benzene exposure and lymphohematopoietic cancers of all cell types, 2) an increased risk and/or increased standard mortality rate over a factor of 2 in patients exposed to benzene with the development of non-Hodgkin’s lymphoma, leukemias, and other lymphohematopoietic malignancies, and 3) benzene is carcinogenic with a linear dose response demonstrating no threshold. 20.5.9 SOLVENTS AND BENZENE Solvents commonly used in the industry have been shown to contain benzene. Elkins, et al.65 found that from time to time analyzed solvents for benzene content showed anywhere from 1% to 2% benzene. In that paper, which was published in 1956, the authors state that the TLV value from a regulatory point of view, at that time, was 35 ppm compared with 100 ppm previously. According to their calculations, they found that a benzene content below 3.5% will be necessary, for instance in solvents containing naphtha, hexane, and toluene, otherwise the permissible level for benzene vapor will be exceeded over 35 ppm, which
20.5 Lymphohematopoietic study of workers exposed to benzene
1371
we now know is extremely and significantly higher than the standard allowed today from a regulatory point of view. At the request of the petrochemical companies, the authors decided to reevaluate the content of benzene in solvents and for this purpose a total of 8 samples of low boiling petroleum naphtha were obtained. After utilizing methodology which included, among others, mass spectrometry for benzene content, the authors concluded that, in general, the benzene content of solvents ranged from 1% to 4% in volume. They have further shown that in the air of one plant where hexane with a relatively low benzene content (1.5%) was used as a solvent in a fabric-spreading operation, a benzene vapor concentration of 1 ppm was found. Since exposure to benzene is cumulative, if a worker is exposed to hexane containing 1.5% benzene, with both inhalation and skin contact, the cumulative exposure over a certain period of time increases the risk of developing benzene related cancers as described by the Rinsky model. The study by Pagnotto,66 looked at and analyzed 32 naphtha solvents. The benzene concentration ranged from 1.5% to 9.3% by weight. Excessive benzene exposure was found at 3 out of 4 plants during their operations on a daily basis. On one occasion the concentration of benzene vapor was as high as 125 ppm (extremely high), and the urinary phenol excretion of the workers in these 3 plants were the highest that these investigators report ranging from 370 to 917 mg per liter of urine. These study indicate that solvents which contain benzene, even at levels of 1.5% per volume, can be associated with significant atmospheric exposure to benzene, shown as causing human exposure with significant excretion of phenol in the urine indicative of heavy benzene exposure. The investigators recommended additional ventilation, and on a follow-up visit the benzene exposure was found to have been reduced to about 70 ppm with urinary phenols of less than 70 mg per liter, still significantly elevated and considered a significant risk. These investigators also looked at blood examinations of 47 men at these plants. Five employees showed lower hemoglobin. One man showed a low hemoglobin at the age of 28, having been employed for 3 years in the environment preparing mixes for the saturating machine. While leukemia was suspected due to bone marrow disease, the patient was treated with iron and recovered. The authors conclude that excessive benzene exposures is consistently found on saturators using naphtha containing more than 3% benzene. This study further shows the importance of assessing benzene concentration is other solvents, from a dosimetry point of view. The manuscript entitled A Recommended Standard for Occupational Exposure to Refined Petroleum Solvents from the U.S. Department of Health, Education and Welfare, Public Health Service, Center for Disease Control, National Institute for Occupational Safety and Health, July 1977,67 recommended standards to be applied to occupational exposure of workers to the following refinery petroleum solvents: petroleum ether, rubber solvent, varnish maker’s and painter’s naphtha, mineral spirits, Stoddard solvents, and kerosene are all included in the term refined petroleum solvents. According to these standards petroleum ether and rubber solvents which contain 1.5% benzene, varnish maker’s and painter’s naphtha which contain 1.5% benzene, mineral spirits which contain 13-19% aromatics, Stoddard solvent which contains 0.1% benzene, 140 Flash Aliphatic Solvent which contains 0.7% benzene, kerosene. NIOSH indicated that some of the refined petroleum solvents contain aromatic hydrocarbons including, in some cases, benzene. Standards were applied, among others, to reduce the benzene exposure. Among others, the use of respirators and skin protective devices were required to protect from the effects of the solvents, as well as the benzene component.67 In his testimony in front of the Occupational Safety and Health
1372
Nachman Brautbar
Administration,68 Proctor testified that, among others, refining operations are continuously changing. Many refineries obtain crude of differing characteristics from various producing areas which sometimes must be processed individually due to crude incompatibility and produce requirements. This means the operation of a crude fractionation unit is altered frequently; a single crude run may be as short as 2 days. Consequently, the crude tankage, crude fractionation units, and all downstream processing units frequently contain benzene. Proctor further stated that it should be clear by now that benzene is a naturally occurring compound in crude oil and is also found in the catalytic and cracking process, and therefore will always be a contaminant of these solvents. Benzene levels in gasoline today are running about 1.1% on the average across the nation but occasionally may reach 4% on individual samples. Reduction of benzene levels in gasoline is technically possible through employment of a number of physical processing schemes to the various gasoline component streams. He further testified that any attempt of physical separation of hydrocarbons, such as distillation, solvent extraction, or adsorption, the separation is not 100% complete. Therefore, some residuals of benzene will always be present in the remaining fraction. Therefore he recommended that benzene should be converted to cyclohexane by hydrogenation which would require an expensive catalyst, expensive high pressure reaction vessels, and consumption of valuable hydrogen. The testimony further indicates that it is believed that it is almost physically impossible to reduce these streams below 0.1% benzene. The 1978 OSHA69 indicate that “The record establishes that there is a wide variation in the benzene content of petroleum solvents used in the rubber, paints, coatings, adhesives, sealants, and other downstream industries. As reported by Smith, reporting on behalf of MCA,69 the benzene content of petroleum solvents of all types generally range from under 0.1% to 4%. Data submitted by downstream industries confirms that benzene is present in virtually all petroleum solvents, at levels which approach and even exceed 3.5% in some cases.” It was stated that in the rubber industry, solvent benzene content appears to range from 0.1% to 0.7% or slightly higher. Similarly, solvents used by adhesive manufacturers show broad variations from less than 0.1% to 3.5%.69 Representatives of the paint industry report variations from under 0.1% to as high as 3.7%.69 Smith in his testimony emphasized that solvent benzene content is likely to vary substantially among supplies, among different plants of the same supplier, and among deliveries from the same plant. Because refinery processes are not designed to precisely control benzene content, variations will inevitably occur. These data clearly show that benzene contents in solvents are difficult to control and vary depending on the sources, processes and therefore solvent exposure must take into account the level of benzene concentration in these solvents. These data, taken together with the most recent study of Brenner et al.6 show that industrial toluene solvent does contain benzene and contributed significantly to the exposure via the skin to benzene. One must remember the importance of benzene exposure through the use of solvents produced through the petrochemical refining processes. 20.5.10 GENETIC FINGERPRINT THEORY Benzene and its metabolites have long been known to cause chromosomal aberrations of various types of cell cultures of exposed humans. (To be discussed in Chapter 20.6 in this book entitled as Benzene Exposure and Sister Chromatoid Changes.) While it is true that genetic changes have been described and frequently effect chromosomes 5 and 7, and others, there is no scientific evidence that these are required for the diagnosis of benzene exposure related cancers. Specifically, many cases of patients who have been exposed to
20.5 Lymphohematopoietic study of workers exposed to benzene
1373
benzene and have developed hematopoietic malignancies do not have changes in chromosomes, therefore the chromosomal changes cannot be used as a “genetic fingerprint”. Indeed, Irons’ publication in the Journal of Toxicology and Environmental Health70 concluded that the significance of these chromosomal alterations with respect to bone marrow damage or leukemogenesis of benzene is unclear. It is not possible today to determine whether leukemia is caused by benzene based on changes in chromosomes, specifically chromosomes 5 and 7. Smith from the University of California at Berkley, who is a leading authority in the biological markers of benzene exposure, opined that the data which supposedly suggest that one must have changes in chromosomes 5 and 7 to assume benzene causation is unreliable and obsolete.71 In summary, while it is true that benzene and related products have been described with changes in chromosomes, DNA adducts, and cell cycle, by no means can they not be used as a diagnostic tool to address benzene causation or not. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
18
19
20 21 22 23 24 25 26 27 28 29
NIOSH, Revised Recommendation for Occupational Exposure Standards for Benzene, Cincinnati, OH, DHEW Publications (NIOSH), 76-76-137-A (1976). Fishbein L, Scan J Work Environ Health, 8, Supplement 1, 5-16 (1992) ATSDR Toxicological Profile for Benzene- Update, U.S. Department of Health & Human Services, Atlanta, Georgia (1996) Browning E. Toxicity and Metabolism of Industrial Solvents, Amsterdam, Elsevier Publishing, Chapter 1 (1965) Wallace, LA, Environ Health Perspect, 82, 165-169 (1999) Brenner D, et al., Eur J Oncol, 3(4), 399-405 (1998) Henderson RF, et al., Environ Health Perspect, 82, 9-17 (1989). Schneider CA, et al., Toxiocology & Applied Pharmacology, 54, 323-331 (1980) Schneider R, Damitriadis E, and Guy R, Environ Health Perspect, 8, 31-35 (1989) Goldstein BD, Adv Mod Environ Toxicol, 4, 51-61 (1983) Santesson CG, Arch Hyg,, 31, 336-349 (1897) Delore P and Borgomano C, Journal Medicine Lyon, 9, 227-233 (1928) Lignac GOE, Krankeitsforsch, 9, 426-453 (1932) Vigliani EC, and Saita G, New England Journal of Medicine, 271, 872-876 (1974) Aksoy M, Erdem S, and Dincol G, Blood, 44, 837-841 (1974) API Toxicological Review, Benzene, September, (1948) Maltoni C, Myths and Facts in the History of Benzene Carcinogenicity : in Advances in Modern Environmental Toxicology, Volume IV, M. Mehlman (Ed). Princeton Scientific Publishing Company, Princeton, 1-15 (1983) NIOSH, Criteria for a Recommended Standard. Occupational Exposure to Benzene. U.S. Department of Health, Education and Welfare, Public Health Service, Center for Disease Control, Cincinnati, OH, Pub. No.74-137 (1974) NIOSH, Occupational Safety and Health Guidelines for Benzene: Potential Human Carcinogen. U.S. Department of Health and Human Services, Public Health Services, Cincinnati, OH, Pub No. 89-104, Supp II (1988) Aksoy M, Environ Res, 23, 181-190, (1980) Aksoy M, Environ Health Perspect, 82, 1931-198 (1989) USEPA, Ambient Water Quality Criteria for Benzene, Environmental Criteria and Assessment Office, Cincinnati, OH, EPA 440/5-80-018, NTIS PB81-117293 (1980) Goldstein BD and Snyder CA, Environ Sci Res, 25, 277-289 (1982) IARC and IACR, In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans: Some Industrial Chemicals and Dyestuffs. IARC, Lyon, France, 28, 183-225 (1982) Aksoy M, et al., Br J Ind Med, 44, 785-787 (1987) Rinksy RA, et al., N Engl J Med, 316, 1044-1050 (1987) Goldstein BD, Benzene Toxicity, In: Occupational Medicine: State of the Art Reviews, 3(3), 541-554, NK Weaver (ed), Hanley and Belfus, Inc, Philadelphia PA (1988) Paci E, et al., Scand J Work Environ Health, 15, 313-318 (1989) Rinksy RA, Environ Health Perspect, 82, 189-192 (1989)
1374
Nachman Brautbar
Ciranni R, Barale R, and Adler ID, Mutagenesis, 6, 417-421 (1991) Cox LA, Risk Anal, 11, 453-464 (1991) Ruis MA, Vassallo J, and Desouza CA, J Occup Med, 33, 83 (1991) Midzenski MA, et al., Am J Ind Med, 22, 553-565 (1992) ATSDR Toxicological Profile for Benzene- Update, U.S. Department of Health & Human Services, Atlanta, Georgia (1993) 34 Medinsky MA, Schlosser PM and Bond JA, Environ Health Perspect, 102 Suppl 9:119-124 (1994) 35 Travis LB, et al., Leuk Lymphoma, 14(1-2), 92-102 (1994) 36 Niculescu R and Kalf GF, Arch Toxicol, 69(3),141-148 (1995) 37 Wintrobe MM, Lee GR and Boggs DR, Wintrobe’s Clinical Hematology, 554, Lea and Febiger, Philadelphia, PA (1981) 38 Schottenfeld D and Fraaumeni JF Jr, Cancer: Epidemiology and Prevention, Saunders, Philadelphia, PA (1982) 39 Zoloth SR, et al., J Natl Cancer Inst., 76(6),1047-1051 (1986) 40 Mehlman MA (ed.) Advanced in Modern Environmental Toxicology, Volume IV, Princeton Scientific Publishing Company, Princeton, NJ (1983) 41 Mehlman MA, American Journal of Industrial Medicine, 20, 707-711 (1991) 42 Nilsson RI, et al., Occupational & Environmental Medicine, 55, 517-521 (1998) 43 Rinsky A, et al., New England Journal of Medicine, 316, 17, 1044-1050 (April 23, 1987) 44. Wong O, British Journal of Industrial Medicine, 44, 382-395 (1987) 45 Linet MS, et al., Environmental Health Perspectives, 104, Supplement 6, 1353-1364 (December 1996) 46 Yin SN, et al., American Journal of Industrial Medicine, 29, 227-235 (1996) 47 Hayes RB, et al., Journal of the National Cancer Institute, 89, 1065-1071 (1997 48 Wong O, British Journal of Industrial Medicine, 44, 365-381 (1987) 49 Huff JE, et al., Environ Health Perspec., 82, 125-163 (1989) 50 Lewis JG, Odom B and Adams DO, Toxicol Appl Pharmacol, 92, 246-254 (1988) 51 Thomas DJ, Reasor MJ and Wierda D, Toxicol Appl Pharmacol, 97, 440-453 (1989) 52 Kalf GF, Crit Rev Toxicol, 18, 141-159 (1987) 53 Smith MT, et al., Environ Health Perspect, 82, 23-29 (1989) 54 Gist GL and Burg JR, Toxicol and Industrial Health, 13(6), 661-714 (1997) 55 Reid LL, et al., Drug Chem Toxicol, 17(1):1-14 (1994) 56 Winternitz MC and Hirschfelder AD, J Exp Med, 17, 657-664 (1913) 57 USEPA, 5.0 Benzene, 5.1 Chemical and Physical Properties, EPA (1988) 58 International Agency for Research on Cancer (IARC), On the evaluation of carcinogenic risk of chemicals to humans: overall evaluations of carcinogenicity: An updating of IARC monographs 1 to 42, IARC, Lyon, France, Suppl 7 (1987) 59 Infante PF, et al., Lancet, 2:76:1977 60 Vigliani EC and Forni A, J Occup Med, 11, 148 (1969) 61 McMichael AJ, Carcinogenicity of benzene, toluene, and xylene: Epidemiological and Experimental Evidence, IARC, Environmental carcinogens methods of analysis and exposure measurement, Volume 10-Benzene and Alkylated Benzens, Fishbein L and O’Neill IK (eds) IARC Scientific Publications, No. 85 (1988) 62 USEPA, Attachment to IRIS file for benzene, 10/14/98, Response to the Peer-Review Draft Carcinogenic Effects of Benzene: An Update, EPA/600/P-97/001A, June 1997 63 Statement Submitted to the OSHA Benzene Hearing by Otto Wong, ScD., FACE., Environmental Health Associates, Inc, 03/04/86. 64 Decoufle P, Blattner WA and Blair A, Environ Res, 30, 16-25 (1983) 65 Elkins HB, and Pagnotto LD, Archives of Industrial Health, 13, 51-54 (1956) 66 Pagnotto LD, et al., Industrial Hygiene Journal, 417-421 (December 1961) 67 NIOSH. A Recommended Standard for Occupational Exposure to Refined Petroleum Solvents from the U.S. Department of Health, Education and Welfare, Public Health Service, Center for Disease Control, National Institute for Occupational Safety and Health (July 1977) 68 Testimony of R.S. Proctor in regards to Proposed Revised Permanent Standard for Occupational Exposure to Benzene, OSHA Docket No. H-059A, (July 11, 1977) 69 Post-Hearing Comments and Records Designations in OSHA Docket H-059 In re: Proposed Amendment to the Permanent Standard for Occupational Exposure to Benzene (June 12, 1978) 70 Irons R, Toxicology and Environmental Health, 16, 673-678 (1985) 71 Kathleen Lavender, et al., v Bayer Corporation, et al., Civil Action No. 93-C-226K, in the Circuit Court of Marshall County, West Virginia 30 31 32 33 34
20.6 Chromosomal aberrations
1375
20.6 CHROMOSOMAL ABERRATIONS AND SISTER CHROMATOID EXCHANGES Nachman Brautbar University of Southern California, School of Medicine, Department of Medicine, Los Angeles, CA, USA
Several technologies have developed in the last 10 years to look at chromosomal changes and DNA changes caused by environmental exposures, as well as a marker of environmental exposures. The use of chromosomal translocation as a biological marker of exposure in humans have become an important tool in the research, as well as in some instance a marker of exposure. Several methodologies have utilized and include structural chromosomal aberrations, sister chromatoid exchanges (SCEs) and micronuclear changes. These are markers of changes in the cellular genetic materials, and represent damage induced by chemicals. These methodologies are viewed as cytogenetic assays, and by themselves cannot provide a diagnosis, but they complement other methodologies which include gene mutation analysis, and DNA changes. Among the important uses of cytogenetics as a biomarker is the relationship between chromosomal aberrations secondary to chemicals and carcinogeneses. Since the scope of this chapter is not addressing mechanisms of carcinogeneses, the reader is referred to other sources.1 Studies in patients with acute non-lymphocytic leukemia whose bone marrow analyzed for chromosomal changes have shown that 50% of them had changes in chromosomes 5, 7, 8 and 21.2 Mitelman et al.3 have looked at patients with acute non-lymphocytic leukemia and looked at the occupational history. They have studied a group of 56 patients. Twenty-three out of the 56 patients had a history of exposure to chemical solvents, insecticides and petroleum products. They have further shown that in males with acute non-lymphocytic leukemia the frequency of exposure to petroleum products was as high as 36%. In their study, Mitelman et al.3 found a striking differences between the chromosomal findings in non-exposed versus exposed groups. In the non-exposed group only 24.2% of the patients had chromosomal aberrations in their bone marrow cells, while 82.6% of the exposed patients had chromosomal aberrations. The authors concluded, based on these studies that the difference between the exposed and non-exposed group strongly indicates that the karyotypic pattern of the leukemic cells were, in fact, influenced by the exposure. The authors further suggested that the prognosis in those patients with normal chromosomes was significantly better than those with abnormal chromosomes, something which has been suggested by previous investigators.4,5,6 Based on their observations and others,7-11 the authors suggested that certain chromosomal regions possible being specifically vulnerable to the chromosome damaging actions of different chemicals which are carcinogens. Indeed, the changes of chromosomes 5, 7, 8 and 21 in workers exposed to different chemical solvents, among them also petrol and pesticides, supports this hypophysis. Studies in cultured lymphocytes from 73 workers in chemical laboratories and the printing industry were found to have a significantly increased frequency of chromatoid and sister-chromatoid breaks in comparison to 49 control subjects.12 The authors suggested that the observed cytogenetic changes is reasonably assumed to be the result of strong factors in the working environment which induced chromosome breaks and sister-chromatoid exchange. The considerable vari-
1376
Nachman Brautbar
ation in cytogenic changes between subjects suggested varying degrees of exposure to mutagenic agents. Brandt et al.13 studied the effects of exposure to organic solvents and chromosomal aberrations. Ten patients had a history of daily handling of organic solvents for at least one year preceding the diagnoses of non-Hodgkin's lymphoma. All have been exposed to a variety of solvents to include aromatic and aliphatic compounds. Forty-four patients for only a shorter period of time worked with organic solvents, and therefore served as the control group. There was a statistically significant increase in chromosomal changes in the exposed group compared to the non-exposed group. The authors suggested based on their results, that at diagnosis of non-Hodgkin's lymphoma, patients with a history of significant occupational exposure to organic solvents tends to have a larger number of chromosomal aberrations in the lymphoma cells. Furthermore, certain aberrations may be characteristic for the exposed patients. They have concluded that the over representation of certain chromosome aberrations in non-Hodgkin's lymphoma patients occupationally exposed to organic solvents supports the concept that these exposure may be relevant for the subsequent development of non-Hodgkin's lymphoma. Their data are in agreement with the studies published previously, indicating that workers handling organic solvents and other petroleum products have an increased frequency of chromosomal aberrations in lymphocytes.12,14,15,16 Indeed, association between exposure to solvents and other chromosomal changes in the cells have been studied by other investigators, describing chromosomal changes in acute non-lymphocytic leukemia.3,17,18,19 In cultured cells, Koizumi20 examined the effects of benzene on DNA syntheses in chromosomes of cultured human lymphocytes. They have shown an increased incidents of chromatoid gaps and breaks in cultures treated with benzene, compared to those who were untreated with benzene. They also showed changes in DNA metabolism in the form of inhibition of syntheses in those samples which were treated with benzene. This observation was confirmed by other investigators21 who have shown an increased chromosomal aberrations in cultured human leukocytes exposed to benzene. From the in vitro to the in vivo experimental animals, it was shown that treated rats with benzene, were found to have increased chromosomal changes taken from the bone marrow as compared to the non-treated animals.22 It was of interest that the degree of chromosomal changes that were induced by benzene, were similar to that induced by toluene23 (probably secondary to the benzene in toluene). Experimental rabbits treated with benzene also showed a significant amount of bone marrow chromosomal changes persisting up to 60 days after the end dosing with benzene.24 A patient who has developed aplastic anemia after exposure to benzene, was shown to have significant chromatoid fragments.25 A cytogenic study which was carried out later,26 on a patient who developed leukemia after 22 years of continuous exposure to a high concentration of benzene, showed that later in the process there were changes in 47 chromosomes in the bone marrow. Sellyei et al.27 studied patients who developed pancytopenia after having been exposed for 18 months to benzene. Significant chromosomal changes were detected even 7 years after remission from the anemia and the presentation of leukemia. In line with these changes, Forni et al.28 have studied 25 subjects with a history of hematopoietic abnormalities and benzene exposure, and compared these to 25 matched controls. They have shown that 18 years after clinical and hematological symptoms chromosomal aberrations were increased as compared to the control group. In 1965, Tough et al.29 have studied chromosomes of workers exposed to benzene for periods varying from 1 to 18 years. They have also shown a small but significant increase in chromosomal changes com-
20.6 Chromosomal aberrations
1377
pared to a control group. These same investigators looked at workers exposed to benzene levels from 25 to 120 ppm, and found that they had significant chromosomal aberrations as compared to the normal population (which has a general background exposure to benzene levels). The study of Forni et al.14 showed significant chromosomal aberrations in those patients who were exposed mainly to benzene, and not to those who were exposed to toluene only. Hartwich et al.30 looked at 9 healthy refinery workers who were exposed to low levels of benzene, and also found significantly increased chromosomal changes compared to the control group. The National Research Council Advisory Center and Toxicology Study31 concluded that close correlation between occupational exposure to benzene and persistence of chromosomal aberrations can be discussed only when there is an association between benzene induced hematopoietic disease and chromosomal aberrations, however, the absence of chromosomal changes, cannot be a determinant in the temporal relationship between exposure to benzene and hematopoietic diseases.31 While some studies suggested that the chromosomal changes require heavy exposure to benzene, the study by Picciano, 1979,32 looked at chromosomal changes in 52 workers exposed to a mean benzene concentration of less than 10 ppm compared to non-exposed controls. There was a statistical significant increase in chromosomal aberrations in exposure as low as less than 10 ppm. Furthermore, these same investigators33 reported a dose response increase in the aberrations when the exposed workers were divided into smaller groups by the exposure levels (less than 1 ppm, 1-2.5 ppm, and 2.5-10 ppm). Drivers of petrol tankers and crew members of gasoline tankers, ships and petrol station attendants were studied for chromosomal changes.15 The degree of exposure to benzene of the three groups was estimated to be at a mean of 0.4 ppm, and the crew members were estimated to be at 6.56 ppm while engaging in handling of gasoline. The frequency of chromosomal and chromatoid aberrations in the petrol tanker drivers was significantly greater than in those of petrol attendants and the crew members. The effects of long-term benzene exposure from the incidents of chromosomal changes were studied in 16 female workers who were exposed to a maximum of 40 ppm benzene between 1-20 years.34 The cytogenetic study was conducted 6 months after benzene was eliminated from the work environment, and they have found no significant increase in chromosomal changes. Clare et al.35 looked at chromosomal changes in the peripheral lymphocytes of workers after a single, one exposure to benzene. Exposure levels were described as high after a spillage of a large amount of benzene during the loading of a ship. Three months after the incident, chromosomal analysis showed no significant abnormalities. The authors concluded that there was no evidence of lasting chromosomal damage in the peripheral blood lymphocytes of these exposed workers. Golomb et al.18 reviewed the literature and reported the results in regards to exposure to benzene and chromosomal changes. They have stated that they studied exposure data on 74 patients with acute leukemia. They describe that 75% of the exposed patients had an abnormal karyotype, whereas only 43% of the patients characterized as non-exposed had an abnormal karyotype. While it is true that these findings are in agreement with previous studies18 they still could not explain the 43% of the patients who were not exposed, and still had abnormal chromosomal changes. This is a very important observation, since some investigators in the field claimed that the “absence of chromosomal changes” in benzene exposed individuals negates the clinical causative diagnosis of benzene induced hematopoietic disease. Essentially, all of the studies show that benzene can cause chromosomal changes, but does not cause it in all the patients, and the absence of chromosomal changes cannot and does not rule out the ex-
1378
Nachman Brautbar
posure to benzene as a causative factor. In this same paper, Golomb et al.18 looked at the chromosomal changes of patients treated with chemotherapeutic agents for other malignancies. Essentially, they looked at a secondary leukemia developing as a result of alkylating agents. For some reason, they have proposed that losses of part or all of chromosomes numbers 5 and 7 are the specific change resulting from mutageneses, leukemogeneses associated with various chemicals including insecticides, petroleum products and alkylating agents. While this interpretation is compatible with the various animal studies, as well as observations in patients, there is certainly a lack of scientific connection between the benzene exposed chromosomal changes, and the chromosomal changes reported in patients treating with alkylating agents. Smith, in a recent paper36 suggested that oxidation of benzene to multiple metabolites plays a role in producing benzene induced toxicity of DNA damage in bone marrow, and adds further weight to the hypophysis that multiple metabolites are involved in benzene toxicity. They also described DNA changes which have been shown to be cause-point mutation. The investigators measured mutation frequency in 24 workers heavily exposed to benzene, and 23 matched controls. They found that benzene caused a highly significant increase in one variant of mutation, suggesting that benzene produces gene duplicating mutations, but no gene inactivating ones. They suggested that the most-likely consequence of aberrant recombination caused by benzene metabolites is the production of stable chromosomal translocation. Indeed, there are several chromosomal abnormalities shown in leukemic cells. This includes Philadelphia chromosome which results from reciprocal translocation between chromosome 9 and 22, and has been associated with chronic myeloid leukemia, and reciprocal translocation between chromosomes 8 and 21. From these studies it is concluded that benzene is a genotoxic carcinogen, but that other genetic phenomena may mediate benzene induced hematopoietic toxicity. Based on the available data up to date, it is proposed36 that benzene is a carcinogen that does not produce cancer through simple gene mutations, but rather through a separate class of carcinogens (metabolites of benzene) that act by a similar mechanism. In summary, the studies in experimental animals, in vitro, and patients show that benzene and a wide range of organic solvents are associated with changes in chromosomes and DNA adducts. While these changes may be helpful in epidemiological studies, the absence or presence of genetic changes or DNA adducts, cannot be used in a specific case to rule out or establish causation. The biomarkers described in this chapter in the form of genetic biomarkers, can be helpful in identifying individual susceptibility, and in some cases understanding of the mechanism of the disease process. They have a significant number of limitations, and these include measurements, errors and confounding factors. REFERENCES 1 2 3 4 5 6 7
Biomarkers: Medical and Workplace Applications, Medelson ML, Moor LC and Peeters JP (eds), John Henry Press, Washington, D.C., (1998) Rowley JD and Potter D, Blood, 47:705, (1976) Mitelman F, Brandt L and Nilsson PG, Blood, 52(6):1229-1237, (December 1978) First International Workshop on Chromosomes in Leukaemia: Chromosomes in acute non-lymphocytic leukaemia. Br J Haematol, 39:311, (1978) Golomb HM, Vardiman J and Rowley JD, Blood, 48:9, (1976) Nilsson PG, Brandt L and Mitelman R, Leukemia Res, 1:31, (1977) Mitelman F, et al., Science, 176:1340, (1972)
20.7 Hepatotoxicity
8 9 10 11 12 13 14 15 16 17 18 19
20 21 22 23 24 25 26 27 28 29 30 31 32 33
34 35 36
1379
Mitelman F, The Rous sarcoma virus story: Cytogenetics of Tumors induced by RSV in German J (ed): Chromosomes and Cancer, New York, Wiley, page 675, (1974) Levan G and Levan A, Hereditas, 79:161, (1975) Levan G and Mitelman F, Hereditas, 84:1, (1976) Sugiyama T, et al., J Natl Cancer Inst, 60:153 (1978) Funes-Cravioto F, et al., The Lancet, 322-325, (August 13, 1977) Brandt L, et al., European Journal of Haematol, 42:298-302, (1989) Forni A, Pacifico E, and Limonta A, Arch Environ Health, 22:373-378, (1971) Fredga K, Reitalu J, and Berlin M, Chromosome studies in workers exposed to benzene. In: Berg K (ed), Genetic Damage in Man Caused by Environmental Agents, New York, Academic Press, 187-203, (1979) Hogstedt B, et al., Hereditas, 94:179-184, (1981) Mitelman F, et al., Cancer Genet Cytogenet, 4:194-214, (1981) Golomb HM, et al., Blood, 60(2):404-411, (1982) Fourth International Workshop on Chromosomes in Leukemia, 1982: The correlation of karyotype and occupational exposure to potential mutagenic/carcinogenic agents in acute nonlymphocytic leukemia. Cancer Genet Cytogenet, 11:326-331, (1984) Koizumi A, et al., Jap J Ind Health, 12:23-29, (1974) Morimoto K, Jap J Ind Health, 17:106-107, (1975) Philip P and Jensen MK, Acta Pathol Microbiol Scand, Section A, 78:489-490, (1970) Dobrokhotov VB, Gig Sanit, 37:36-39, (1972) Kissling M and Speck B, Helv Med Acta, 36:59-66, (1971) Pollini G and Colombi R, Med Lav, 55:244-255, (1964) Forni A and Moreo L, Eur J Cancer, 3:251-255, (1967) Sellyei M and Kelemen E, Eur J Cancer, 7:83-85, (1971) Forni AM, et al., Arch Environ Health, 23:385-391, (1971) Tough IM and Court Brown WM, Lancet, I, 684, (1965) Hartwich G and Schwanitc G, Dtsch. Med. Wschr, 87:45-49, (1972) National Research Council Advisory Centre on Toxicology, Washington D.C., Health Effects of Benzene: A Review. Prepared for the Environmental Protection Agency. Report No. NAS/ACT/P-829, (June, 1976) Picciano DJ, Environ Res, 19:33-38, (1979) Picciano DJ, Monitoring Industrial Populations by Cytogenetic Procedures. In: Infante PF and Legator MS (eds). Proceedings of the Workshop on Methodology for Assessing Reproductive Hazards in the Workplace, U.S. Government Printing Office, Washington DC, pages 293-306, (1980) Watanabe T, et al., Environ Health, 46:31-41, (1980) Clare MG, et al., Br J Ind Med, 41:249-253, (1984) Smith MT, Environ Health Persp, 104(6):1219-1225, (December 1996)
20.7 HEPATOTOXICITY Nachman Brautbar University of Southern California, School of Medicine Department of Medicine, Los Angeles, CA, USA
20.7.1 INTRODUCTION Solvents which are inhaled or gain access to the blood circulation via skin absorption or at times ingestion largely are metabolized by the liver. The liver has a complex mechanism composed of the cytochrome P450 enzyme, and other enzymes related to conjugation pathways such as glutathione conjugation. This is represented schematically in Figure 20.7.1 It is therefore no surprise that in the occupational setting as well as non-occupational setting, liver damage anywhere from transient to subacute to chronic, and at times terminal liver damage has been described. The circumstances of exposure to hepatotoxic agents are divided to:
1380
Nachman Brautbar
Figure 20.7.1. The phases of detoxification. [Adapted, by permission, from DS Jones, Institute for Functional Medicine, Inc., 1999]
1. Occupational: Either through a routine daily inhalation or skin absorption of solvents which have been shown to be toxic to the liver (accidental exposure). 2. Domestic during either accidental or intentional suicidal exposure, ingestion in foods or as a toxic contaminant of food, exposure to toxic agents such as in the form of glue sniffing. 3. Environmental, most commonly exposure through contaminated water with solvents (drinking water contamination) or through atmospheric pollution such as release to the environment from plants utilizing solvents. Historically the first cases attributing chloroform to liver toxicity were described in 1887, 1889 and 1904.2,3,4 The role of carbon tetrachloride and liver injury has been originally described in 1967 and 1973.5,6 In general, the understanding of hepatotoxicity is extremely complex, and the reader is referred to the outstanding text by Hyman J. Zimmerman.7 A typical example of how metabolism and toxicity of a water takes place is the aromatic chemical such as benzene attached to bromine. The effect on the liver has been originally studied by Mitchell in 19758 who have shown that a change in the rate of the metabolism of this compound is required to create its toxic products. While bromobenzene and carbon tetrachloride share a similar place of metabolism in the liver, the toxicity of bromobenzene and carbon tetrachloride are different, since the bromobenzene toxicity is related to the metabolic capacity of the liver, while that of carbon tetrachloride is not. Several factors contribute to the handling of the solvents by the liver and affect the final toxicity, including species differences. For instance, rats are vulnerable to a wide variety of toxic agents such as carbon tetrachloride and bromobenzene due to the ability of the liver
20.7 Hepatotoxicity
1381
to convert these agents to their respective toxic metabolites.6,9 Among other mechanisms responsible for the species differences are liver blood flow, protein binding, and the points of binding intracellularly. Genetic factors in humans are of extreme importance. Genetic factors are most-likely responsible for the various levels of adverse effects of alcohol in different individuals due to induced activity of detoxification enzymes in the liver in some and lack of those or reduced activity in others. Another important factor is age. The effect of age on the susceptibility has been shown in experimental animals. For instance, the neonatal rats are less susceptible to carbon tetrachloride and bromobenzene toxicity as compared to adult animals.9 In humans, liver necrosis after the administration of Halothane was rare in children, but more common in more elderly patients. Factors such as sex and endocrine status have also influence and different toxic effects of solvents in this case. Nutritional status is a major factor in the effects of solvents on the liver. For instance, protein malnutrition leads to reduced activity of cytochrome P450. Increasing the percentage of fat in the liver has been shown to increase the susceptibility of toxicity to such agents as carbon tetrachloride. Several studies have looked into the histopathological injury of some solvents and solvent-like agents and the liver, and are shown in Table 20.7.1. Table 20.7.1. Partial list of agents that produce hepatic necrosis in experimental animals. [Adapted, by permission, from HJ Zimmerman, Hepatotoxicity, 1978] Site of necrosis Centrizonal
Midzonal
Peripheral Zone
Massive
Steatosis
Bromobenzene
+
(+)
Bromotrichloromethane
+
+
CCl4
+
+
Chlorobenzenes
+
Chloroform
+
+
Dichloropropane
+
+
Dinitrobenzene
+
(+)
+
Dinitrotoluene
+
(+)
+
Ethylene dichloride
+
Methylene chloride
(+)
Naphthalene
+
(+)?
+ (+)
(+) +
The specific mechanism of hepatotoxicity of many solvents are unknown, however, the knowledge have been gathered from experimental studies now available for the reader for review.10,11,12,13 Due to the lack of specific information for many solvents, I have decided to discuss in this chapter some of the most typical ones which have been used in the past heavily, or are used currently.
1382
Nachman Brautbar
Table 20.7.2. Some occupational solvents that produce acute and chronic liver disease Carbon tetrachloride Chloroform Trichloroethylene Tetrachloroethylene Dinitrobenzene
Halothane Trichlorodiphenyl Trichloroethene Trinitrotoluene
Ethyl alcohol A mixture of solvents such as toluene and xylene Dichloromethane
Tables 20.7.3 and 20.7.4 describe two major phases in the liver metabolism and detoxification of drugs, foreign agents and solvents. Carbon tetrachloride was found to cause liver injury in man and in experimental animals.14 Carbon tetrachloride is a known and potent liver toxicant, and therefore has been studied extensively in experimental animals. Recknagel6 and Reynolds15 have shown that single doses will lead to areas of necrosis in the liver within a few minutes. This has been shown to be associated with changes in liver enzymes which are known to indicate liver damages.10,11 Prolonged exposure to carbon tetrachloride has been shown to lead to liver cirrhosis and to liver cancers. In order to become toxic the carbon tetrachloride has to undergo metabolic changes in the liver.6,16 The lesions described initially in animals have been shown in humans poisoned with carbon tetrachloride.14,17 It has also been shown that alcohol enhances the susceptibility to carbon chloride toxicity.18 Several factors play a role in the susceptibility to toxicity by carbon tetrachloride, among them are sex, age, diet, underlying preexisting liver dysfunction and alcoholism. Over the years in both clinical and experimental studies and observations, it has been shown that carbon tetrachloride induced liver damage is divided to fatty metamorphosis, and independently liver necrosis. Fat starts to accumulate in experimental animals as early as one hour after administration of a high dose of carbon tetrachloride. Liver necrosis occurs as early as 6 to 12 hours, and a maximum of 24 to 36 hours. The concept of steatosis (fat accumulation in the liver) is a common one for looking at the effect of solvents (those which are known to be toxic to the liver). The fat accumulation is the result of abnormal transport of lipids and as a result, accumulation of lipids in the liver. Therefore clinically industrial exposure to hepatotoxic solvents is associated with liver steatosis, among others. Necrosis, which is the second most common effect of liver damage of solvents toxic to the liver, is the result of destruction of the cell architecture as well as biochemical pathways. It has been shown in many experimental studies that the toxicity of carbon tetrachloride (and some of the other solvents which are toxic to the liver and other organs, such as benzene and the hematopoietic system) requires several reactions, in order to produce the toxic metabolites which are causing the damage to the liver. Most studies point to the responsibility of cytochrome P450 system.6 The metabolite responsible for the liver damaging effect of carbon tetrachloride is a C Chloride III which is formed from carbon tetrachloride.6,19 There is however also information that non-metabolized carbon tetrachloride contributes to the injury, especially that of the cell membrane,20,21 something which is logical since solvents are a mechanistic injury to various tissues is through the effects on the cell membrane which is either dissolved or damaged by the solvent. A consolidation of the data available and views on the pathogenesis of carbon tetrachloride liver damage has been eloquently described by Zimmerman.7
20.7 Hepatotoxicity
1383
Table 20.7.3. Phase I reactions. [Adapted, by permission, from HJ Zimmerman, Hepatotoxicity, 1978]
1384
Nachman Brautbar
Table 20.7.4. Phase II (conjugation) reactions. [Adapted by permission, from HJ Zimmerman, Hepatotoxicity, 1978] Type of conjugation
Glucuronidation
Endogenous substance
Transferring enzyme and location
UDP-glucuronic UDPG-transferase acid (microsomes)
Type of xenobiotics & metabolites conjugated Phenols; alcohols; carboxylic acids; primary amines; hydroxylamines; sulfonamides, etc.
Dihydrodiol formation Water
(Epoxide) hydrase (cytosol)
Epoxides and arene oxides
GSH conjugation
GSH
GSH-S-transferase (cytosol)
Epoxides; arene oxides; halides; nitro groups; hydroxylamines, etc.
Glycine conjugation
Glycine
Acyl CoA-glycine transferase (mitochondria)
CoA derivatives of carboxylic acids
Sulfate conjugation
PAP-sulfate
Sulfotransferase (cytosol)
Phenols; alcohols; aromatic amines
Methylation
S-adenosylmethionine
Transmethylase (cytosol)
Catechols; phenols; amines; histamine
The carcinogenic effect of carbon tetrachloride will not be discussed in this chapter, and the reader is referred to other texts.22 20.7.2 INDIVIDUAL VARIABILITY AND HEPATOTOXICITY OF SOLVENTS The issue of individual variability based on various factors as described above is important, especially in medical monitoring and risk assessment in occupationally exposed patients. The fact that workers in industrial environments are not the same and are subject for differences such as body build, underlying kidney function differences (genetic or acquired), exposure to other solvents or other liver toxicants may effect the results. A recent study23 evaluated a population response to solvent exposure. These investigators have shown that body fat is the most important body compartment for fat soluble solvents. Body fluids and physical work load effect the blood flow, alveoli ventilation and therefore will effect the amount of solvent inhaled as well as absorbed through the blood and delivered to the liver. They have developed the physiological model which takes into account variability in the form of exposure, physical overload, body build, liver function and renal function. Other factors which have been taken into account are solubility in blood and tissue. Investigators suggested that such a model should be useful in improving our understanding of the complex and multifactorial system and to generate a hypothesis, and to improve our assessment of occupational exposure. This has significance from a clinical toxicology point of view. A patient who has an increased body fat will be at a higher risk of solvent toxicity. If this same patient also has a habit of heavy alcohol consumption the risk for solvent liver toxicity is significantly increased. Epidemiological studies are commonly not designed to evaluate the individual hepatotoxicity of solvents and therefore the issue of cause and effect must be viewed taking into account individual variability, other risk factors, and medical common sense, following the well established criteria by practitioners of medicine. Chloroform is another haloalkane which has been typically used as an example to understand and study the toxic effects on the liver. Studies in experimental animals in 186624 have shown that the chloroform causes liver toxicity. In 1923 Meyer et al.14 have shown that
20.7 Hepatotoxicity
1385
toxicity of chloroform to the liver in humans. Chloroform was used years ago as an anesthetic, and has been used successfully, however, due to its’ toxicity the use of it has been abandoned. Acute exposure and toxicity has been associated with liver necrosis, liver steatosis, and chronic exposure has been associated with liver cirrhosis. The mechanism of injury most-likely is the result of metabolic changes of chloroform by the liver. Different effects and level of toxicity between carbon tetrachloride and chloroform is most-likely the result from the solubility in lipid and water, and the mechanisms by which these two agents are metabolized and then cause liver toxicity. Table 20.7.5 summarizes liver damage described in the literature as a result of halogenated aliphatic hydrocarbons. Table 20.7.5. Lesions produced by halogenated aliphatic hydrocarbons. [Adapted, by permission, from HJ Zimmerman, Hepatotoxicity, 1978] Steatosis and centrizonal necrosis CCl4 CI4 CCl3Br CHCl3 CHI3 CHBr3 CHCl2CHCl2 CH2ClCH2Cl CH2BrCH2Br CH3CCl3 CHCl2CCl3 CHCl=CCl2 CH3CHClCH3 CH3CHClCH2Cl
Steatosis only CH2ClBr CH2Cl2 CHCl=CHCl (cis) CCl2=CCl2 CH3CH2CHClCH3
Slight steatosis or no injury CH3Cl CH3Br CH3I CCl2F2 CHCl=CHCl (trans) CH3CH2Cl CH3CH2I CH3CH2Br CH3CH2CH2CH2Cl
20.7.3 NON-HALOGENATED SOLVENTS While the halogenated hydrocarbons discussed here include carbon tetrachloride, chloroform, 1,1,1-trichloroethene, trichloroethylene are significantly hepatotoxic, the literature on the toxicity of the non-halogenated hydrocarbons is a combination of positive and negative studies. Several studies looking into the hepatotoxicity of both aliphatic solvents such as kerosene, hexane and aromatics such as xylene, toluene and styrene have reported mixed results. Xylene is an aromatic hydrocarbon which is used heavily in the industry, as well as medical technology as a solvent. Xylene commonly has been reported with impurities in varying amounts which include ethyl benzene, prime ethyl benzene, phenol, benzene and toluene.25 To evaluate the effects of xylene in experimental animals, Toftgard et al.26 studied rats who were exposed for three days by inhalation to xylene and to a mix of xylene isomers. Hepatic cytochrome P450 concentration increased as well as C reductase activity, and NADPH cytochrome C reductase activity. Furthermore, xylene and its isomers were able to modify the metabolism of other potentially toxic substances. In addition to these biochemical changes, the investigators found that xylene increased liver body weight, most-likely secondary to proliferation of the endoplastic reticulum. These studies show that at these levels xylene induced cytochrome P450 activity and NADPH cytochrome C reductase activity, but was not associated with significant pathological abnormalities. On the other hand, these same investiga-
1386
Nachman Brautbar
tors, concluded that the capacity of xylene and xylene isomers to induce hepatic cytochrome P450 suggests the possibility of synergistic toxic effects from simultaneous exposure to xylene and substances metabolically activated by these cytochromes P450. Therefore, exposure to a mixture of solvents which also include xylene will increase the toxicity and xylene is a synergistic liver toxicant. For instance, the formation of 2-hexanoyl a metabolic precursor of 2,5-hexanedione, which is the main metabolite found in urine of workers exposed to n-hexane, is increased following xylene treatment.27 20.7.4 SOLVENT MIXTURES Various solvent mixtures have been reported as hepatotoxic, Fishbien et al.28,29 reported abnormal liver functions in chemical workers exposed to a mixture of solvents. It was suggested that bile acids as indicators of hepatic function will be utilized as markers of injury. Franco et al.30 examined a group of workers exposed to organic solvents, and used the criteria addressing exposure to solvent mixtures for over two years, daily ethanol consumption less than 50 grams, no history of hepatic disease, no drug intake in the previous three months. Workers were exposed to between 6 and 9 solvents, mostly toluene, xylene, acetone, methyl acetate, and butanol ethyl acetate. The mean levels of liver enzyme activities in the exposed and the control group were similar. The mean serum bile acid contents was statistically and significantly increased in the exposed group compared to the controlled group. The authors concluded that the observation of higher serum bile acid levels in the group of workers currently exposed to organic solvents might be explained by a change in hepatocyte function, and that the commonly followed parameters of liver enzymes may be insensitive to these preliminary initial changes in liver function. Conventional liver function tests seem to be rather insensitive to early detect liver damage from solvents exposure. Early detection is crucial, but is probably missed since the standard liver functions tests are not sensitive enough to detect early liver damage from solvents. What that means is that by the time the patient is seen by the doctor with liver fibrosis or necrosis and solvent toxicity, it is already in advanced stages. Joung-Dar-Chen et al.31 evaluated the effects of solvents exposure on liver functions, specifically looking at gamma glutamyl transferase activity. They have studied the effects of xylene and toluene. The median air concentration was evaluated in the exposed workers who used mixed solvents in the process of spray booth car painting. These investigators showed that gamma glutamyl transferase activities increased independently with both an increased consumption of alcohol and exposure to a mixture of solvents. They have concluded that an increase in GGT activity may be a form of enzyme induction rather than a marker for cellular damage. Kurpper et al.32 examined the effect of mixed organic solvents on liver enzymes activity in car painters, and found that at the exposure level of that study (at that time was 1/2 the level recommended by the regulatory agencies) the liver enzyme activities of car painters were not effected by exposure to mixed solvents. The continuation of the previous study by Franco et al.33 examined serum bile acid concentrations as a marker of liver functions in a group of workers exposed to organic solvents. They have shown a significantly increased concentration of serum bile acids with normal liver enzymes, and concluded that this indicated a very sensitive and early marker of liver function abnormality in patients exposed to mixed solvents, and might be explained as an early sign of liver dysfunction. While it was not possible to state which solvent caused what type of abnormality, the authors concluded that conventional liver function tests seem to be rather insensitive for early liver disease detection, and normal measurement does not rule out the existence of subclinical disease, and therefore, an elevation of serum bile acid indicates
20.7 Hepatotoxicity
1387
early potential hepatotoxicity from mixed organic solvents. From a clinical point of view this study suggests that exposure to mixture of solvents in an industrial setting may cause liver damage which is subclinical and initially undetected, unless liver enzymes and bile acids are measured. It seems likely that this initial stage of liver damage in patients exposed to solvent mixtures is commonly missed. 20.7.5 TRICHLOROETHYLENE The liver is a target organ toxicity for trichloroethylene in experimental animals. Data in humans are limited.34 Case reports describe trichloroethylene induced hepatitis and liver necrosis.35 Guzelian et al. described both hepatic necrosis and fatty metamorphasis.36 As early as 1962, the hepatotoxicity of trichloroethylene has been studied in humans. Trichloroethylene has been found to cause liver damage after both acute and chronic exposures.37,38,39,40,41 Several studies reported a history of pathological changes including individual or focal necrosis after treatment of experimental animals with trichloroethylene.42 In addition to these histopathological changes Berman also found a dose response relationship to the histopathological changes. Since trichloroethylene is commonly present as a contaminant in ground water (from degreasing, paint thinning and plastic metal processes) Barton has evaluated risks assessment of trichloroethylene and liver toxicity,43 and showed that exposure was associated with continuous response in the form of liver toxicity. There was a connection between increased liver weight over body weight and liver toxicity, this effect also appears to be a sensitive indicator of liver toxicity. The authors concluded that trichloroethylene is toxic to the liver, based on their analysis of their findings, and that the use of liver enzymes by themselves may miss the early signs of toxicity to the liver by trichloroethylene. From these studies on trichloroethylene, and the studies on solvent mixtures described above, it seems reasonable to conclude that the early subclinical stages of solvent hepatotoxicity are commonly missed. When patients present to the doctor they will already be in more advanced stages such as liver steatosis and/or chronic hepatitis, and at times liver fibrosis. The toxicity of trichloroethylene is dependent upon metabolism and induction of cytochrome P450. Trichloroethylene is metabolized through chloral hydrate to compounds including trichloroacetic acid and dichloroacetic acid which alter intercellular communication, induce peroxisome proliferation and may promote tumor production.44 Significant variability in trichloroethylene metabolism in 23 human haptic microsomal samples was reported by Lipscomb et al.44 It was also demonstrated that the trichloroethylene metabolism is dependent on enzymatic activities of the cytochrome system, and they conclude that their data indicates that humans are not uniform in their capacity for CPY dependent metabolism of trichloroethylene and increased activity may increase susceptibility to trichloroethylene induced toxicity in humans. These observations are compatible with the variability reaction which is depending on nutritional factors, enzyme induction factors, hormonal factors and interaction with other environmental chemicals, prescription medications and general health conditions, and explains the variable reports as far as trichloroethylene and level of liver toxicity in the various individuals studied. In a predisposed individual (for example, a patient who is on medications or alcohol) it is highly likely that exposure to trichloroethylene will be a substantial factor in the genesis of a wide variety of liver diseases.
1388
Nachman Brautbar
20.7.6 TETRACHLOROETHYLENE Tetrachloroethylene is synthetic chemical used for dry cleaning fabrics, and has also been named as perchloroethylene and tetrachloroethene. The liver is the target organ in humans, mainly in reports of accidental exposure to a high concentration. Meckler et.al.45 has shown liver damage in a woman exposed occupationally to tetrachloroethylene fumes documented by a liver biopsy. Other investigators also have shown elevation of liver enzymes, jaundice and enlarged liver.46,47 Experimental animal studies also have shown liver damage by inhalation of tetrachloroethylene.48,49,50,51 Liver necrosis occurred in experimental mice exposed to 100 and 200 ppm of tetrachloroethylene for 103 weeks.52 Experimental animals exposed orally to tetrachloroethylene have been shown to develop liver changes similar to those produced by inhalation studies, and mice are more sensitive than rats to tetrachloroethylene induced liver toxicity. Humans exposed by oral routes to tetrachloroethylene except for heavy doses commonly have not shown significant changes other than obstructive jaundice and enlarged liver reported in an infant exposed to tetrachloroethylene via breast milk.53 Issues of carcinogenicity will not be addressed in this chapter, and the interested reader is referred to the toxicological profile for tetrachloroethylene.54 It is highly likely that tetrachlorethylene is a hepatotoxic agent in high doses, and probably in low doses in susceptible individuals with either other environmental exposures, prescription medications, alcoholism, nutritional and/or genetic factors, and preexisting disease of the liver. 20.7.7 TOLUENE Industrial use of toluene is wide, commonly in paint, paint thinners, fingernail polish, lacquers, adhesives, rubbers and in the printing letter industry. Toluene is extensively metabolized by the liver; however, the liver does not appear to be a primary target for toluene toxicity. A study of printing factory workers who were exposed to toluene at a concentration of less than 200 ppm, showed minimal changes of liver enzymes.55 The cohorts included 289 men, of which 8 showed elevated liver enzymes, and 6 of them had enlarged livers. Seven of those patients had liver biopsies which showed some centrally lobular and periportal fat accumulation, and Kupffer cell hyperplasia. The study by Svensson et al.56 has looked at 47 rotogravure workers occupationally exposed to toluene at a concentration of 80 ppm for 3-39 years, and showed a significant elevation of liver enzymes, finding of chemical hepatitis. Seiji et al.57 has examined a group of 157 female shoemakers who were exposed to toluene at levels of 7-324 ppm from 2-14 months, and showed no significant elevation of commonly measured liver functions. Another study that looked at 47 Swedish paint industry workers who were exposed for more than a 10 year period of time to organic solvents which included xylene, toluene, isobutanol, n-butanol, mineral spirits, methyl acetates, dichloromethane, methyl ethyl ketone and isopropanol did not show any changes in liver enzymes.58 However, this study cannot be relied upon as a specific study, since the cohort size was small, and there were multiple exposures to multiple solvents, and therefore, the study had only limited power to detect the effects of toluene on the liver of exposed workers. Experimental animals exposed to toluene at concentrations of 533 to 800 ppm for 7 days showed increased liver weights, but no significant morphological changes by microscopy. Electron microscopical examination revealed ultrastructural changes which were compatible with changes in the cytochrome P-450 concentrations. Others have shown no effect on liver size or liver functions.59,60,61,62 Overall the data seems to suggest that toluene may cause liver damage in certain industries, and especially in synergism with other sol-
20.7 Hepatotoxicity
1389
vents. It is highly likely that, in predisposed individuals, toluene can cause liver damage from chemical hepatitis to necrosis and fibrosis. 20.7.8 DICHLOROMETHANE Dichloromethane also called dichloromethane, is a colorless liquid that has a mild sweet odor. It is used widely in the industry as a solvent and a paint stripper. It is commonly found in spray painting operations, automotive degreasing, in cleaners and in household products. Stewart et al.63 showed no changes in liver enzymes in patients exposed for a period of 6 weeks to levels of dichloromethane via inhalation from 50-500 ppm. On the other hand, Ott et al.64 has shown an elevation in bilirubin in workers exposed to dichloromethane up to 475 ppm. In experimental animals dichloromethane exposure has been associated with fatty changes of the liver and elevated liver enzymes. Norpoth et al.65 have shown hepatic microsomal enzyme elevation at 500 ppm of dichloromethane exposure for 10 days, and others have shown significant fatty changes of the liver upon exposure of mice and rats for 100 days to 75-100 ppm of trichloroethylene.66,67,68 When exposure to dichloromethane continues for 2 years there was increased evidence of pathological changes and fatty liver changes.69,70,71 The overall weight of the data supports a hepatotoxic effect of dichloromethane on the liver. 20.7.9 STODDARD SOLVENT Stoddard solvent is a widely used organic solvent synthetically made, and comes from the refining of crude oil. It is a petroleum mixture made from distilled alkanes, cycloalkanes (naphthenes), and aromatic compounds. In addition, it goes by other names such as Varsol 1, Texsolve S and others. It is commonly used as a paint thinner, as solvents in some types of photocopier toners, printing ink, adhesives, dry cleaning and as a general cleaner and degreaser. Twelve men exposed to 610 mg per cubic meter of vaporized Stoddard solvents for a period of 6 hours revealed no changes in serum glucose, triglycerides, cholesterol or urate.72 Dossing et al.73 described painters who were exposed to non-specified levels of Stoddard solvents and other chemicals for chronic periods, and elevated levels of serum alanine aminotransferase, but other functions were normal and normal liver biopsies. Flodin et al.74 has studied a group of patients exposed to a variety of solvents, including Stoddard solvents and showed normal liver function tests, but an elevated gamma glutamyl transferase. Hane et al.75 has shown that a group of painters exposed to Stoddard solvents had no significant abnormality of liver enzymes. Studies in experimental animals showed minimal fatty changes of the liver, Jenkins et al.76 as did the studies by Carpenter and Phillips.77,78,79 The data from experimental animals and humans suggests a potential hepatotoxic effect of Stoddard solvents, but additional studies and a case by case evaluation is required. 20.7.10 1,1,1-TRICHLOROETHANE 1,1,1,-Trichloroethane is a colorless solvent which is manmade. It is produced by industry and is used in commercial products. It is used as a solvent, and is heavily used in glue and paint, as well as a degreaser and metal parts manufacturing. It is also used in some household products such as spot cleaners, glues and aerosol sprays. It is commonly found in soil and water as a contaminant. Brief single exposures to very high levels of 1,1,1-trichloroethane and a moderate high concentration have been shown to cause elevation of urobilinogen in patients.80 This type of finding indicates reduced bile excretion and some intrinsic liver damage. Stewart et al.81 showed in patients accidentally exposed to a high concentration of 1,1,1-trichloroethane increased levels of urinary urobilinogen for ap-
1390
Nachman Brautbar
proximately 4 days following the exposure. While looking at enzyme levels in the blood after exposure to 1,1,1-trichloroethane, there was no evidence of elevated serum enzyme levels.80,82,83,84 Case studies of people exposed to a high concentration of 1,1,1-trichloroethane did not show elevated liver enzymes.81,85 Histopathological examination of the liver of patients who died following inhalation of a high concentration of 1,1,1-trichloroethane showed minimal changes, mainly those of mild fatty changes of the liver.86,87 Kramer et al.88 studied humans at low levels of exposure and found minor changes in liver enzymes. Experimental animal studies showed mild histopathological changes and effects on liver enzymes.89,90 Truffert et al.91 showed that intermittent duration and intermittent exposure to a low concentration of 1,1,1-trichloroethane produced a 67% increase in the synthesis of DNA of the livers of exposed rats, and concluded that the DNA synthesis measurements may be a more sensitive indicator of liver damage than just measurements of liver enzymes. McNutt et al.92 showed histological damage following exposure to 1,1,1-trichloroethane with hepatocyte necrosis. The most commonly reported effects of 1,1,1-trichloroethane on the liver in experimental animals is increased fat accumulation.93,94,95 The function of duration of exposure played an important role in experimental rats, and was seen in those who were exposed for 7 hours, but was not seen in those exposed for 2 hours in high levels. Savolainen et al.96 showed that exposure to a moderate concentration in experimental animals caused decreased microsomal cytochrome P450 enzyme activity. Overall, the animal studies and human studies suggest an effect of 1,1,1-trichloroethane on the liver, but the severity appears to be related to the dose and duration of the exposure. 20.7.11 SUMMARY In summary, from the available data, it is clear that exposure to solvents and hepatotoxicity must be evaluated in context of the individual variability, exposure to mixture of solvents, and synergistic toxicity. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13
14 15 16 17 18
Jones DS, ed, Detoxification: A Clinical Monograph, Institute for Functional Medicine, Inc., 1999 Ungar, Viertelsjarisch f. gericht. Med., 47:98 (1887) Ostertag R, Virchows Arch., 118:250 (1889) Stiles HJ and McDonald S, Scott Med Surg J, 15:97 (1904) Recknagal RO, Pharmacol Rev, 19:145 (1967) Recknagel RO and Glinde EA, CRC Crit Rev Toxicol, 2:263 (1973) Zimmerman HJ, Hepatotoxicity: The Adverse Effects of Drugs and Other Chemicals on the Liver, Appleton-Century-Crofts, New York, 1978 Mitchell JR, et al. In Concepts in Biochemical Pharmacology, Part 3, JR Gillette, JR Mitchell and PS Randall (eds), Springer-Verlag, Berlin, 383-419 (1975) Mitchell JR, et al., Drug Met Disp, 1:418 (1973) Rouiller CH, In The Liver, Volume II, CH Rouiller (ed), Academic Press, New York, 335-476 (1964) Von Oettingen WF, The Halogenated Hydrocarbons of Industrial and Toxicological Importance, Elsevier, Amsterdam (1964) Browning E, Toxicology and Metabolism of Industrial Solvents, Elsevier, Amerstdam (1965) Von Oettingen WF, The Halogenated Aliphatic, Olephinic Cyclic, Aromatic and Aliphatic-Aromatic Hydrocarbons Including the Halogenated Insecticides. Their Toxicity and Potential Dangers. U.S. Dept. HEW, U.S. Govt Printing Office, Washington, D.C. (1955) Meyer J and Pessoa SB, Am J Trop Med, 3:177 (1923) Reynolds ES, Biochem Pharmacol, 21:2255 (1972) Slater TF, Nature (Lond), 209:36 (1966) Jennings RB, Arch Pathol, 55:269 (1955) Klatskin G, Toxic and Drug Induced Hepatitis. In Diseases of the liver, 4th Ed, L Schiff (ed), JB Lippincott, Philadelphia, 604-710 (1975)
20.7 Hepatotoxicity
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
53 54 55 56 57 58 59 60 61 62
63
1391
Slater TF, Free Radical Mechanisms in Tissue Injury, JW Arrowsmith, Ltd, Bristol, 118-163 (1972) Le Page RN and Dorling PR, Aust J Exp Biol, 49:345 (1971) Rufeger U and Frimmer M, Arch Pharmacol, 293:187 (1976) Agency for Toxic Substances and Disease Registry (ATSDR), Toxicological Profile for Carbon Tetrachloride, Atlanta, Georgia (1997) Droz PO, Wu MM, Cumberland WG and Berode M, Br J Ind Med, 46(7):447-460 (July 1989) Nothnagel, Berlin Klin Wchnschr, 3:31 (1866) International Agency for Research on Cancer (IARC) Monographs on the Evaluation of Carcinogenic Risks to Humans, Volumed 47:125-156; IARC, Lyon, France Volumes 43-48 (1987-1990) Toftgard R and Nilsen OG, Toxicology, 23(2-3):197-212 (1982) Frommer U, Ullrich V and Orrenius S, FEBS Lett, 41(1):14-16 (Apr 15, 1974) Sotaniemi EA, et al., Acta Med Scand, 212:207-215 (1982) Fisbien A, et al., Lancet, 1:129 (1983) Franco G, et al., Br J Ind Med, 46:141-142 (1989) Chen JD, et al., Brit J Ind Med, 48:676-701 (1991) Kurppa K and Husman K, Scand J Work Environ Health, 8:137-140 (1982) Franco G, et al., Int Arch Occup Environ Health, 58:157-164 (1986) Davidson IW and Beliles RP, Drug Metabolism Review, 23:493-599 (1991) Bond GR, Clinical Toxicology, 34(4):461-466 (1996) Guzelian PS, Disorders of the Liver, In Principles and Practice of Environmental Medicine, Aylce Bezman Tacher (eds), Plenum Medical Book Company, New York, 319-333, (1992) Baerg RD and Kimberg DV, Annals of Internal Medicine, 73:713-720 (1970) Clearfield HS, Digestive Digest, 15:851-856 (1970) Fielder RJ, Loweing RK and Shillaker RO, Toxic Rev, 6:1-70 (1982) Litt I and Cohen M, New England Journal of Medicine, 281:543-544 (1969) Sax N, Dangerous Properties of Industrial Chemicals, 4th Edition, Van Nostrand Reinhold Co, Toronto, Canada, 1186 (1975) Berman E, et al., J Tox Environ Health, 45(2):127-143 (June 1995) Barton HA, et al., Regulatory Toxicology & Pharmacology, 24:269-285 (1986) Lipscomb JC, et al., Toxicology & Applied Pharmacology, 142:311-318-1997 Meckler LC and Phelps PK, J Am Med Assoc, 197:662-663 (1966) Coler HR and Rossmiller HR, AMA Arch Ind Hyg Occup Med, 8:227-233 (1953) Hake CL and Stewart RD, Environ Health Perspect, 21:231-238 (1977) Kyline B, et al., Acta Pharmacol Toxicol, 20:16-26 (1963) Kyline B, Sumegi I and Yllner S, Acta Pharmacol Toxicol, 22:379-385 (1965) Carpenter CP, J Ind Hyg Toxicol, 19:323-336 (1937) Rowe VK, et al., AMA Arch Ind Hyg Occup Med, 5:566-579 (1952) National Toxicology Program (NTP) - Technical Report Series No. 311, Toxicology and Carcinogenesis Studies of Tetrachloroethylene (Perchloroethylene) (CAS No. 127-18-4) in F344/N rats and B6C3F1 mice (inhalation studies), Research Triangle Park, NC, U.S. Dept of Health & Human Services, Public Health Service, National Institute of Health, NIH Publication No. 86-2567 (1986) Bagnell PC and Ellenberger HA, Can Med Assoc J, 117:1047-1048 (1977) Agency for Toxic Substances and Disease Registry (ATSDR), Toxicological Profile for Tetrachloroethylene, Atlanta, Georgia (1993) Guzelian P, Mills S and Fallon HJ, J Occup Med, 30:791-796 (1988) Svensson BG, et al., Br J Ind Med, 49:402-408 (1992) Seiji K, et al., Ind Health, 25:163-168 (1987) Lundberg I and Hakansson M, Br J Ind Med, 42:596-600 (1985) Bruckner JV and Peterson RG, Toxicol Appl Pharmacol, 61:27-38 (1981) Bruckner JV and Peterson RG, Toxicol Appl Pharmacol, 61:302-312 (1981) Kjellstrand P, et al., Acta Pharmacol Toxicol, 57:242-249 (1985) National Toxicology Program (NTP) - Technical Report Series, Toxicology and Carcinogenesis Studies of Toluene (CAS No. 108-88-3) in F344/N rats and 86C3F mice (inhalation studies), Research Triangle Park, NC, U.S. Environmental Protection Agency, U.S. Dept of Health & Human Services, No. 371. PB90-256371 (1990) Stewart RD, et al., Methylene Chloride: Development of Biologic Standard for the Industrial Worker by Breath Analysis. Report of the National Institute of Occupational Safety and Health, Cincinnatti, OH, by the Medical College of Wisconsin, Milwaukee, Wisconsin, NTIS No. PB83-245860 (1974)
1392
64 65 66
67 68
69 70 71
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
Nachman Brautbar
Ott MG, et al., Scand J Work Environ Health, 9(Supple 1):17-25 (1983) Norpoth K, Witting U and Springorum M, et al., Int Arch Arbeitsmed, 33:315-321 (1974) Haun CC, et al., Continuous Animal Exposure to Low Levels of Dichloromethane. In: Proceedings of the Third Annual Conference on Environmental Toxicology. Wright Patterson Air Force Base, OH: Aerospace Medical Research Laboratory, 199-208, AMRL-TR-72-130, (1982) Kjellstrand P, et al., Acta Pharmacol Toxiocl (Copenh), 59:73-79 (1986) Weinstein RS and Diamond SS, Hepatotoxicity of dichloromethane (methylene chloride) with continuous inhalation exposure at a low dose level. In: Proceedings of the Third Annual Conference on Environmental Toxicology. Wright Patterson Air Force Base, OH: Aerospace Medical Research Laboratory, 209-220, AMRL-TR-72-130, (1972) Burek JD, et al., Fund Appl Toxicol, 4:30-47 (1984) Nitschke KD, et al., Fundam Appl Toxicol, 11:48-59 (1988) National Toxicology Program (NTP) - Technical Report Series No. 306, Toxicology and Carcinogenesis Studies of Dichloromethane (Methylene Chloride) (CAS No. 75-09-2) in F344/N rats and B6C3F1 mice (inhalation studies), Research Triangle Park, NC, U.S. Dept of Health & Human Services, Public Health Services, Centers for Disease Control, National Institute of Health, (1986) Pederson LM and Cohr KH, Acta Pharmacol Toxicol, 55:317-324 (1984) Dossing M, Arlien-Soborg P and Petersen LM, Eur J Clin Invest, 13:151-158 (1983) Flodin U, Edling C and Axelson O, Am J Ind Med, 5:287-295 (1984) Hane M, et al., Scand J Work Environ Health, 3:91-99 (1977) Jenkins LJ, et al., Toxicol Appl Pharmacol, 18:53-59 (1971) Carpenter CP, et al., Toxicol Appl Pharmacol, 32:246-262 (1975) Carpenter CP, et al., Toxicol Appl Pharmacol, 32:282-297 (1975) Phillips RD and Egan GF, Fundam Appl Toxicol, 4:808-818 (1984) Stewart RD, et al., Am Ind Hygn Assoc J, 22: 252-262 (1961) Stewart RD, J Am Med Assoc, 215:1789-1792 (1971) Stewart RD, et al., Arch Environ Health, 19:467-472 (1969) Dornette WHL and Jones JP, Anesthesia and Analgesia, 39:249-252 (1960) Torkelson TR, et al., Am Ind Hygn Assoc J, 19:353-362 (1958) Wright MF and Strobl DJ, J Am Osteopath Assoc, 84:285-288 (1984) Caplan YH, Backer RC and Whitaker JQ, Clin Toxicol, 9:69-74 (1976) Hall FB and Hine CH, J Forensic Sci, 11:404-413 (1966) Kramer CG, et al., Arch Environ Health, 33:331-342 (1978) Carlson GP, Lif Sci (United States), 13:67-73 (1973) Gehring PJ, Toxicol Appl Pharmacol, 13:287-298 (1968) Truffert L, et al., Arch Mal Prof Med Trav Secur Soc, 38:261-263 (1977) McNutt NS, et al., Lab Invest, 32:642-654 (1975) Takahara K, Okayama Igakkai Zasshi, 98:1099-1110 (Japanese) (1986) Adams EM, et al., Am Med Assoc Arch Ind Hyg Occup Med, 1:225-236 (1950) Herd PA, Lipsky M and Martin HF, Arch Environ Health, 28:227-3 (1974) Savolainen H, et al., Arch Toxicol, 38:229-237 (1977)
20.8 Solvents and the liver
1393
20.8 SOLVENTS AND THE LIVER David K. Bonauto Occupational Medicine, University of Washington Seattle, Washington, USA
C. Andrew Brodkin Department of Internal Medicine and Department of Environmental Health, University of Washington, Seattle, Washington, USA
William O. Robertson Washington Poison Center, University of Washington Seattle, Washington, USA
The toxic effects of organic solvent compounds on the liver are dependent on the intensity and duration of exposure, route of exposure, the intrinsic toxicity of the specific compound, as well as individual susceptibility factors.1 There are a number of pathologic manifestations of solvent induced hepatotoxicity, including inflammation, fat accumulation in the liver (steatosis), hepatocellular necrosis and carcinogenesis. Functional disturbances in liver physiology have also been associated with solvent exposure. The purpose of this chapter is to review the known hepatotoxicity of commonly used industrial solvents.2 A brief review of normal anatomic and physiologic function of the liver will be provided as a background for understanding histopathologic and biochemical changes associated with solvent toxicity. The final segment includes a discussion of solvents known to cause liver injury with a review of the available medical evidence suggestive of human hepatotoxicity of solvents at present day exposure levels. Solvent induced hepatotoxicity is almost exclusively encountered in an occupational setting and thus this review will focus on evidence culled from that setting. 20.8.1 NORMAL ANATOMIC AND PHYSIOLOGIC FUNCTION OF THE LIVER The liver is the largest internal organ and is involved in many physiologic processes including nutrient homeostasis, synthesis and excretion of bile, lipid metabolism and lipoprotein and protein synthesis.3 Most importantly for purposes of this chapter, the liver is the site of the biotransformation of a wide variety of endogenous and exogenous toxins.4 The ability of the liver to biotransform various chemicals is due to the multiple different enzyme systems contained within the hepatocytes.3,5 One such enzyme system is the cytochrome p 450 enzyme system. It consists of a large group of enzymes which biotransform many different substances by either oxidation or reduction to facilitate excretion from the body. Specifically different members of the cytochrome p 450 family catalyze reactions involving aromatic and aliphatic hydroxylation, epoxidation, dehalogenation, dealkylation, N-, Soxidation as well as O-, N-, S- dealkylation reactions.3,5 The diverse metabolic activities of the liver make it susceptible to solvent induced injury, particularly from reactive intermediates which damage cellular macromolecules. The microscopic anatomy of the liver provides an explanation for this susceptibility. The basic unit of the liver is the hepatic lobule which consists of a central vein surrounded radially by sinusoids of liver cells (hepatocytes). Portal triads consisting of a hepatic artery, a hepatic vein and a bile canniliculus are located at the periphery. Liver cells closest to the vascular
1394
David K. Bonauto, C. Andrew Brodkin, William O. Robertson
supply or the portal triad, zone one, are more resistant to oxidative stress, while hepatocytes near the central vein, zone three or centrizonal region, are most susceptible to solvent induced injury. 20.8.1.1 Factors Influencing Solvent Hepatotoxicity Bioavailability: The physical and chemical properties of a solvent and its toxicokinetics determine its availability to hepatic tissues. The primary route of absorption of most solvents which cause hepatotoxicity into human biological systems is via the lung. Therefore, the greater the volatility of the solvent, the greater its concentration in the air, and subsequently the larger the potential dose.4 While accidental or intentional ingestion of solvents is reported in the medical literature, it is an uncommon route of exposure in the occupational setting. Dermal absorption should be considered a significant route of exposure for most solvent compounds based on their lipid solubility. The degree of exposure can often be modified by the use of personal protective equipment such as gloves or a respirator and engineered exposure controls such as building ventilation. The lipid solubility of solvents also favors their deposition of into lipid rich organ systems such as the liver. The toxicity of a particular solvent may be enhanced by its long residence time in the liver.6 Genetic and environmental factors: While some solvents are directly hepatotoxic, frequently biotransformation of solvents by hepatic mixed function oxygenases, such as the cytochrome p-450 system, result in toxic intermediates.7 A variety of genetic and environmental factors inhibit or induce the activity of these hepatic enzyme systems, effecting the biotransformation and resulting toxicity. Genetic factors thought to determine the activity or even the presence of an enzyme within an organism center around human polymorphisms or variations in the genetic code.7 As the activities of the liver enzymes are changed so will the rate of formation of the metabolite thus increasing or decreasing the toxicity of the foreign substance.3 Environmental factors which determine the activity of liver enzyme systems include co-exposure to other drugs and toxins or characteristics of the individual particularly disease states which induce or inhibit the activity of liver biotransformation enzymes.1,3,7-9 Individual characteristics such as age, nutritional status, pregnancy or disease states such as diabetes or obesity may also change the activity of cytochrome p-450 enzymes.7,9 The assessment of an individual’s susceptibility to exposure should attempt to account for these variables in determining risk. 20.8.1.2 Microscopic, Biochemical and Clinical Findings Associated with Liver Injury due to Solvents Hepatotoxic manifestations associated with acute solvent exposure are dose dependent. Acute cytotoxic injury of a solvent directly or by its metabolites causes an alteration in the normal physiologic function leading to ballooning fatty change and ultimately cellular necrosis. If the dose is minimal and doesn’t exceed the ‘regenerative capacity’ of the liver, inflammatory changes generally resolve within two weeks to several months. Metabolic derangement results in the accumulation of fats in the liver, termed steatosis.10 A less common form of acute cytotoxic injury is related to cholestatic injury with disruption of normal biliary flow.11 Severe long term exposures can lead to fibrosis or scarring and cirrhosis which distort the hepatic architecture and lead to altered liver function.
20.8 Solvents and the liver
1395
Current research focuses on the effect of low doses of solvents on the liver, with concern that low grade prolonged solvent exposure could lead to chronic injury and eventual impairment.5,12 Tests used to evaluate and screen for liver injury can be divided into three general categories: serum biomarkers of disease, tests of hepatic clearance, and anatomic evaluation.11 The hepatic enzymes most commonly screened for related to hepatocellular necrosis and inflammatory changes are aspartate aminotransferase (AST), and alanine aminotransferase (ALT). Elevation of these enzymes in the setting of significant exposure is indicative of hepatotoxic injury, though alternative causes such as alcohol and viral hepatitis should be excluded. Importantly, serum hepatic transaminase levels indicate hepatocellular necrosis or inflammation, but may not indicate more subtle metabolic alterations in hepatic function. Measures of other hepatic enzymes, gamma glutamyl transpeptidase (GGT), alkaline phosphatase (Alk Phos), total and direct bilirubins may also be suggestive of solvent induced hepatotoxicity. Specifically if hepatic excretion of bile, is diminished, the resulting intrahepatic cholestasis is associated with elevations in GGT, Alk Phos and serum bile acids. Significant elevations of bilirubins leads to the clinical observation of jaundice or yellowing of the skin. However, pathologic obstruction of the biliary tract is not a common finding in solvent induced hepatotoxicity.5 Clearance tests of liver function assess a number of physiologic activities including hepatic uptake, hepatic metabolism, and hepatic excretion. Typical clearance tests of liver function include indocyanine green (ICG), antipyrine clearance test and 14C aminopyrine breath test. These tests give an estimation of the ability of the liver to extract and detoxify exogenous toxins (xenobiotics). Measuring the excretion of endogenously produced serum bile acids is an additional measure of hepatic clearance and has been used as a sensitive measure of early solvent hepatotoxicity.13,14 Anatomic evaluation of solvent hepatotoxicity centers on physical examination of the liver, radiologic study and liver biopsy. Physical exam is nonspecific as to the cause and characterization of the disease. Radiologic studies such as ultrasound can identify hepatobiliary disease and liver parenchymal disease, namely steatosis and fibrosis. Steatosis and fibrosis are noted on ultrasound by a change in the echogenicity of the liver. While liver biopsy is the ‘gold standard’ for anatomic evaluation of the liver, the invasiveness of the test, the morbidity and discomfort of the procedure, and its cost make it prohibitive for routine screening. It is usually reserved for definitive diagnostic and prognostic purposes. Algorithmic strategies for screening for liver injury and evaluation of abnormal results have been reported in several references.11,12,15,16 20.8.2 HEPATOTOXICITY ASSOCIATED WITH SPECIFIC SOLVENTS The following section presents specific classes of organic solvents strongly associated with hepatotoxicity in human populations or animal studies. While there is more limited evidence of hepatotoxicity related to inhalational and dermal exposure to aliphatic hydrocarbons, ketones, alcohols, aldehydes, esters and ethers, potential hepatotoxicity related to these agents must be assessed on an individual basis with regard to concentration, duration, and bioavailability of exposure.5,11 Variations in individual susceptibility must also be considered with regard to concurrent use of alcohol, mixed solvent exposure, underlying liver diseases (e.g., viral hepatitis, hemochromatosis, hypertriglyceridemia and diabetes) as well as demographic differences in hepatic metabolism.11 Given these limitations, the organic solvents of primary concern with regard to hepatotoxicity are the haloalkanes, haloalkenes,
1396
David K. Bonauto, C. Andrew Brodkin, William O. Robertson
dimethylformamide, and nitroparaffins.5,11 Other agents such as styrene have been associated with hepatotoxicity in some studies.17,18 Potential interactive effects of solvent mixtures should always be considered in the assessment of hepatotoxicity, even if composed of solvents not commonly associated with hepatotoxicity. 20.8.2.1 Haloalkanes and haloalkenes Some of the most extensively studied and most concerning hepatoxins are the haloalkane solvents. Major haloalkanes encountered industrially, with documented animal and human hepatotoxicity, are carbon tetrachloride, chloroform, 1,1,2,2-tetrachlorethane, methyl chloroform and 1,1,2-trichloroethane, tetrachloroethylene, and trichloroethylene.5,6,19 The relative hepatotoxicity of each is correlated inversely with the carbon chain length, and carbon halogen bond energy and correlated directly with the number of halogens on the molecule and the atomic number of the halogen.5,20 Some of these solvents have been eliminated from common industrial use due to their deleterious environmental and human effects, though they may still be encountered in specific processes and regions (e.g., developing countries). Carbon tetrachloride is the most extensively studied and serves as a model for hepatotoxicity for other haloalkanes.21,22 20.8.2.2 Carbon tetrachloride Carbon tetrachloride hepatotoxicity has been reported since the early twentieth century.23 The toxicological literature is extensive with regard to carbon tetrachloride hepatotoxicity in animals. 21,23 Human toxicological information derives primarily from accidental or intentional ingestion in humans or by inhalational exposure in groups of workers.21 The industrial use of carbon tetrachloride has declined precipitously, due to the recognized health effects and regulatory policy.21 Historically, it was used as a solvent in the manufacture of industrial chemicals, in the dry cleaning industry and even as an antiparasitic medication.5 Presently the main means of exposure is in research laboratory settings, or as low level environmental contaminant.21 Because it is so volatile, the main mode of carbon tetrachloride exposure in occupational setting is via inhalation, although exposure by the dermal route also occurs. Animal and human susceptibility to carbon tetrachloride hepatotoxicity is dependent on many different factors. There is substantial interspecies variation in carbon tetrachloride induced hepatotoxicity in animals due to differences in metabolic pathways among species.21 Based on animal models, hepatotoxicity in humans is most likely mediated from the trichloromethyl radical formed from the metabolism of carbon tetrachloride by hepatic cytochrome p 450 2E1.24 Animal studies suggest differential hepatotoxicity based upon the animal’s age and gender, with greater toxicity demonstrated in adult rats compared to newborns,25,26 and males compared with females.5 Cytochrome p-450 enzyme systems are present in the human fetus suggesting a potential for in utero liver toxicity.27 Human gender differences in the metabolism of carbon tetrachloride have not been demonstrated despite potential sex steroid influences on the cytochrome p-450 system.28 The hepatotoxic effects of carbon tetrachloride are more severe in the setting of alcohol consumption.21,29,30 Animal studies demonstrate that the temporal relationship between ethanol ingestion and carbon tetrachloride exposure determines the severity of toxicity.31-33 Maximal hepatotoxicity is derived from ethanol ingested eighteen hours preceding exposure to carbon tetrachloride,32,33 whereas exposure to ethanol three hours prior to carbon tetrachloride exposure leads to minimal hepatotoxicity.33 The mechanism for this interaction is
20.8 Solvents and the liver
1397
the induction of cytochrome p-450 enzymes leading to greater formation of toxic intermediates.34 In contrast, exposure to ethanol immediately preceding carbon tetrachloride exposure leads to competitive inhibition of carbon tetrachloride metabolism.34 Several other alcohols (e.g., isopropanol,32,35 t-butyl alcohol36) and ketones37 potentiate the effect of carbon tetrachloride hepatotoxicity. Exposures to other haloalkanes38 or haloalkenes39 potentiate carbon tetrachloride hepatotoxicity while carbon disulfide exposure is ‘protective’ of carbon tetrachloride hepatototoxicity.40 Dietary factors, medications, chronic diseases and persistent halogenated environmental contaminants such as PCB’s and DDT have been shown to modulate or exacerbate the hepatotoxicity of carbon tetrachloride.5,21 Cellular disruption leading to hepatocellular necrosis results from damage to cellular macromolecules by trichloromethyl radicals.24 Cellular disruption involves alteration of calcium homeostasis,41 impaired oxidative phosphorylation,42 and trichloromethyl radical binding to cellular proteins, nucleic acids, and induction of lipid peroxidation.6,21 Histologically there is preferential necrosis of zone three hepatocytes in the liver acinus so called centrizonal necrosis as well as zone three steatosis. As with other halogenated hydrocarbons, carbon tetrachloride is an intrinsic hepatotoxin with adverse effects occurring at predictable exposure levels. The American Conference of Governmental and Industrial Hygienists (ACGIH) Threshold Limit Value (TLV), for carbon tetrachloride, based upon animal and human exposure data where limited adverse health effects are observed, is 5 ppm over an 8 hour time weighted average and a 40 hour work week for carbon tetrachloride.43 In human population studies, elevations in hepatic transaminase levels occur at carbon tetrachloride concentrations averaging 200 ppm, with small but significant elevations of ALT, AST, Alk Phos and GGT occurring at exposure levels below the TLV.44-46 Carbon tetrachloride also affects many other organ systems, specifically the central nervous system, the gastrointestinal tract, the liver and the kidney.5,6 Hepatic manifestations of carbon tetrachloride include serum AST and ALT elevations as early as three hours following exposure. Clinical evidence of hepatic disease occurs approximately twenty four hours following exposure, and is manifest in half the cases as jaundice accompanied by hepatic enlargement. In severe poisonings, progressive hepatic injury leads to coma and death within a week of exposure. Fortunately, non lethal exposures are often associated with significant clinical recovery in two to three weeks. Treatment is limited to supportive care, in a hospital setting. Chronic exposures to carbon tetrachloride have been associated with hepatic fibrosis and cirrhosis in animals and documented as well in several case reports in humans.47-50 20.8.2.3 Chloroform Medical and industrial use of chloroform has also declined significantly.51 Today, industrial use is limited to the manufacture of refrigerants and fluoropolymers.51 Chloroform metabolism involves the same cytochrome p-450 2E1 as carbon tetrachloride but with oxidation of chloroform to trichloromethanol with spontaneous formation of phosgene via the elimination of hydrochloric acid.3 In turn, phosgene reacts with hepatic lipids and microsomal proteins and depletes cellular glutathione, a cellular antioxidant.7,51 Factors potentiating chloroform hepatotoxicity include ethanol and other alcoho1s,52-54 hypoxia,53 ketones,55 fasting state,56,57 concomitant chronic medical disease and chronic medication use, or those with repeated exposures to chloroform.5 The pattern of human liver injury associated from chloroform poisoning is centrilobular necrosis and steatosis.23
1398
David K. Bonauto, C. Andrew Brodkin, William O. Robertson
The ACGIH has set a TLV of 10 ppm over an 8 hour time weighted average and a 40 hour work week for chloroform. Because chloroform is a potential carcinogen, the lowest possible exposure is recommended. Occupational hepatotoxicity below the ACGIH TLV has been demonstrated, with evidence of adverse effects between 2 and 10 ppm.51,58 Clinical manifestations of chloroform toxicity involve multi-organ system effects including damage to the central nervous system, the kidney and lung as well as the liver.59 Fulminant toxic hepatitis appears within one to three days following exposure, with death at approximately one week in severe poisonings.57 In nonfatal cases, hepatic inflammatory changes,with hepatomegaly and transaminitis can occur within hours.57 Ingestion or significant inhalational exposure should be managed in a closely monitored hospital setting. 20.8.2.4 Dichloromethane Dichloromethane is commonly used as a degreaser and a paint stripper. It is metabolized in the liver by the cytochrome p-450 pathway to produce carbon monoxide.60 An independent pathway of metabolism occurs via conjugation with glutathione.60 Animal experimentation has demonstrated hepatotoxicity at near lethal concentrations of dichloromethane.61,62 Dichloromethane potentiates carbon tetrachloride hepatotoxicity in rat livers.38 Short term exposure to both ethanol and dichloromethane demonstrate an antagonistic relationship, while chronic exposure potentiates hepatotoxicity.63 Cases of human hepatotoxicity to dichloromethane have been reported.62,64 Workers in an acetate fiber production plant, exposed to 140 to 475 ppm of dichloromethane, with concomitant exposures to acetone and methanol, were observed to have elevated bilirubin and ALT levels relative to workers exposed to acetone alone.64 Bilirubin elevations were dependent on the level of dichloromethane exposure.64 Other studies have shown no significant effects in the range of 5 to 330 ppm of dichloromethane.65 Chronic exposure (greater than 10 years) to dichloromethane levels greater than 475 ppm was not associated with significant elevations in liver function tests.66 There is minimal evidence of human hepatotoxicity of dichloromethane less than the ACGIH TLV of 50 ppm over an 8 hour time weighted average.43 20.8.2.5 Trichloroethanes There are two isomers of trichloroethane, namely methyl chloroform and 1,1,2-trichloroethane. Animal hepatotoxicity to 1,1,2-trichloroethane is documented in the literature67 with potentiation of toxicity in association with acetone,68 isopropyl alcohol69 and ethanol.70 Hepatotoxicity, with steatosis, necrosis, elevated serum enzymes, and increased liver weight have been observed in animal models exposed to 1000 ppm of methyl chloroform.71 Human studies consist of case reports documenting hepatotoxicity, with elevated serum transaminases and fatty liver disease related to 1,1,1-trichloroethane exposure.72,73 Epidemiologic evidence suggests little hepatotoxicity related to this agent at exposure levels <350 ppm.74,75 20.8.2.6 1,1,2,2-Tetrachloroethane Though rarely used in current practice, this solvent was an important cause of hepatotoxicity in the past. Its hepatotoxic potential was first identified during its use in the first World War.5 Animal hepatotoxicity with fatty degeneration of the liver has been documented in multiple species.76 Human inhalational exposures manifest in liver enlargement, jaundice, steatosis with subsequent liver failure in severe poisonings.77,78 Subacute exposure periods of weeks to months is generally required for hepatic injury.77 Liver regeneration oc-
20.8 Solvents and the liver
1399
curs after nonfatal exposures.77 The mechanism of hepatotoxicity has not been elucidated in humans but the reactive metabolites 1,1-dichloroacetyl chloride with binding to hepatic macromolecules may play a role.76 Metabolism of 1,1,2,2-tetrachloroethane is potentiated by fasting and ethanol in rats.79,80 There is little documentation of any precise inhalational exposure levels necessary to cause hepatotoxicity. 20.8.2.7 Tetrachloroethylene and trichloroethylene This widely used dry cleaning agent and degreasing agent is associated with hepatotoxic effects.81,82 Cases of human hepatotoxicity to tetrachloroethylene at exposure levels greater than 100 ppm have been reported in the literature.83,84 Humans exposed to tetrachloroethylene at dosages up to 150 ppm for durations of one to five 8 hour shifts had no difference in hepatic enzyme levels from baseline levels.85 Studies of workers chronically exposed to concentrations of tetrachloroethylene less than 50 ppm showed no difference in liver enzyme levels, relative to groups of workers who did not have the exposure.82 However, dry cleaning workers chronically exposed to low levels of tetrachloroethylene at less than 25 ppm had evidence of an alteration in hepatic echogenicity relative to non-exposed workers.80 This is suggestive evidence that steatosis may occur at levels below the ACGIH TLV, without associated alterations in serum hepatic enzymes. The long term effects of exposures have not been well characterized. Wide spread use of trichloroethylene occurs in the dry cleaning industry and industrially as a degreasing agent. Historical use as an anesthetic generally suggests little acute hepatotoxicity.86,87 Longer term exposures in an occupational setting are associated with elevations in serum transaminases, with variable findings in epidemiologic studies.88-90 Exposures below the ACGIH TLV of 50 ppm in workers using trichloroethylene as a cleaning agent found elevated levels of serum bile acids.45,91 Hepatotoxicity is potentiated by alcohol,92 isopropanol and acetone.69 The long term effects of subclinical exposures are not known. 20.8.2.8 Other halogenated hydrocarbons Vinyl chloride, a gas at normal temperature and pressure, has solvent properties at high pressures; its industrial use as a monomer in the manufacture of polyvinylchloride and hepatotoxicity with chronic exposure make it an important public health risk. Vinyl chloride is associated with angiosarcoma,93,94 a rare highly malignant hepatic tumor, hepatic fibrosis,6 hepatocellular injury95 and hepatoportal sclerosis, a form of noncirrhotic portal hypertension.96,97 Appearance of angiosarcoma and hepatoportal sclerosis occurred in workers after decade long exposures.94,98 Measures to limit both occupational and environmental exposures have been instituted to decrease potential hepatic outcomes, with effective screening programs using indocyanine green clearance tests.6 Haloalkanes other than the chloroalkanes, especially those with structural homology to known hepatotoxic chloroalkanes, should be considered potentially hepatotoxic despite little industrial use as solvents.5,20 Case reports of bromoethane and hydrochlorofluorocarbon poisonings with hepatotoxicity have been reported in the literature.99-102 20.8.2.9 Styrene and aromatic hydrocarbons Styrene is not only used as a monomer in the production of polystyrene but also as a reactive solvent in the manufacture of unsaturated polyester resins.103 The hepatic metabolism of styrene involves the formation of the reactive intermediate styrene 7,8-oxide.104 In rat models, styrene 7,8-oxide binds to hepatic macromolecules and lipids causing hepatocellular in-
1400
David K. Bonauto, C. Andrew Brodkin, William O. Robertson
jury.105,106 Epidemiologic investigations of workers exposed to high concentrations (greater than 50 ppm) of styrene have shown elevations in GGT, AST, ALT,107-109 and serum bilirubin levels.110 At the ACGIH TLV of 50 ppm or less, evidence of transaminase and GGT elevations111,112 are lacking but elevated levels of serum bilirubins113 and bile acids17,110 have been demonstrated. There is no evidence of alterations in hepatic echogenicity at exposure levels less than 50 ppm.18 Toluene, benzene and xylenes are generally considered to have limited hepatotoxicity.6,114-118 Exposure to xylene is reported to cause mild steatosis.5 Exposure to a mixture of solvents, inclusive of xylene and toluene have been reported to produce elevated serum bile acids.13 20.8.2.10 N-substituted amides Two important N-substituted amides are dimethylformamide and dimethylacetamide. Dimethylformamide is used in the fabrication of synthetic textiles such as rayon. Its hepatotoxicity has been well demonstrated in occupational settings.119-121 Evidence of dose dependent alcohol intolerance and subjective gastrointestinal symptoms (abdominal pain, anorexia and nausea) have been described.122 Objective clinical and biochemical signs include elevations of transaminases, AST and ALT, hepatomegaly and abnormal liver biopsy findings demonstrating hepatocellular necrosis and steatosis.120,123 Workers with acute toxicity related to DMF have more severe symptoms and higher transaminase levels than workers with toxicity related to chronic exposures.124 Of significance, symptoms may occur under the ACGIH TLV of 10 ppm. Dermal absorption is a main exposure pathway in addition to inhalation.119-121 Dimethylacetamide is used as a solvent in the manufacture of plastics and as a paint remover. Occupational poisoning and hepatotoxicity to extreme concentrations of dimethylacetamide (DMA) are reported in the medical literature.125 Decreases in hepatic clearance measures and alterations in hepatic transaminases with hepatomegaly have been reported at lower doses.126 Like dimethylformamide, DMA is readily absorbed through the skin. Chronic exposures in workers exposed to low air concentrations of DMA of less than 3 ppm and with biological monitoring assessments to measure dosages by dermal absorption demonstrated little evidence of hepatotoxicity by clinical chemistries.127 20.8.2.11 Nitroparaffins The well known hepatotoxicity of nitroaromatic compounds such as trinitrotoluene lends suspicion to the hepatotoxicity of the nitroparaffins.114 Nitromethane and nitroethane produce steatosis in animal models, but there is limited evidence of hepatotoxicity of these agents in humans.114 Evidence for the hepatoxicity of 2-nitropropane has been raised by case reports and case series of occupational fatalities in settings of severe exposure.128,129 In these cases the lack of appropriate industrial hygienic measures such as adequate ventilation, and personal protective equipment contributed to the severity of the exposures.130 Autopsies of the fatal cases revealed hepatocellular necrosis and fatty infiltration of the liver.128 No significant evidence of hepatotoxicity has been demonstrated below the ACGIH TLV of 10 ppm.131 Medical surveillance of workers exposed to less than 25 ppm of 2-nitropropane have not shown alterations in liver chemistries.132
20.8 Solvents and the liver
1401
20.8.2.12 Other solvents and mixed solvents Suggestive evidence for hepatotoxicity of many compounds exist in the literature.5 Two solvents with some suspicion for hepatotoxic potential in humans are tetrahydrofuran and 1,4-dioxane, both solvents used in industry.133-135 Cases of tetrahydrofuran induced hepatotoxicity have been reported in the literature.133 Tetrohydrofuran's inhibition of the cytochrome p-450 enzyme system lends biologic credibility to it being a hepatotoxin.134 1,4-dioxane is reported to be hepatotoxic but epidemiologic evidence in human populations for this is limited.135 Rarely do solvents exist in isolation and thus evaluation of hepatotoxicity must consider the effects of mixtures of solvents.136 Alterations in the hepatotoxic potential of a chemical may exist, especially when the biotransforming enzymes are modulated or effected by various components of the mixture. Usual mechanisms for the potentiation of toxicity by alcohols, ketones may be altered when solvents are mixed. In such settings hepatotoxicity may occur below recommended levels.137 Much remains unknown regarding the hepatotoxic effects of compounds. For this reason, vigilance regarding the potential adverse hepatic effects of chemicals is appropriate. Maintaining active surveillance for solvent induced hepatotoxicity is important in protecting workers’ health, and will further our knowledge of the hepatotoxic effects of solvents. With emerging knowledge, occupational and environmental standards can be refined to further protect the health of workers and the public. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
H. Zimmerman in Schiff’s Diseases of the Liver, E. Schiff, M. Sorrell and W. Maddrey, Eds., LippincottRaven Publishers, Philadelphia, 1999, pp.973 -1064. J. Rosenberg in Occupational and Environmental Medicine, 2nd ed., J. LaDou. Ed., Appleton and Lange, Stamford, Conn., 1997, pp. 359-386. A. Parkinson in Casarrett and Doull’s Toxicology; The Basic Science of Poisons, 5th ed., C. Klaasen, Ed., McGraw-Hill, New York, 1996, pp. 113-186. M. Ellenhorn, S. Schonwald,G. Ordog, and J. Wasserberger eds., Ellenhorn’s Medical Toxicology: Diagnosis and Treatment of Human Poisoning. Williams and Wilkins. Baltimore, 1997. H. Zimmerman, Hepatotoxicity: The Adverse Effects of Drugs and Other Chemicals on the Liver. 2nd Ed., Lippincott, Williams and Wilkins, Philadelphia, 1999. N. Gitlin in Hepatology; A Textbook of Liver Disease. 3rd ed., D. Zakim and T. Boyer, Eds., WB Saunders Co., Philadelphia, 1996, pp. 1018-1050. J. Raucy, J. Kraner, and J. Lasker, Crit. Rev. Toxicol., 23, 1 (1993). P. Watkins, Semin. Liver Dis., 10, 235 (1990). D. Vessey in Hepatology; A Textbook of Liver Disease. 3rd ed., D. Zakim and T. Boyer, Eds., WB Saunders Co., Philadelphia, 1996, pp. 257-305. G. Lundqvist, U. Flodin, and O. Axelson, Am. J. Ind. Med., 35, 132 (1999). C. Redlich and C. Brodkin in Textbook of Clinical Occupational and Environmental Medicine, L. Rosenstock and M. Cullen, Eds., WB Saunders, Philadelphia, 1994, pp. 423-436. C. Tamburro and G. Liss, J. Occup. Med., 28, 1034 (1986). G. Franco, R. Fonte, G. Tempini, and F. Candura, Int. Arch. Occup. Environ. Health, 58, 157 (1986). G. Franco, Br. J. Ind. Med., 48, 557 (1991). D. Herip, Am. J. Ind. Med., 21, 331 (1992). R. Harrison, Occupational Medicine: State of the Art Reviews, 5, 515 (1990). C. Edling and C. Tagesson, Br. J. Ind. Med., 41, 257 (1984). C. Brodkin, J. Moon, D. Echeverria, K. Wang, and H. Checkoway, ICOH 2000 Congress, (Abstract In Press). R. Harbison, Hamilton and Hardy’s Industrial Toxicology. 5th ed., Mosby, St. Louis, 1998. G. Plaa and W. Hewitt in Toxicology of the Liver, G. Plaa and W. Hewitt, Eds., Raven Press, New York, 1982, pp. 103-120.
1402
21
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59
60 61 62
63 64
David K. Bonauto, C. Andrew Brodkin, William O. Robertson
O. Faroon, J. Riddle, Y. Hales and W. Brattin, Toxicological Profile for Carbon Tetrachloride. Agency for Toxic Substances and Disease Registry (ATSDR), US Department of Health and Human Services, Atlanta, 1994. R. Recknagel, Pharmacol. Rev., 19, 145 (1967). W. Von Oettingen, The Halogenated Hydrocarbons of Industrial and Toxicological Importance, Elsevier, Amsterdam, 1964. R. Recknagel and E. Glende, Jr., Crit. Rev. Toxicol., 2, 236 (1973). M. Dawkins, J. Pathol. Bacteriol., 85, 189 (1963). S. Cagen and C. Klaasen, Toxicol. Appl. Pharmacol., 50, 347 (1979). G. Mannering, Fed. Proc., 44, 2302 (1985). J. Gustaffson, Annu. Rev. Physiol., 45, 51 (1983). M. Manno, M. Rezzadore, M. Grossi, and C. Sbrana, Hum. Exp. Toxicol., 15, 294 (1996). P. New, G. Lubash, L. Scherr, and A. Rubin, JAMA, 181, 903 (1962) H. Ikatsu, T. Okino, and T. Nakajima, Br. J. Ind. Med., 48, 636 (1991). G. Traiger and G. Plaa, Toxicol. Appl. Pharmacol., 20, 105 (1971). H. Cornish and J. Adefuin, Am. Ind. Hyg. Assoc. J., 27, 57 (1966). T. Castillo, D. Koop, S. Kamimura, G. Triadafilopoulos, and H. Tsukamoto, Hepatology, 16, 992 (1992). D. Folland, W. Schaffner, H. Ginn, O. Crofford, and D. McMurray, JAMA, 236, 1853 (1976). R. Harris and M. Anders, Toxicol. Appl. Pharmacol., 56, 191 (1980). G. Plaa, Fundam. Appl. Toxicol., 10, 563 (1988). Y. Kim, Fundam. Appl. Toxicol., 35, 138 (1997). D. Pessayre, B. Cobert, V. Descatoire, C. Degott, G. Babany, C. Funck-Brentano, M. Delaforge and D. Larrey, Gastroenterology, 83, 761 (1982). Y. Masuda and N. Nakayama, Biochem. Pharmacol., 31, 2713 (1982). E. Glende and R. Recknagel, Res. Commun. Chem. Pathol. Pharmacol., 73, 41 (1991). G. Christie and J. Judah, Proc. Roy. Soc. Lond. B., 142, 241 (1954). American Conference of Governmental Industrial Hygienists (ACGIH), Threshold Limit Values(TLVs) for chemical substances and physical agents and Biological Exposure Indices (BEIs), American Conference of Governmental Industrial Hygienists, Cincinnati, 1996. R. Barnes and R. Jones, Am. Ind. Hyg. Assoc. J., 28, 557 (1967). T. Driscoll, H. Hamdan, G. Wang, P. Wright, and N. Stacey, Br. J. Ind. Med., 49, 700 (1992). J. Tomenson, C. Baron, J. O’Sullivan, J. Edwards, M. Stonard, R. Walker, D. Fearnley, Occup. Env. Med., 52, 508 (1995). G. Cameron and W. Karunatne, J. Pathol. Bacteriol., 42, 1 (1936). C. Poindexter and C. Greene, JAMA, 102, 2015 (1934). N. Gitlin, S.A. Med. J., 58, 872 (1980). T. Paerez, Hepatology, 3, 112 (1983). S. Chou, W. Spoo, Toxicological Profile for Chloroform. Agency for Toxic Substances and Disease Registry (ATSDR), US Department of Health and Human Services, Atlanta, 1997. S. Kutob and G. Plaa, J. Pharmacol. Exposure Therap., 135, 245 (1962). K.Hutchens and M. Kung, Am. J. Med., 78, 715 (1985). S. Ray and H. Mehendale, Fundam. Appl. Toxicol., 15, 429 (1990). J. Brady, D. Li, H. Ishizaki, M. Lee, S. Ning, F. Xiao, and C. Yang, Toxicol. Appl. Pharmacol., 100, 342 (1989). D. McMartin, J. O’Connor Jr., and L. Kaminsky, Res. Commun. Chem. Pathol. Pharmacol., 31, 99 (1981). S. Winslow and H. Gerstner, Drug Chem. Toxicol., 1, 259 (1978). L. Li, X. Jiang, Y. Liang, Z. Chen, Y. Zhou, and Y. Wang, Biomed. Environ. Sci., 6, 179 (1993). Occupational Safety and Health Administration(OSHA), Occupational Safety and Health Guidelines for Chloroform, available at http://www.osha.slc.gov/SLTC/ healthguidelines/chloroform/recognition.html, 2000. R. Snyder and L. Andrews in Casarrett and Doull’s Toxicology; The Basic Science of Poisons, 5th ed., C. Klaasen, Ed., McGraw-Hill, New York, 1996, pp. 737-771. K. Mizutani, K. Shinomiya, and T. Shinomiya, Forensic Sci. Int., 38, 113 (1988). Occupational Safety and Health Administration, OSHA Federal Register: Occupational Exposure to dichloromethane-62: 1494-1619, available at http://www.osha-slc.gov/FedReg_osha_data/FED19970110.html, 2000. M. Balmer, F. Smith, L. Leach, and C. Yuile, Am. Ind. Hyg. Assoc. J., 37, 345 (1976). M. Ott, L. Skory, B. Holder, J. Bronson, and P. Williams, Scand. J. Work Environ. Health, 9 S(1), 1 (1983).
20.8 Solvents and the liver
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
104 105 106 107 108 109 110
1403
H. Anundi, M. Lind, L. Friis, N. Itkes, S. Langworth, and C. Edling, Int. Arch. Occup. Env. Health, 65, 247 (1993). K. Soden, J. Occup. Med., 35, 282 (1993). Syracuse Research Corporation, Toxicological Profile for 1,1,2-Trichloroethane, Agency for Toxic Substances and Disease Registry (ATSDR), US Department of Health and Human Services, Atlanta, 1989. J. MacDonald, A. Gandolfi, I. Sipes, and J. MacDonald, Toxicol. Lett., 13, 57 (1982). G. Traiger and G. Plaa, Arch. Environ. Health, 28, 276 (1974). C. Klaasen and G. Plaa, Toxicol. Appl. Pharmacol., 9, 139 (1967). M. Williams and F. Llados, Toxicological Profile for 1,1,1-Trichloroethane. Agency for Toxic Substances and Disease Registry (ATSDR), US Department of Health and Human Services, Atlanta, 1995. M. Hodgson, A. Heyl, and D. Van Thiel, Arch. Intern. Med., 149, 1793 (1989). C. Cohen and A. Frank, Am. J. Ind. Med., 26, 237 (1994). C. Kramer, H. Imbus, M. Ott, J. Fulkerson, and N Hicks, Arch. Environ. Health, 38, 331 (1978). R. Stewart, H. Gay, A. Schaffer, D. Erley, and V. Rowe, Arch. Environ. Health, 19, 467 (1969). L. Smith and J. Mathews, Toxicological Profile for 1,1,1,2-Tetrachloroethane, Agency for Toxic Substances and Disease Registry (ATSDR), US Department of Health and Human Services, Atlanta, 1994. H. Coyer, Ind. Med., 13, 230 (1944). R. Gurney, Gastroenterology, 1, 1112 (1943). T. Nakajima and A. Sato, Toxicol. Appl. Pharmacol., 50, 549 (1979). A. Sato, T. Nakajima, and Y. Koyama., Br. J. Ind. Med., 37, 382 (1980). C. Brodkin, W. Daniell, H. Checkoway, D. Echeverria, J. Johnson, K. Wang, R. Sohaey, D. Green, C. Redlich, D. Gretch, and L. Rosenstock, Occup. Env. Med., 52, 679 (1995). P. Gennari, M. Naldi, R. Motta, M. Nucci, C. Giacomini, F. Violante, and G. Raffi, Am. J. Ind. Med., 21, 661 (1992). G. Saland, N.Y.S. J. Med., 67, 2359 (1966). L. Meckler, and D. Phelps, JAMA, 197, 662 (1966). R. Lauwerys, J. Herbrand, J. Buchet, A. Bernard, and J. Gaussin, Int. Arch. Occup. Environ. Health, 52, 69 (1983). G. Smith, Br. J. Ind. Med., 23, 249 (1966). R. Defalque, Clin. Pharmacol. Ther., 2, 665 (1961). R. McCunney, Br. J. Ind. Med., 45, 122 (1988). G. Bond, J. Toxicol. Clin. Toxicol., 34, 461 (1996). T. Nagaya, N. Ishikawa, H. Hata, and T. Otobe, Int. Arch. Occup. Environ. Health, 64, 561 (1993). M. Neghab, S. Qu, C. Bai, J. Caples, and N. Stacey, Int. Arch. Occup. Environ. Health, 70, 187 (1997). G. Müller, M. Spassowski, and D. Henschler, Arch. Toxicol., 33, 173 (1975). J. Creech Jr., and M. Johnson, J. Occup. Med., 16, 150 (1974). F. Lee, P. Smith, B. Bennett, and B. Williams, Gut, 39, 312 (1996). S. Ho, W. Phoon, S. Gan, and Y. Chan, J. Soc. Occup. Med., 41, 10 (1991). P. Smith, I. Crossley and D. Williams, Lancet, 2 (7986), 602 (1976). P. Bioulac-Sage, B. LeBail, P. Bernard, and C. Balabaud, Semin. Liver Dis., 15, 329 (1995). W. Lelbach, Am. J. Ind. Med., 29, 446 (1996). A. Van Haaften, Am. Ind. Hyg. Assoc. J., 30, 251 (1969). P. Hoet, M. Graf, M. Bourdi, L. Pohl, P. Duray, W. Chen, R. Peter, S. Nelson, N. Verlinden, and D. Lison, Lancet., 350, 556 (1997). D. Anders and W. Dekant, Lancet, 350, 1249 (1997). G. Rusch, Lancet, 350, 1248 (1997). P. Pfaffli and A. Saamanen in Butadiene and Styrene: Assessment of Health Hazards, IARC Scientific Publications No 127, M. Sorsa, K. Peltonen, H. Vainio, and K. Hemmiki Eds. International Agency for Research on Cancer, Lyon, 1993, pp. 15 - 33. J. Bond, Crit. Rev. Toxicol., 19, 227 (1989). J. Marniemi in Microsomes and Drug Oxidations, V. Ullrich ed., Pergamon, New York, 1977, pp. 698-702. J. Van Anda, B. Smith, J. Fouts, and J. Bend, J. Pharmacol. Exp. Ther., 211, 207 (1979). O. Axelson O and J. Gustavson, Scand. J. Work Environ. Health, 4, 215 (1979). G. Triebig, S. Lehrl, D. Weltle, K. Schaller, and H. Valentin, Br. J. Ind. Med., 46, 799 (1989). A. Thiess, and M. Friedheim, Scand. J. Work Environ. Health, 4 (S2), 220 (1978). R. Vihko in Biological Monitoring And Surveillance Of Workers Exposed To Chemicals, A. Aitio, V. Riihimäki, and H. Vainio Eds., Hemisphere Publishing Corp, Washington, D.C., 1984, pp. 309-313.
1404
Kaye H. Kilburn
111 W. Lorimer, R. Lilis, W. Nicholson,, H. Anderson, A. Fischbein, S. Daum, W. Rom, C. Rice, and I. Selikoff, Environ. Health Perspect., 17, 171 (1976). 112 H. Harkonen, A. Lehtniemi, and A. Aitio, Scand. J. Work Environ. Health, 10, 59 (1984). 113 C. Brodkin, Personal Communication. 114 H. Zimmerman and J. Lewis, Gastroenterol. Clin. North. Am., 24, 1027 (1995). 115 L. Low, J. Meeks, and C. Mackerer, Toxicol. Ind. Health, 4, 49 (1988). 116 P. Guzelian, S. Mills, and H. Fallon, J. Occup. Med., 30, 791 (1988). 117 R. Morris, J. Occup. Med., 31, 1014 (1989). 118 C. Boewer, G. Enderlein, U. Wollgast, S. Nawka, H. Palowski, and R. Bleiber, Int. Arch. Occup. Environ. Health, 60, 181 (1988). 119 V. Scailteur and R. Lauwerys, Toxicology, 43, 231 (1987). 120 C. Redlich, W. Beckett, J. Sparer, K. Barwick, C. Riely, H. Miller, S. Sigal, S. Shalat and M. Cullen, Ann. Int. Med., 108, 680 (1988). 121 A. Fiorito, F. Larese, S. Molinari, and T. Zanin, Am. J. Ind. Med., 32, 255 (1997). 122 S. Cai, M. Huang, L. Xi, Y. Li, J. Qu, T. Kawai, T. Yasugi, K. Mizunuma, T. Watanabe, and M. Ikeda, Int. Arch. Occup. Environ. Health, 63, 461 (1992). 123 L. Fleming, S. Shalat, and C. Redlich, Scand. J. Work Environ. Health, 16, 289 (1990). 124 C. Redlich, A. West, L. Fleming, L. True, M. Cullen, and C. Riely, Gastroenterology, 99, 748 (1990). 125 G. Marino, H. Anastopoulos, and A. Woolf, J. Occup. Med., 36, 637 (1994). 126 G. Corsi, Med Lav, 62, 28 (1971). 127 G. Spies, R. Rhyne Jr., R. Evans, K. Wetzel, D. Ragland, H. Turney, T. Leet, and J. Oglesby, J. Occup. Med., 37, 1102 (1995). 128 C. Hine, A. Pasi, and B. Stephens, J. Occup. Med., 20, 333 (1978). 129 R. Harrison, G. Letz, G. Pasternak, and P. Blanc, Ann. Int. Med., 107, 466 (1987). 130 D. Hryhorczuk, S. Aks, and J. Turk, Occup Med: State of Art Reviews, 7, 567 (1992). 131 T. Lewis, C. Ulrich, and W. Busey, J. Environ. Pathol. Toxicol., 2, 233 (1979). 132 G. Crawford, R. Garrison, and D. McFee, Am. Ind. Hyg. Assoc. J., 46, 45 (1985). 133 R. Garnier, N. Rosenberg, J. Puissant, J. Chauvet, and M. Efthymiou, Br. J. Ind. Med., 46, 677 (1989). 134 D. Moody, Drug Chem. Toxicol., 14, 319 (1991). 135 C. DeRosa, S. Wilbur, J. Holler, P. Richter, and Y. Stevens, Toxicol. Ind. Health., 12, 1 (1996). 136 F. Tomei, P. Giuntoli, M. Biagi, T. Baccolo, E. Tomao, and M. Rosati, Am. J. Ind. Med., 36, 54 (1999). 137 E. Sotaniemi, S. Sutinen, S. Sutinen, A. Arranto, and R. Pelkonen, Acta Med. Scand., 212, 207 (1982).
20.9 TOXICITY OF ENVIRONMENTAL SOLVENT EXPOSURE FOR BRAIN, LUNG AND HEART Kaye H. Kilburn School of Medicine, University of Southern California Los Angeles, CA, USA
This chapter considers the neurobehavioral effects of environmental exposures to organic solvents. Much information applicable to environmental or community exposures usually at home came from animal experiments, brief human exposures in chambers and prolonged workplace exposures. The mode of entry of solvent chemicals into the body is almost always by inhalation not by contact or ingestion.1 While inhalational exposures to single chemicals occur in the community mixtures are usual making measurements more complex. Effects from animal experiments, and human exposures in chambers, and workplace exposures are usually consistent and help predict environmental effects. The major categories of environmental exposures to solvents are from petroleum refining to consumer use indoors, Table 20.9.1. Sometimes adverse human effects are from surprisingly small environmental doses, an order of magnitude or two lower than those needed for workplace
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
1405
effects. One possible explanation is greater sensitivity of measurements but many of the methods were adapted from studies of workers.2 Table 20.9.1. Sources of environmental exposure to solvents Processes
Chemicals
Media
Example
Losses during refining and chemical production
MTBE
air
Seymour, IN, Santa Maria, CA
Losses from use in industry
TCE + toluene
water surface, ground water, air, water
Phoenix, AZ, Motorola, Printers, Baton Rouge, LA, Abuse-glue sniffers
air
Avila Beach, CA, Livington Parish, LA
Leaks and spills during transporta- toluene, tion (pipeline, truck, rail, ship) xylene, PAH Combustion: a. Fires as incidents, b. Incineration of fuel air pollution
air
TCE
water, ground water 800 national hazardous acid pits
c. Incineration of garbage Contaminated sites
Wilmington, CA, San Bernardino, CA, Los Angeles, Houston, TX, Mexico City Oak Ridge, TN, Walker, LA
hydrocarbon particles
Outgassing indoors of forest products-like particle board, carpets, TCA drapes, adhesives
air
Indoor air incidents, Sick building syndrome
+ - other chlorinated solvents; TCA - trichloroethane; MTBE - methyl ter butyl ether, additive in gasoline; PAH polyaromatic hydrocarbons, example benzo(a)pyrene
Table 20.9.2. Differences in environmental and occupational toxicology Occupational Subjects health age, years selection positive
Environmental
healthy 18-60 selected for employment attenuated by losses of sick
chronic illness 0-100 unselected sick collect
Duration in a week total
40 hours years
168 hours lifetime
Source
raw materials processes
leaks and spills (gasoline) outgassing of consumer goods fuel combustion
Chemical exposure
one or few
many
Monitoring exposure
area or personal
rarely possible
Environmental transformation of agents
unusual
frequent
1406
Kaye H. Kilburn
Differences between exposure at work and in the community are important, Table 20.9.2. Most worker groups were younger and healthier, met job-entry criteria, have more reserve function so are less likely to manifest damage. Workers have had selective attrition of affected or less fit people to accentuate the difference.3 In contrast community people are unselected and include more susceptible groups: infants, children, the aged and the unwell.1 Some differences in exposure are obvious, work exposure is rarely longer than 40 of the weeks 168 hours. This time away allows work acquired body burdens of chemicals (and their effects) to diminish or disappear while workers are at home. In contrast, home exposures may be continuous or nearly so.4 At work the time that elapsed between exposure and effect is short, making measurement of the dose of a toxic agent easy. It is less obvious what should be measured in community exposures. Good detective work is needed to specify the chemicals to search for and measure in air, water or soil. Community effects may take years to be recognized as a problem. Opportunities for pertinent environmental measurements were overlooked and have disappeared with time, often simply evaporated. Thus measurement of relevant doses are seldom possible and dose-response curves can rarely be constructed. The logical surrogates for dose such as distance and direction from a chemical source and duration are rarely satisfactory.1 Thus plausible estimates of dose are needed to focus the association of measured effects and the chemicals that are probably responsible. The realistic starting place is people’s symptoms-complaints that indicate perception of irritation from chemicals.3 These serve as sentinels to alert one to a problem but cannot be interpreted as impairment or damage without measurements of brain functions. The inability to characterize exposure should not postpone or prevent adequate investigation for adverse human health effects. It is intuitive and ethical to suggest that absent of adverse human effects should be the only reason for stopping inquiry. People’s complaints and upset moods (anxiety, depression, anger, confusion and fatigue) frequently reflect or parallel impairment. The question then becomes how to measure effects on the brain to decide whether it is damaged and if so how much, Table 20.9.3. Table 20.9.3. Useful tests of evaluation of brain damage from solvents Tests
Part of brain measured
Simple reaction time & visual two choice reaction time
retina, optic nerve and cortex integrative radiation to motor cortex
Sway-balance
inputs: ascending proprioceptive tracts, vestibular division 8th cranial nerve, cerebellum, vision, visual integrative and motor tracts
Blink reflex latency
sensory upper division trigeminal nerves (V), pons, facial nerves (VII)
Color confusion index
center macular area of retina, with optic cones, optic nerve, optic occipital cortex
Visual fields
retina-optic nerve-optic cortex occipital lobe
Hearing
auditory division of 8th cranial nerve
Verbal recall memory
limbic system of temporal lobe, smell brain
Problem solving culture fair & digit symbol
cerebral cortex: optic-occipital and parietal lobe cortex
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
Tests
1407
Part of brain measured
Vocabulary
long-term memory, frontal lobes
Information, picture completion & similarities
long-term memory, frontal lobes
Pegboard performance
optic cortex to motor cortex
Trail making A & B
(eye-hand coordination)
Fingertip number writing
parietal lobe, sensory area of pre-Rolandic fissure
Profile of mood states
limbic system for emotional memory
The effects of ethyl alcohol are familiar to most people so I will start with this best studied of mind altering solvents. Measurements of alcoholic patients in the mid-twentieth century at New York’s Bellevue Hospital helped David Wechsler formulate his adult intelligence scale, 11 tests that measure attention, problem solving, concept juggling and memory including vocabulary.5 Many other tests were devised to estimate intelligence, how the mind works as defined by AR Luria and others.6 Ward Halstead assembled and created function tests to measure the effects of traumatic damage to the brain by wartime missiles and by neurosurgery, prefrontal lobotomy.7 Application of these tests, by Reitan,8 helped differentiate the organic brain disorders from schizophrenia and other mental illnesses. Thus the starting place for testing became brain diseases recognized by the neurologist using simple bedside qualitative tests. The tests were not used to detect impairment before it was clinically recognized. The first steps were Benjamin Franklin’s recollections of his own brain poisoning by lead while he was a printer and Lewis Carroll’s mad hatter, from mercury used in felting beaver hair. The next steps were taken in Nordic countries in the 1960’s.9 Carbon disulfide was the first solvent studied and had adverse effects observed by Delpech in 1863. Neuropsychiatric abnormalities were described 13 years later by Eulenberg in workers in the rubber and viscose rayon industries.2 A Finnish psychologist, Helen Hanninen tested 100 carbon disulfide exposed workers in 1970, 50 were poisoned, 50 exposed and compared them to 50 unexposed.10 She found intelligence, tasks of attention, motor skill vigilance and memory were impaired in clinically poisoned and exposed men compared to unexposed. Digit symbol substitution from the Wechsler’s scale5 showed the most effect of exposure. Additional studies of spray painters in the 1970’s and compared to computer augmented tomography (CT) scans of the brain and function tests. Symptomatic painters after 20 years or more of exposure had brain atrophy associated with impairment.11-14 The key to progress in this field was sensitive tests to measure brain function, Table 20.9.3. Fortunately, the Finnish, Danish and Swedish occupational-environmental health centers units included cooperative neurologists, neurophysiologists and psychologists who did not defend disciplines to limit activities. The obvious reality that the nervous system regulates and controls many essential functions helped select measurements to assess vision, hearing, vibration, odor perception, balance, reaction time including automatic responses that are measured as blink reflex latency,15,16 heart rate variation17 and peripheral nerve conduction and Hoffmann’s (H) reflex.18 Tests must be sensitive and reliable, easily understood and economical of time, taking 3 to 4 hours with rest periods.
1408
Kaye H. Kilburn
Sensitivity’s main dimensions are time and mapping. For example, balance is measured using the classic Romberg stance (1850) standing feet together with eyes ahead open and then closed and using a force (displacement) platform or even simpler the position of the head from a sound emitter secured to a headband and recorded by two microphones to inscribe the distance swayed and the speed, in centimeters per second.19 From three performances for 20 seconds with the eyes closed alternating with the eyes open the minimal speed of sway is selected. The inscribed path, the map, may provide more information but how to interpret this is unclear. Eye-hand choice reaction time is tested as the speed to cancel by tapping a keypad, a 4 inch letter that appears on the screen of a laptop computer.20 Twenty trials repeated twice and the median time of last 7 trials in each run is recorded. Simple, same letter, reaction time takes 1/4 of a second, 250 ms while choice between 2 letters takes twice as long, 500 ms. Many tests are faster in women, most deteriorate with aging after 25 years and for people with more years of educational attainment scores are higher.21 Vision is measured by mapping for color perception which is a central retinal cone function. This consists of placing 15 pale colors in a spectral array, the Lanthony desaturated hue test. Retinal rod function which is light perception was mapped for the central 30o of each visual field at 80 points using an automated perimeter recording to a laptop computer.22 This standardized and speeded up the fields that had been done by the tangent screen and a skilled operator for 100 years. It was logical to consider the 12 nerves of the head, cranial nerves as the scaffold for organizing tests and for reviewing brain functions that are adversely affected by chemicals. Smell (Nerve I, olfactory), is tested by recognition of familiar odors and of threshold concentration for detecting them. Smell disorders include loss and disturbed perception. Nerve II, the optic was described above. Nerves III, IV and VI move the eyes and rarely show effects of chemicals. An exception is the optokinetic effects of styrene. In contrast the faces sensory nerve, the trigeminal, number V and motor nerve, the facial, number VII are needed to blink and are tested by blink reflex latency which is measured electromyographically in milliseconds (10 to 15 ms) after stimulation by a tap, that is mechanical or an electrical impulse. Blink is slowed by exposures to chlorinated solvents like trichloroethylene (TCE), by chlorine and by arsenic. Nerve VIII has hearing and vestibular (balance) divisions which are tested by audiometry and by sway speed for balance. Nerve IX, the glossopharyngeal innervates the throat and is needed for the gag reflex and baroreceptor. Nerve X, the vagus X is evaluated by recording variations of heart rate with breathing. Nerves XI, spinal accessory is tested as strength of neck muscles and XII tongue’s hypoglossal nerve by speech. Using these tests implies comparing scores observed to a standard, an expected value. Ideally that would be to the same subject which is rarely possible, although it works for before and after exposures of workers. The next best comparisons are to suitable unexposed normal subjects who can be called controls.21 We developed over several years a national sample of unexposed people, tested their performance and calculated expected values using prediction equations with coefficients for age, sex, education and other factors such as height and weight that affected some tests. Thus individual observed values for each subject are compared to predicted values (observed/predicted x100) equal percent predicted. Frequently, we needed to be sure that comparison groups of apparently unexposed control people were normal because adverse effects are widespread. Next from the standard deviations
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
1409
of the mean, each tests confidence intervals were developed that included 95% of values, excluded as abnormal approximately 5% of unexposed subjects on each test that defined abnormal precisely. For these tests they were values outside the mean plus 1.5 x standard deviations (sd) that defined normal. The next concern what was the best summary for each subject. The number of abnormalities was best adjusting balance and vision above other tests and given grip strength, blink reflex and color discrimination 0.5 for right and left sides of the body. The attributes of plausible association leading to attribution of effect include temporal order, strength of association, exposure intensity and duration, specificity, consistency of findings and coherence and plausibility.1 As noted earlier the fact of exposure or suspected exposure may be the only certainty about exposure so its plausibility is important based on chemical properties, experiments and studies of workers. Consistency with results of occupational exposures and animal experiments is helpful. Koch’s 4 postulates developed to judge causation of infectious agents (1, organism present in every case; 2, grown in pure culture; 3, produces the disease when inoculated and 4, recovery and growth in pure culture) are usually inapplicable. This reality is discomforting to some interpreters of the new observations. The next section reviews the neurobehavioral affects of solvents found in the environment in the order of importance.2,4 We begin with trichloroethylene (TCE) and related short chain chlorinated agents.22 Next are ring compounds toluene including related xylene and styrene with comments on creosols or phenols. The chlorinated ring compounds follow: dichlorophenol and polychlorinated biphenyls and their highly neurotoxic derivatives, the dibenzofurans. Other straight chain solvents leading off with n-hexane move through white solvent (paint thinner) and solvent mixtures. Before studies of effects of TCE on many brain functions came the measurement of blink reflex latency in 22 people exposed at home to solvents rich in TCE at Woburn, MA. They showed significant delay of blink but no other functions were measured.16 In France about this time workers exposed to TCE had similar delays of blink.24 Earlier experimental exposure of 12 subjects to TCE at 1,000 parts per million (ppm) for 2 hours in a chamber had produced rapid flickering eye movements when following figures on a rotating drum (optokinetic nystagmus), a lowered fusion limit.22 Thus TCE induced dysfunction of several cranial nerves VI (with III, IV) and V and VII. Nystagmus normalized after a washout and recovery time. Blink is the easier and quicker measurement.16 A community within Tucson, AZ of over 10,000 people who depended on well-water for drinking and bathing had developed many complaints and had excesses of birth defects and cancers that associated with TCE in their water. The source was metal cleaning that included stripping off protective plastic coatings, from demothballing aircraft stored on the desert with TCE. This had dumped vast quantities of TCE on the porous, desert floor that drained into the shallow Santa Cruz River aquifer. Testing of 544 people from this water exposure zone showed increased blink reflex latency, impaired balance, slowed simple and choice reaction times, reduced recall, poor color discrimination, and impaired problem solving in making designs with blocks, digit symbol substitution and Culture Fair (consisting of 4 subtests: selection of designs for serial order, for difference, for pattern completion and refining defined relationships).25,26 Also peg placement in a slotted board and trail making A (connecting 25 numbers in ascending sequence and B connecting numbers alternating with letters).27 TCE concentrations at the well heads and distribution pipes to homes were
1410
Kaye H. Kilburn
measured to calculate with duration of exposure, lifetime peak levels, lifetime averages, and cumulative exposure. Possible relationships of neurobehavioral test scores to these surrogates for dose were searched by regression analysis. No relationships were found for these dose surrogates which was disappointing. TCE dominated the mixture of chlorinated aliphatic solvents in air and water of northeast Phoenix around two Motorola microchip manufacturing plants that began production in 1957. Neighbor’s complaints of adverse health effects started amelioration effects in 1983. An underwater TCE solvent plume spread west and south of the plants in the Salt River aquifer. Test wells showed concentrations of TCE from 50 ppm to 1.4%. Also air dispersal was important for direct exposure as TCE escaped into the air from the plant. It drained into dry wells, sewers and into a canal running northwest through the neighborhood. In 1993, 236 exposed adults were compared to 161 unexposed ones from a town 80 km northwest across the mountains at a higher elevation. The exposed group showed delayed blink reflex, faster sway speed, slowed reaction time, impaired color discrimination and reduced cognitive function and perceptual motor speed and reduced recall. Airway obstruction was shown by pulmonary function testing. Adverse mood state scores and frequencies of 32 symptoms were also increased.4,28 Remedial efforts directed at dumping and ground water had not reduced the effect suggesting either these were ineffective or impairment was permanent and had developed after 1983. Additional groups of subjects on plume but not in the lawsuit were not different from clients so there was no client bias. Phoenix residents off the plume had only abnormal slowing of blink interpreted as due to TCE and abnormal airway obstruction compared to the unexposed population of Wickenburg, AZ. Airway obstruction was attributed to Phoenix wide air pollution. Proximity within 1.6 km seemed to increase impairment. In 1998 retesting of 26 people from original groups showed improvement with faster blink reflexes but worse airway obstruction that had persisted (ref. 4 and unpublished). The improvement to normal in blink paralleled that seen in chlorine exposed people 3 years after exposure and first evaluations that were abnormal.4 The perceptual motor tests, trail making A and B and peg placement were improved, as were cognitive function measured as Culture Fair and verbal recall. We deduced that diminished TCE releases from Motorola after 1993 allowed recovery of cranial nerves V and VII so blink latency decreased, accompanied by some improvement in vigilance and tracking for the better scores. A possible reversal of effect is so important that these observations should be verified in other groups. Workers welding and grinding on jet engines in a repair shop were an unusual way to focus attention on the Gerber-Wellington aquifer in Oklahoma.28 Our attention was temporarily on metals in alloys but when testing of 154 workers showed impaired balance, slowed choice reaction times and impaired color discrimination compared to 112 unexposed subjects, the priority became effects on the brain. These worker’s cognitive function, perceptual motor and recall were all abnormal using the same tests as in Phoenix. We probed for their exposures after observing these effects and found that these workers had used TCE, trichloroethane, methanol and Freon FC-113 in metal cleaning. The 112 control subjects had not worked with solvents or TCE, but many people, both workers and control groups, lived on and drew well water from the Gerber-Wellington aquifer that is contaminated with TCE. Mapping the blink reflex latencies in the control group showed the people with normal blink lived outside the aquifer. The aquifer was TCE contaminated (national priority
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
1411
list 1990) Thus here we had an example of a probable environmental exposure to TCE with a superimposed occupational one.4 TCE leakage had produced these same effects in Joplin, MO neighbors of a company manufacturing ball bearings and cleaning with this solvent. The companies decision to clean and reuse TCE rather than dump it ameliorated the effects.4 In San Gabriel and San Fernando Valleys in California similar observations confirm adverse effects from TCE contamination of groundwater and air. Clearly the observations were replicated and each time TCE was associated with reasonable timing and proximity. More than half of the federal superfund sites in the US are contaminated with TCE suggesting 800 potential replications.29 Experience with several patients have shown me that the effects of dichloroethylene and of 1,1,1-trichloroethane are indistinguishable from those of TCE. Toluene is the most toxic and best studied of the aromatic ring compound solvents. Both acute and chronic effects were observed by 1961 from inhaling “huffing” toluene30,31 or lacquer thinner,32 especially in children sniffing airplane glue.30 Chronic impairment was shown shoemakers33 and rotogravure printing workers34 using neurobehavioral testing. Toluene exposed experimental animals, mainly rats and mice showed enhanced motor activity, abnormal movements, altered sleep patterns and electroencephalograph (EEG) changes from an integrative brain loop, the hippocampus.2 Occupational exposures produced memory disturbances, poorer performance on block design assembly, embedded figures, visual memory and eye-hand coordination. CT scans showed some generalized brain swelling34 that correlated with impaired psychological functions.35 Women working in electronic assembly had environmental air levels of toluene of 88 ppm compared to 13 ppm for controls and comparable differences in blood levels. These workers were less apt at placing pegs in a grooved board, at trail making, digit symbol, visual retention and reproduction and verbal memory.36 They were tested during the day after being away from exposure for at least 16 hours. Protracted sniffing of solvents alone or in glue has produced intention tremor and titubating gait30 consistent with cerebellar degeneration which continued after 5 years with ataxia, EEG slowing and cerebral atrophy.31 Polyneuropathy was observed in 2 glue sniffers in Japan whose exposures were to n-hexane and toluene.37 Many such descriptions2 outweigh one epidemiological study that found no differences in performance when comparing 12 glue sniffing boys, ages 11 to 15, mean 13.8 years and 21 controls, ages 11 to 15, mean 12.6 years. Four non-standard tests and the Benton visual retention and design reproduction test were used but the exposed group was 1.2 years older and should have outperformed younger controls whose skills were less developed.38 Some published data are difficult to interpret. For example, 26 men were exposed in tanks and holds of two merchant vessels being painted (solvents) and sprayed with malathion 20% and pyrethrin 1.5%, with piperonyl butoxide in toluene. They showed losses of concentration, unawareness of danger and unconsciousness at toluene levels estimated as 10,000 to 12,000 ppm and up to 30,000 ppm below waist level.39 Additive effects of the neurotoxic insecticides were not discussed. Effects of toluene in 52 men and paint solvents in 44 men were contrasted with unexposed men. Painters had impaired reading scores, trails B, visual search, block design, grooved pegboard, simple reaction time and verbal memory.40 Toluene exposed men had only abnormal reading scores reduced significantly, although scores on all tests were lower.
1412
Kaye H. Kilburn
Levels of toluene were less than 200 ppm for 4 years prior to this study, although above 500 ppm earlier. The axiom that environmental exposures are “never” to pure chemicals is matched by another that the mixtures are frequently so complex as to defy description. The few observations suggest that effects seen from mixtures may be due to one or two specific neurotoxic agents. Judgment must be exercised to curb bias and accept the most plausible attribution as the above studies illustrate. Studies of a population exposed in Louisiana to toluene rich solvents and other chemicals distilled from a site for 17 years were contrasted to unexposed people living 55 km to the east.41 The Combustion site accepted 9 million gallons of used motor oil in 1975-1976 and 3 to 4 million gallons from 177 to 1983. Tons of liquid chemical waste from over 100 chemical factories was consigned to this site including toluene, xylene, styrene and benzene, many chlorinated aliphatics solvents like TCE and chlorinated aromatics including PCBs and dibenzofurans. Lead, cadmium, mercury and other metals were present in samples of sludge in ponds after the site closed in 1983 but they were rich in toluene, benzene and other aromatics. Modeling based on toluene and benzene and using standard Environmental Protection Agency assumptions and a windrose showed symmetrical spread eastward. Excesses of leukemia in school children, cancers and neurobehavioral symptoms in the about 5,000 neighbors of the site led to neurobehavioral testing for impairment in 131 adult subjects within 2 km of the site and 66 adult controls from voter registration rolls of a town 50 km east. The exposed group matched controls for age but were 1.4 years less educated. There were adverse effects from exposures while living within 2 km from the site for 4 to 17 years.41 that were shown by slowed simple and choice reaction times and abnormal sway speeds. Cognitive function in Culture Fair and block design was decreased and pegboard and trail making A and B were diminished, as was recall of stories. Profile of mood states (POMS) scores were 2.5 fold increased with low vigor and high depression, tension, confusion, anger and fatigue. Thirty of 32 symptoms inquiring about chest complaints, irritation, nausea and appetite associated, balance, mood, sleep, memory and limbic brain were significantly more frequent in exposed people and the other two were rare in both groups. When differences were adjusted for age, color discrimination and similarities became abnormal were added and trail making A became normal. The second study was designed to answer how large an area-population was affected, was direction important and were abnormalities related to the duration of exposure.42 I examined 408 subjects selected to fill 3 distances outward to 1.6, 3.2 and 4.8 km in 8 compass octants, thus 24 sectors. The same tests were given by the same staff and results replicated the earlier study. Regression analysis of each test against distance showed no significant coefficient and comparison of inner and other sectors found no differences, thus there was no evidence of a diminished effect from distance. There were no effects of direction. A possible lessening of effect for durations of exposure of less the 3 years was seen only for peg placement and trail making B scores. Distance, direction and duration as surrogates for exposure did not influence impairment as measured. We concluded that the periphery of effect was beyond 4.8 km meaning a health impact area larger than 75 km2. There was no gradient of effect from the distilling plant outward suggesting airborne spread and mixing had produced even dosing from a large “cloud”.4 Peoples migration inside the exposure zone did not influence effects. Bias of examiners
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
1413
was unlikely. Was the control group suitable? Their average measurements and the distribution were like three other groups in different parts of the country. The possibility of confounding exposures was considered from two sites that were beyond 4.8 km and to the south. Unfortunately, the resources were unavailable to extend testing beyond 4.8 km to find the rim of Combustion’s effect on people and detect effects of other nearby sites. Xylene is a solvent for paints, lacquers and adhesives and is a component of gasoline. In human volunteers in exposure chambers xylene at 70 ppm for 2 hours had no effect on reaction time or recall memory but levels of 100 to 400 ppm for 2 hours impaired body balance, memory span, critical flicker fusion and cause eye irritation.43-45 Alcohol and 1,1,1-trichloroethane had adverse effects on balance that show synergism with xylene.46 and increased the latencies for visual and auditory evoked potentials.47 Occupational studies have focused on psychiatric symptoms in photogravure workers who also showed headache, nausea, vomiting and dizziness.48 Only one study showed impairment for recall memory attributed to xylene but workers were also exposed to formaldehyde.49 Xylene toxicity has received less study than that of toluene, but appears considerable less which supports attributed the neurotoxicity to toluene of mixtures of xylene, benzene and toluene with straight chain hydrocarbons such as gasoline. Styrene’s major use is in reinforced fiberglass plastics in constructing boats and bathtubs and showers and in styrene-butadiene rubber.2 Small amounts are used in polystyrene foam cups and packing materials. Styrene inhalation increased locomotion activity in rats and grip strength at the highest 700 to 1,400 ppm concentrations.2 Studies of workers showed hearing loss (increased high frequency hearing thresholds at 16 kHz).50 Color discrimination is also reduced.51 Other observers found abnormal hearing and by posturography-larger sway areas and poor rotary visual suppression-inhibition or vestibulatory nystagmus.52 In 25 studies of workers2 some showed slowing of reaction time, poor performance on block design, short-term memory, EEG abnormalities and neuropathy. These relatively mild effects made me predict less than the severe impairment than observed in 4 women from a factory making styrene-fiberglass shower-bathtubs. Two sprayed styrene and the other 2 who had developed skin and airway symptoms on initial exposure did lay-up and assembly. Five weeks after her first exposure one woman became lightheaded and dizzy, felt hot and her vision blacked out. On testing reaction times were slow, sway speed was increased. Problem solving was impaired as was verbal recall and POMS scores were elevated. She left work stopping exposure. Ten days later, on a trip to the mountains 4,000 feet above sea level she collapsed and became unconscious. Retesting showed constricted visual fields and worse performance of the above tests. Testing on the second woman who had developed asthma showed multiple blind spots in her visual fields, diminished problem solving ability, grip strength, excessive fingertip writing errors and failure to recall stories after 30 minutes. A third woman also had asthma and severe airway obstruction showed abnormal balance with eyes open and closed, diminished hearing, bilaterally constricted visual fields and decreased vibration sense. The fourth woman had a skin rash and red welts that had kept her away from direct contact with epoxy and styrene. She had abnormal color discrimination, decreased vibration sensation, a blind spot in the retina of the left eye and decreased recall of stories. She was the least impaired although her POMS score and symptom frequencies were increased.
1414
Kaye H. Kilburn
Air sampling was not permitted, concentrations of styrene are unknown and contributions of other chemicals to this exposure cannot be excluded. However, exposures to formaldehyde and phenol are unlikely as these workers did not “lay-up” fiberglass resin. Inhalation of sprayed styrene is the most attribution for the neurobehavioral impairments. The impairment exceeded that found in a review of boat building and other studies but tests were more sensitive and the styrene levels may have been higher. We encourage more neurobehavioral evaluations of styrene spraying workers using such sensitive tests. Polychlorinated biphenyls (PCBs), the ultimate (poly)chlorinated solvents are 2 membered ring compounds that when heated to 270o produce dibenzofurans (DBFs) that are 1,000 or more times as neurotoxic.53 Initial evaluations were of a few PCB exposed individuals and 14 firemen exposed to DBFs who showed severe impairment measured after a medical schools power plant transformers cooked and exploded. Most of the firemen could not pass the physical, balance and truck driving requirements to return to duty and were retired on disability.4,54 A community study explores effects of environmental exposures. PCBs were used as pump lubricants in natural gas pipelines running north from Texas and Louisiana from 1950 to the middle 1970’s. One pumping station was at Lobelville, TN and at least 16 other US communities had them.4 Ninety-eight adult village dwellers were compared to 58 unexposed subjects from 80 km east or 35 km north. The exposed people were the most abnormal group I have studied. They had abnormal simple and choice reaction times, balance, hearing, grip strength and the visual function of color discrimination, contract sensitivity and visual field performance. The cognitive functions of Culture Fair, digit symbol were abnormal as were vocabulary, information, picture completion and similarities. Story recall was diminished and peg placement and trail making A and B and fingertip number writing errors were decreased. Other possible associations were ruled out and there were no other causes of impairment. This exposure had caused the most severe neurobehavioral impairment for these people that I have observed.4 It exceeded that from distilling chemical waste rich in toluene, from TCE and from other solvents. n-hexane by inhalational or through the skin causes peripheral nerves to die-back. Glue sniffing exposure frequently combines n-hexane and toluene. Twenty-five percent of workers using glue in shoes and leather goods with n-hexane, 40 to 99.5% had symptomatic polyneuropathy, slowed nerve conduction and neurological signs.55 Abnormal findings increased with age and durations of exposure and were accompanied by lower limb weakness and pain, abnormal sensations (parenthesis) in the hands and muscles spasm. In another shoe plant exposure group upper extremity nerve conduction was slowed, frequently after 5 years of exposure.56 Sensormotor distal neuropathy characterized 98 of 654 workers in the Italian shoe industry, 47 had decreased motor conduction velocity with headache, insomnia, nausea and vomiting irritability and epigastric pain.57 Most workers improved when removed from exposure.56 In Japan beginning in 1964 several studies found polyneuropathy in polyethylene laminating printers58 and makers of sandals and slippers.59 A major metabolite of n-hexane and of methyl butyl ketone is 2,5-hexanedione that is more neurotoxic than these precursors causing swelling of nerve axons and accumulations of neurofilaments in mid-portions of peripheral nerves.60 Methyl ethyl ketone studied in workers lengthened choice reaction time and motor nerve conduction and decreased vibration sensation signs of neuropathy. These effects were also seen in glue sniffers.2
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
1415
The focus on polyneuropathy in 14 studies2 has usurped studies of central nervous system (CNS) functions except for one showing increased latencies of visual and auditory evoked potentials.61 Neurophysiological and psychological assessments62 show narcotic effects that match those in animals.2 Until restricted from foods n-hexane was used to extract oil from soybean meal and exposed US workers. They had headache, dysesthesia, insomnia, somnolence and memory loss but testing for appraise brain damage was not done. Gasoline is a mixture of aliphatic straight and branched chains and aromatic hydrocarbons with toxicity attributable to toluene, xylene and perhaps hexane and additives including methyl ter butyl ether and tri-orthocresyl phosphate.63 Effects of the lungs of inhaled solvents simplify to the consideration of agents affecting airway cells that include n-hexane and PCBs. Both cause proliferation and transformation of distal airway lining cells to produce mucus and obstruct airways64-66 and cause inflammatory cells to pour into the lungs distal alveolar spaces interfering with for gas exchange. Cardiac effects of solvents are of three types A, alterations in rhythm B, cardiomyopathy and C, hypertension directly and via renal changes as in interstitial nephritis, glomerulonephritis and Goodpasture’s syndrome.67 Alterations in heart rhythm have been attributed to anesthesia with TCE and were serious, especially when administered in soda lime CO2 absorbing anesthesia machines to stop the use of TCE for anesthesia in the 1960’s.4 Rhythm disturbances have also been observed in some groups of workers exposed to TCE. Knowing this we did electrocardiograms (ECG’s) on the Tucson TCE exposed population and found no arrhythmias. A loss of respiratory variation in heart rate has been associated with exposure to organic solvents, including carbon disulfide, acrylamide and alcohol but not toluene and with diabetes mellitus and syndromes of autonomic nervous system dysfunction.17,68,69 Later freons, volatile chlorofluorocarbons were associated with arrhythmias and withdrawn from use to propel therapeutic aerosols used for asthma.70 Cardiomyopathy, heart muscle dysfunction and enlargement have been associated with alcohol ingestion. Two epidemics were ascribed to cobalt used to color beer.9 But in the most common cardiac muscle disorder from alcohol, cobalt is not incriminated. Hypertension has been associated with solvent exposure in workers, an association that needs further study. Associations with hypertension were absent in the authors studies of TCE, toluene rich waste and PCBs discussed earlier. Workers who used methylene chloride in making acetate film had sleepiness and fatigue and decreased digit symbol substitution scores and lengthened reaction time.71 Use of methylene chloride in closed spaces has been fatal with brain edema, elevated blood levels of carboxyhemoglobin and caused temporary right hemispheric paralysis and/or unconsciousness.2 Chronic exposure has been associated with dementia, headache, dizziness and disturbed gait.72 Chloromethane exposures from foam production caused tremor and decreased attention and ability to do arithmetic.73 Environmental exposures from leaks in refrigerating systems2 caused deaths, convulsions, myoclonus and personality changes. One fishing boat exposure of 15 men left profound neurological residuals, fatigue, depression and alcohol intolerance.2 The effects resemble those of methyl bromide poisoning. Methanol has profound and specific toxic effects on the optic nerve and vision causing central blind spots and ingestion of methanol for the intoxicating effects of ethanol has
1416
Kaye H. Kilburn
caused blindness.74 Its metabolism to formic acid suggests the possibility that of other central nervous system effects. White spirit is a mixture of straight and branched chain paraffins, naphthalenes and alkyl aromatic hydrocarbons is used widely as a paint solvent. Ten studies of painters, mostly in Nordic countries, have shown increased neurobehavioral symptoms and several showed decreased performance on psychological tests.11-14 Longitudinal studies showed an almost doubled risk for neuropsychiatric disability pension in painters compared to construction workers. Several such studies support the concept of neuropsychiatric impairment and disability linked to the painting trade in many countries.75-78 Several women in my consulting practice had profound neurobehavioral impairment after entering their homes during spray painting including unconsciousness which suggest there may be a considerable problem from environmental exposures. Many industrial painters exposures are to solvent mixtures. Those painting airplanes where dust and hence fume exposure is limited by strict cleanliness, which means good air hygiene for the workers, have little trouble compared to symptoms, impairment, disability and brain atrophy with dementia in car and refrigerator painters. These groups supplied the clear evidence of solvent effects in workers in Nordic countries that established how to assess human subject’s neurobehavioral status and detect impairment that were discussed early in this chapter. Many cross sectional studies showed adverse effects, excessive neuropsychiatric symptoms and several longitudinal studies show greatly increased likelihood of receiving a pension for neuropsychiatric disability.2,78-80 Chemical companies fight the concept that chemicals damage human subjects. They are more combative and better defended than are bacteria and other infectious agents. In the past 25 years companies learned from asbestos litigation, the bankruptcy of Johns Manville Company and the banning of asbestos to contest observations and their scientific basis and frequently hire scientists to support their position of null effects-not harm and sponsor environmental meeting and advertise their concern and sense of responsibility. They avoid or shift responsibility for damage to the victim or community and the social security system. The necessary banning of PCBs and chlordane enforced their strategy of “controversy” even about incontrovertible facts. Perhaps, they count on having the 50 years that tobacco companies enjoyed before having to accept responsibility for adverse effects of tobacco smoking. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
K Kilburn, and R Warshaw, Epidemiology of adverse health effect from environmental chemicals, Princeton Scientific Publishing Co., 1995, pp 33-53. P Arlien-Soberg, Solvent Neurotoxicity, CRC Press, Boca Raton, 1992. J Angerer, Scan. J. Work. Environ. Hlth., 11 (Suppl 1),45(1985). K Kilburn, Chemical Brain Injury, John Wiley & Sons, New York, 1998. D Wechsler, Adult Intelligence Scale Manual, (revised), The Psychological Corporation, New York, 1981. AR Luria, Higher Cortical Function in Man, Travistock, London, 1966. W Halstead, Brain and Intelligence, The University of Chicago Press, Chicago, 1947. R Reitan, Percept. Motor Skills, 8, 271(1958). D Hunter, Diseases of Occupations, 4th Edition, Little Brown, Boston, 1969. H Hanninen, Brit. J. Indust. Med., 28, 374(1971). B Knave, B Kolmodin-Hedman, H Persson, and J Goldberg, Work Environ, Health, 11, 49(1974). S Elofsson, F Gamberale, T Hindmarsh, et al., Scand. J. Work Environ. Health, 6, 239(1980). O Axelson, M Hane, and C Hogstedt, Scand. J. Work Environ. Health, 2, 14(1976). P Arlien-Soberg, P Bruhm, C Gyldensted, and B Melgaard, Acta Neurol. Scand., 60, 149(1979). R Feldman, J Chirico-Post, and S Proctor, Arch. Environ. Health, 43, 143(1988).
20.9 Toxicity of environmental solvent exposure for brain, lung and heart
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66
1417
K Kilburn, J Thornton, and B Hanscom, Electromyograph Clin. Neurophysiol., 38, 25(1998). E Matikainen, and J Juntunen, Neurobehavioral Methods in Occupational and Environmental Health. World Health Organization, Copenhagen, 1985, pp. 57-60. S Oh, Clinical Electromyography; nerve conduction studies, 2nd Edition, Wilkins and Wilkins, Baltimore, 1993, pp.89. K Kilburn, and R Warshaw, Occup. Environ. Med., 51, 381(1994). J Miller, G Cohen, R Warshaw, J Thornton, and K Kilburn, Am. J. Indust. Med., 15, 687(1989). K Kilburn, J Thornton, and B Hanscom, Arch. Environ. Health. 53, 257(1998). B Kylin, K Axell, H Samuel, and A Lindborg, Arch. Environ. Health, 15, 48(1967). K Kilburn, Neurotoxicology, 21, (2000). L Barrett, S Garrel, V Danel, and J Debru, Arch. Environ. Health, 42, 297(1987). R Cattell, S Feingold, and S Sarason, J. Educational Psych., 32, 81(1941). R Cattell, J. Consulting Psych., 15, 154(1951). K Kilburn, and R Warshaw, J Toxicol. Environ. Health, 39, 483(1993). K Kilburn, Environ. Res., 80, 244(1999). National Research Council, Public Health and Hazardous Waste, Environ. Epid., National Academy Press, Washington DC, 1991. D Grabski, Am. J. Psychiatry, 118, 461(1961). J Knox, and J Nelson, New Eng. J. Med., 275, 1494(1966). L Prockop, M Alt, and J Tison, JAMA, 229, 1083(1974). T Matsushita, Y Arimatsu, A Ueda, K Satoh, and S Nomura, Ind. Health, 13, 115(1975). J Juntunen, E Matikaninen, M Antti-Poika, H Suouanta, and M Valle, Acta Neurol. Scand., 72, 512(1985). H Hanninen, M Antti-Poika, and P Savolainen, Int. Arch. Occup. Environ. Health, 59, 475(1987). S Foo, J Jeyaratnam and D Koh, Brit. J. Indust. Med., 47, 480(1990). T Shirabe, T Tsuda, A Terao, and S Araki, J. Neurol. Sci., 21, 101(1974). J Dodds, and S Santostefano, J. Pediatr., 64, 565(1964). E Longley, A Jones, R Welch, and Lomaev, Arch. Environ. Hlth., 14, 481(1967). N Cherry, H Hutchins, T Pace, and H Waldron, Brit. J. Indust. Med., 42, 291(1985). K Kilburn, and R Warshaw, Neurotoxicol. Terato., 17, 89(1995). K Kilburn, Environ. Res., 81, 92(1999). C Carpenter, E Kinkead, D Geary Jr, L Sullivan, and J King, Toxicol. Appl. Pharmacol., 33, 543(1976). K Savolainen, V Riihimaki, and M Linnoila, Int. Arch. Occup. Environ. Hlth., 44, 201(1979). K Savolainen, V Riihimaki, Omuona, and J Kekoni, A Laine, Arch. Toxicol., Suppl., 5, 96(1982). K Savolainen, V Riihimaki, O Muona, J Kekoni, R Luukkonen, and A Laine, Acta Pharmacol. Toxicol., 57, 67(1985). A Seppalainen, K Savolainen, and T Kovala, Electroencephalo. Clin. Neurophysiol., 51, 148(1981). D Klaucke, M Johansen, and R Vogt, Am. J. Induct. Med., 3, 173(1982). K Kilburn, R Warshaw, and J Thornton, Arch. Environ. Health, 42, 117(1987). H Muijser, E Hoogendijk, and J Hooisma, Toxicol., 49, 331(1988). F Gobba, and A Cavalleri, In: Butadiene and Styrene: Assessment of Health Hazards, IARC Scientific Publications No. 127, Lyon, France, 1993. C Moller, L Odkvist, B Larsby, R Tham, T Ledin, and L Bergholtz, Scand. J. Work. Environ. Hlth., 16, 189(1990). O Hutzinger, G Choudhry, B Chittim, and L Johnston, Environ. Health Perspect., 60, 3 (1985). K Kilburn, R Warshaw, and M Shields, Arch. Environ. Health, 44, 345(1989). E Buiatti, S Cecchini, O Ronchi, P Dolara, and G Bulgarelli, Brit. J. Ind. Med., 35, 168(1978). I Aiello, G Rosati, G Serra, and M Manca, Acta Neurol., 35, 285(1980). S Passero, N Battistini, R Cioni, F Giannini, C Paradiso, F Battista, F Carboncini, and E Sartorelli, Ital. J. Neurol. Sci., 4, 463(1983). S Yamada, Jpn. J. Ind. Health, 6, 192(1964). Y Yamamura, Folia Psychiatr. Neurol. Jpn., 23, 45(1969). P Spencer, and H Schaumburg, J. Neurol. Neurosurg. Psychiatr., 38, 771(1975). A Seppalainen, C Raitta, and M Huuskonen, Electroencephalogr. Clin. Neurophysiol., 47, 492(1979). S Sanagi, Y Seki, K Sugimoto, and M Hirata, Int. Arch. Occup. Environ. Health, 47, 69(1980). T Burbacher, Environ. Hlth. Perspect., 101(Suppl 6),133 (1993). R Warshaw, A Fischbein, J Thornton, A Miller, and I Selikoff, Ann. NY Acad. Sci., 320, 277(1979). P Houch, D Nebel, and S Milham Jr, Am. J. Indust. Med., 22, 109(1992). N Shigematsu, S Ishimaru, and R Saito, Environ. Res., 16, 92(1978).
1418
67 68 69 70 71 72 73 74 75 76 77 78 79 80
Kaye H. Kilburn
N Benowitz, Cardiotoxicity in the workplace, Occup. Med., Vol 7, No. 3, Hanley & Belfus, Inc., Philadelphia, 1992. Y Kuriowa, Y Shimada, and Y Toyokura, Neurology, 33, 463-467(1983). K Murata, S Araki, K Yokoyama, K Yamashita, F Okajima, and K Nakaaki, NeuroToxicol., 15, 867-876(1994). W Harris, Arch. Intern. Med., 131, 1621(1973). N Cherry, H Venables, H Waldron, and G Wells, Brit. J. Ind. Med., 38, 351(1981). B Friedlander, T Hearne, and S Hall, J. Occup. Med., 20, 657(1978). J Repko, P Jones, L Garcia, E Schneider, E Roseman, and C Corum, Behavioral and neurological effects of methyl chloride, CDC-99-74-20, National Institute for Occupational Safety and Health, 1976. W Grant, Toxicology of the Eye, Charles C Thomas Publisher, Springfield, Illinois, 1974. S Mikkelsen, Scand. J. Soc. Med., (Suppl. 16), 34,(1980). H Rasmussen, J Olsen, and J Lauritsen, J. Occup. Med., 27, 561(1985). K Lindstrom, Scand. J. Work Environ. Health, 7 (Suppl. 4), 48(1981). J Olsen, and S Sabroe, Scand. J. Soc. Med., 44 (Suppl. 16), 44,(1980). K Lindstrom, H Riihimaki, and K Hanninen, Scand. J. Work Environ. Health, 10, 321(1984). T Riise, and B Moen, Acta Neurol. Scand., 77 (Suppl. 116), 104(1988). Note: New general source for neurotoxicity for solvents. R Feldman, Occupational and Environmental Neurotoxicology, Leppincott-Raven, Hagerstown, Maryland, 1999.
21
Substitution of Solvents by Safer Products and Processes 21.1 SUPERCRITICAL SOLVENTS Aydin K. Sunol and Sermin G. Sunol Department of Chemical Engineering University of South Florida, Tampa, FL, USA
21.1.1 INTRODUCTION Significant and steady inroads towards wider and more effective utilization of supercritical fluids have been made over the last two decades, especially for high value added differenti-
Figure 21.1.1. Pressure-temperature and pressure-density behavior of matter.
1420
Aydin K. Sunol and Sermin G. Sunol
ated products. Furthermore, a new wave of second-generation supercritical technologies started to emerge, bringing forth new roles for dense gases. The motivation of this chapter is to assess the current status of the technology in an effort to extricate the challenges both in the current practice of processing with supercritical fluids and potential areas into which we have as yet to venture. The foundations including transport phenomena, reactions and thermodynamics as well as generic processing principles will be highlighted. “Supercritical” refer to the state of the matter where the temperature and pressure of a single component fluid is above the critical point at which the phase boundaries diminish. A portfolio of chemical and physical operations carried out in the vicinity of this region defines Supercritical Fluid Technology (SFT). The pressure-temperature-volume (PVT) behavior of a substance can be best depicted by pressure-temperature and pressure-density (volume) projections, as shown in Figure 21.1.1. Pressure-temperature diagram identifies the supercritical fluid region, alternative separation techniques that involve phase transition including the associated phase boundaries, and the fact that an isotherm below critical (AB) involves phase transition while one above (A’B’) does not. Pressure density projections illustrate tunability of the solvent density at supercritical conditions (A’B’) and again the continuity of the isotherm that does not encounter any phase transition. The aforementioned behavior is for a pure component, solvent. The critical properties of various organic and inorganic substances are shown in Table 21.1.1. Table 21.1.1. The critical properties of solvents Solvents
Critical temperature, °C
Critical pressure, atm
Critical conditions for various inorganic supercritical solvents Ammonia
132.5
112.5
Carbon dioxide
31.0
72.9
Carbonyl sulfide
104.8
65.0
Nitric oxide
-93.0
64.0
Nitrous oxide
36.5
71.7
Chlorotrifluoro silane
34.5
34.2
Silane
-3.46
47.8
Xenon
16.6
58.0
Water
374.1
218.3
Critical conditions for various organic supercritical solvents Acetone
235.5
47.0
Ethane
32.3
48.2
Ethanol
243.0
63.0
Ethylene
9.3
49.7
Propane
96.7
41.9
Propylene
91.9
45.6
21.1 Supercritical solvents
Solvents
1421
Critical temperature, °C
Critical pressure, atm
Cyclohexane
280.3
40.2
Isopropanol
235.2
47.0
Benzene
289.0
48.3
Toluene
318.6
40.6
p-Xylene
343.1
34.7
Chlorofluoromethane
28.9
38.7
Trichlorofluoromethane
198.1
43.5
Figure 21.1.2. Annual number of publications related to supercritical fluids.
Figure 21.1.3. Annual number of dissertation related to supercritical fluids.
1422
Aydin K. Sunol and Sermin G. Sunol
Figure 21.1.4. Annual number of patents related to supercritical fluids.
21.1.1.1 A promising path to green chemistry One of significant paradigm shifts in chemical processing for the new millennium is the increased use of environmentally benign technology. The effectiveness of green solvents such as supercritical water and supercritical carbon dioxide for carrying out reactions, difficult separations, and materials processing is naturally very promising. The effectiveness of the supercritical solvents is related to their state, critical temperature, and pressure. Obviously, synergy between the physical characteristics of solvent and the conditions favorable for the desired chemistry is of paramount importance for the success of the application. 21.1.1.2 Unique and tunable physico-chemical properties Supercritical fluids have the mobility of gases and the dissolving power of liquid solvents resulting in efficient penetration into porous matrices, high mass transfer rates, and high solvency. Furthermore, these properties are extremely sensitive to perturbations in temperature, pressure and composition resulting in innovative processing concepts. Furthermore, tailored products with tunable performance could be produced through synthesis and creative operating policies. Thus, over the last three decades, a spate of supercritical processes have been developed particularly for manufacture of high value added products that are superior in performance and exhibit conscious regard for a more socially responsible manufacturing practice. 21.1.1.3 Sustainable applications in many different areas Despite the higher capital charges associated with relatively high pressures, the necessity to often add a new component into the processing environment, and operational challenges at conditions foreign to most process engineers, the interest in supercritical fluids had been growing steadily since the early eighties beyond the select number of areas. We see applications in the food and beverage industries, pharmaceutical, biomedical, micro-electronic industries, textiles, forest products, petrochemicals, chemicals, environmental clean-up, syn-fuel production, polymeric materials, ceramics, auto industry, coatings and paint industry, energetic materials, and fuels. As depicted in Figures 21.1.2-4, the number of the publications, dissertations, and patents in the area are still growing from one year to next.
21.1 Supercritical solvents
1423
Figure 21.1.5 Classification of vapor-liquid phase behavior of binary systems.
21.1.2 FUNDAMENTALS 21.1.2.1 Phase behavior with supercritical solvents The first recorded observation of supercritical fluid’s ability to dissolve solids with low volatility goes back to 1879.1 Since then, better understanding of high-pressure phase equilibrium emerged partially as a result of experiments of the last century, especially those of Timmermans,2 Schneider,3,4 Franck5 on aqueous systems, McHugh6 on polymeric systems and more recently the Delft group in Holland.7,8 In classification of phase diagrams, one of the most important analytic contributions is due to Scott and Van Konynenberg.9 They demonstrated that most of the experimental observations could be described qualitatively by the van der Waals equation of state. Their analysis of critical and three phase boundaries in temperature-pressure projection of the phase behavior led to their categorization of possible fluid phase equilibrium into five classes. Rowlinson and Swinton10 later added a sixth class that occurs in some aqueous systems but is not predicted by the van der Waals equation of state. The classes described by Scott illustrate only a few of the known types of phase equilibrium and are shown in Figure 21.1.5. This classification scheme illustrates the principal lines (degrees of freedom = 1) and end points (degrees of freedom=0) that form the boundaries of pressure-temperature space of the surfaces that describe the equilibrium between two components. The types of boundaries are: 1. Solid lines (_______) are pure component vapor pressure curves. 2. The dashed lined (_ _ _ _ _) are for three phase lines. 3. The dotted lines (.........) are for critical lines. The types of points are:
1424
Aydin K. Sunol and Sermin G. Sunol
1. The filled circles are pure component critical points 2. The triangles are critical end points, lower (L) and upper (U) Type I mixtures have continuous gas-liquid critical line and exhibit complete miscibility of the liquids at all temperatures. Mixtures of substances with comparable critical properties or substances belonging to a homologous series form Type I unless the size difference between components is large. The critical locus could be convex upward with a maximum or concave down with a minimum. Examples of Type I mixtures are: Water + 1-propanol, methane + n-butane, benzene + toluene, and carbon dioxide + n-butane. Type II have systems have liquid-liquid immiscibility at lower temperatures while locus of liquid-liquid critical point (UCST) is distinct from gas liquid critical line. Examples include: water + phenol, water + tetralin, water + decalin, carbon dioxide + n-octane, and carbon dioxide + n-decane. When the mutual immiscibility of two components becomes large, the locus of liquid-liquid critical solution moves to higher temperatures and it eventually interacts with gas-liquid critical curve disrupting the gas-liquid locus. This particular class is type III and some examples include: water + n-hexane, water + benzene, carbon dioxide + n-tridecane, and carbon dioxide + water. Type IV systems have three critical curves, two of which are VLL. If the hydrocarbon mixtures differ significantly in their critical properties, they conform to type IV or V. The primary difference between Type IV and V is that type IV exhibits UCST and LCST while type V has LCST only. One important class of systems that exhibit type IV behavior is solvent polymer mixtures such as cyclohexane + polystyrene. Other examples of type IV include carbon dioxide + nitrobenzene and methane + n-hexane while ethane with ethanol or 1-propanol or 1-butanol exhibit type V behavior. Type VI systems are composed of complex molecules with hydrogen bonding or other strong intermolecular forces and result in behavior where LCST and UCST are at temperatures well removed from gas-liquid critical temperature of the more volatile component. Types IV and I are of particular interest in representing behavior exploited in supercritical extraction. There are many more possible classes and subclasses, especially where azeotropic behavior variations are involved, as discussed in the works of Rowlinson10 and King.11 More on the phase behavior applicable to dense gases can be found in reviews12,13 and specialized texts14 in this area. Supercritical extraction often involves separation of relatively non-volatile components, often in the solid phase, through selective solubility in the supercritical gasses. Thus, the critical temperatures of the pure components are likely to be significantly different and the critical temperature of the solvent is likely to be lower than the triple point temperature of the solute. The implication is that there is no common temperature range where both
Figure 21.1.6 Phase behavior of dissimilar systems.
21.1 Supercritical solvents
1425
Figure 21.1.7. The effect of supercritical fluid concentration on polymer-solvent phase behavior.
components will be liquid. The phase diagram of systems with solid and supercritical fluid is shown in Figure 21.1.6.15 The full Figure 21.1.8 Solubility of naphthalene in supercritical ethylene. lines are the sublimation, melting, and vapor pressure curves of the pure components, C1 and C2. Point Q is the quadruple point, where four phases S1, S2, L, and G are in equilibrium. If the temperature of C2 lies far below of that A1 (triple point of C1), the three-phase region S2, L, G curves upward and intersects the gas liquid line to form critical end points Ua and Ub. No liquid phase exits between the temperatures of Ua and Ub. The three-sided region A1C1Ub is a region of gas-liquid phase separation. At temperatures between this region and Ua, the solid phase of component C1 is in equilibrium with C2 rich gas phase. The polymer solution behavior is of significant importance due to large number of applications in the field. The supercritical fluids primarily follow class four behavior. An example related to the effect of supercritical solvent on phase behavior of such systems is shown in Figure 21.1.7. More comprehensive coverage of the field can be found in texts6 and review papers.16 The unique behaviors of supercritical fluids relate to solubility enhancement and how the solubility varies with operating conditions. As can be seen from Figure 21.1.8, the solubility of a solid solute is enhanced several orders of magnitude over the ideal solubility.17 Furthermore, the solubility decreases with increasing temperature at supercritical pressures somewhat above the critical, while at supercritical pressures above a “cross-over pressure”, solubility of a solute increases with temperature. The crossover pressures are different for different solutes. As depicted in Figure 21.1.9, complete fractionation of solutes solubilized in the supercritical solvent is possible by slight increase in temperature.18 Another unique behavior, called “barotropic effect”, exists when highly compressed dense gas reaches pressures higher than the incompressible liquid. In such instances, dense gas phase is below the liquid or the second gas phase. There are three essential elements that make up the thermodynamic foundations of supercritical fluids. These elements are experimental and identification techniques for elucidating the phase behavior, models for dense gases, and computation methods.
1426
Figure 21.1.9 The crossover pressure and retrograde behavior.
Aydin K. Sunol and Sermin G. Sunol
21.1.2.1.1 Experimental methods There are two basic approaches to experimental determination of the high-pressure phase behavior of a system, synthetic and analytic. In the synthetic approach, phase boundaries of a fixed (known) composition system are observed, usually visually, in a cell with sight windows, by manipulation of the system pressure and temperature. In the analytical approach, the components are equilibrated and the compositions of the co-existing phases are determined either through sampling and off-line analysis or on-line spectroscopic techniques. A typical system for determination of vapor/liquid/liquid phase behavior is shown in Figure 21.1.10. The system is a double re-circulation variable volume system. A dynamic variant of this method is for the determination of the solubility of solids in a supercritical fluid. The stationary solid phase is contacted with a continuous supply of
Figure 21.1.10. A high pressure fluid phase equilibrium measurement system.
21.1 Supercritical solvents
1427
Figure 21.1.11. Estimation of solubility and partial molar volume using supercritical fluid chromatography.
supercritical fluid that is ensured to exit with an equilibrium amount of the solid. The chromatographic and calorimetric information can also be used to infer phase behavior information. For example, solubility measurements can be performed on a Supercritical Fluid Chromatograph (SFC) where a relatively short fused silica capillary tubing system replaces the usual column. Sample is coated into this tubing using the same techniques that are used to coat a stationary phase on an analytical capillary column. Raising the solvent pressure stepwise from a selected starting point allows isothermal measurements of solubility. At each step, the ion current in the mass spectrometer, which is used as the detector, is monitored to determine the quantity of solute in solution. Also, solubility may be estimated from the last equations in Approach A and Approach B summarized in Figure 21.1.11.19 This is true provided that the adsorbent is a solid and partial molar volume of the solute (e.g., naphthalene) in the stationary phase is equal to the solid molar volume of the solute. It is worth noting that Supercritical Fluid Chromatography is an important experimental tool for measurement of a portfolio of physical-chemical properties as depicted in Table 21.1.2. However, indirect approaches and methods that involve sampling suffer from magnification of propagated error due to the high sensitivity and non-linear nature of governing fundamental relations in the critical region.
1428
Aydin K. Sunol and Sermin G. Sunol
Table 21.1.2. Classification of physicochemical properties that can be measured by SFC Equilibrium properties * Fluid phase interactions - Second virial coefficients * Solution interactions - Partial molar volumes - Solubilities * Surface interactions - Adsorption isotherms
Kinetic and transport properties * Diffusion coefficients * Mass transfer coefficients
Other properties * Molecular masses
o
Adsorption and desorption rate constants Reaction rate constants
o
*Properties determined by SFC; oProperties determined by GC and can be determined by SFC
Due to the inherent limitation of predictive methods, relative importance of experimental methods become very significant when compared to simple fluids at near ambient condition. Therefore, the area is a popular review3,20 topic and is included in most texts6,21 in the field. 21.1.2.1.2 Computational aspects Computation and definition of critical phenomena is essential in prediction, modeling, and identification of various phases at high pressures. Gibbs presented the definition of critical state in his pioneering paper “On the Equilibrium of Heterogeneous Substances” in 1876.22 For mixtures, mechanical and thermal stability criteria are not sufficient to describe the critical behavior and diffusional (material) stability criteria are required. Various investigators have transformed Gibbs’ criteria to other variable sets and Reid and Beegle23 have developed Gibbs’ criteria in Legendre transforms with suggestions to overcome indeterminacy. Works of Michelsen24 and Heidemann25 combine algorithms with more convenient representation of Gibbs criteria. Generation of complete phase diagrams is not a trivial task and may require utilization of either insight and heuristic guidance or global approaches coupled with Mixed Integer Non-linear Programming (MINLP) or intelligent search algorithms such as genetic algorithms and simulated annealing. The methods used in generation of phase diagrams may either employ the more popular integral approach that is suitable for design purposes or the more insightful differential approach that is preferred for generation of phase diagrams.26 The basic elements of integral approach are the determination of fluid phase partition coefficient, K, for each component as well as stability criterion coupled with a computational algorithm. For determination of fluid phase partition coefficients, K, one may take a symmetric approach that uses Equation of States (EOS) for all fluid phases or an unsymmetrical approach that makes use of liquid activity models for the liquid phase retaining the equation of state models for the gas phase. For symmetric approach:
21.1 Supercritical solvents
Ki =
1429
ϕ$ βi (T, P, x i ) ϕ$ αi (T, P, y i )
For the unsymmetrical approach: Ki
γ iβ (T, P, x i )f i (T, P) ϕ$ αi (T, P, y i )
where: partition coefficient of component i in phases α and β fugacity coefficient of component i in mixture for phase β fugacity coefficient of component i in mixture for phase α activity coefficient of component i in mixture for phase β fugacity of pure component i
Ki ϕ$ βi ϕ$ αi γ βi fi
If one assumes no solvent in the solid phase, the relation for solubility of the solid in the supercritical solvent reduces to: Pisub (T ) 1 yi = P ϕ$ DG i A
B
(
Vi s P − P sub (T ) exp RT
)
C
where: yi Pisub P ϕ$ DG i Vis R T
mole fraction of component i in gas phase sublimation pressure of component i total pressure fugacity coefficient of component i in mixture for dense gas phase partial molar volume of component i in solid phase universal gas constant absolute temperature
The first term A is for ideal gas solubility, the second term B accounts for the non-ideality, while the third term C (Poynting correction) accounts for the pressure effects. The product of the second and third terms is often referred as the Enhancement Factor (over the ideal). In the absence of data, the sublimation pressure may be approximated through extrapolation of the vapor pressure information. 21.1.2.1.3 Modeling The approaches to modeling the high pressure phase behavior follow either use of the equation of state models for fugacity coefficient determination or the treatment of the dense gas as an “expanded liquid” through an activity coefficient model, the former approach being more popular. The macroscopic dense gas modeling approaches include van der Waals family of equation of states, virial family of equation of states, and non-classical approaches. The molecular/theoretical approaches and considerations contribute not only to more comprehensive models but also provide insight bringing forth sound parameters/terms to the macroscopic models. Computer simulation (molecular) can also be used to directly compute phase behavior with some success.27,28 The virial family of equation of states finds limited use in supercritical applications due to the necessity for a large number of terms and
1430
Aydin K. Sunol and Sermin G. Sunol
their inadequacy in representing the critical point. The non-classical scaling approaches29,30 fail when extended beyond the very narrow critical region and their combinations with equations of state (crossover EOSs) are still far from having practical applicability. Therefore, we will focus on the van der Waals family of Equation of States.31 These equations of states have both a repulsive and an attractive term. The most popular variants differ in how the attractive terms are modeled. The success of more recent Peng-Robinson32 (PR) and Soave-Redlich-Kwong33 (SRK) equation of states in phase behavior representation is due to the added parameter, accentric factor that incorporates vapor pressure information into the model. The shortcomings around the critical point are mainly due to the enhanced contribution of repulsive forces that can be modified through Perturbed Hard Chain Theory (PHCT).34 Naturally, one other inherent difficulty is due to the instability of high molecular weight compounds at their critical point. This implies use of other parameters or pseudo critical properties to model the compounds and their mixtures. These equations of states are better in representing molar gas volume then the liquid. The accuracy of molar liquid volume can be improved through volume translation.35 The added terms and parameters naturally complicate the equation of state since the order of the equation increases above cubic. Yet another essential dimension of the equation of states is their representation of mixture properties, which is achieved through mixing rules (models) that often include adjustable binary interaction parameters determined from multi-component, ideally binary, data. Most classical approaches fail to model mixing behavior of systems with dissimilar size components and hydrogen bonding, particularly if one expects to apply them throughout a wide pressure and temperature range that extends from very low to high pressures. Some of the empirical density dependent mixing rules such as Panagiotopoulos-Reid36 can account for the relative size of the solute and the solvent but have theoretical shortcomings violating one fluid model. There are also approaches that combine equation of state with activity models by forcing the mixture EOS to behave like the activity models (GE) at liquid densities.31 EOS-GE models can be combined with group contribution models to make them predictive. More recently, Wong-Sandler37 mixing rules provide an avenue to incorporate more readily available low-pressure information into the equation of states extrapolate throughout the entire pressure range, as well as ensuring fundamentally correct boundary conditions.38 Considering the difficulty associated with representing fugacity coefficients that may vary over ten orders of magnitude by mere change in temperature, pressure and composition; a simple equation of state such as Peng-Robinson, which is given below, do very well in correlating the experimental data and representing the phase behavior well. p=
RT a − Vm − b Vm (Vm + b ) + b (Vm − b ) repulsive term
where:
attractive term
b = ∑ x i bi i
a = ∑ ∑ x i x j( a i a j )
0.5
i
j
(1 − k )
bi = 0.07780 (RTci/pci)
ij
21.1 Supercritical solvents
1431
ai = αi 0.45724 (R2T2ci/pci) αi = [1 + mi(1 - T1/2ri)]2 mi = 0.37464 + 1.54226ωi - 0.26992ωi2 where: p R T Vm b a αi mi ωi kij Tci, Pci xi Tri
pressure universal gas constant temperature molar volume function of Tci and Pci and xi function of Tci, Pci, xi and kij a function of reduced temperature, Tri and accentric factor, ωi a function of accentric factor of component i accentric factor of component i the interaction coefficient of the molecules i and j critical temperature and pressure of component i mole fraction of component i reduced temperature, T/Tci
The polymer solutions39 warrant use of a special class of lattice models such as Florry-Huggins. For correlation purposes Sanchez-Lacombe40 method is sufficient but one may also use Statistical Association Fluid Theory41 (SAFT) models to obtain a better representation. 21.1.2.2 Transport properties of supercritical solvents The transport properties of the supercritical fluids fall somewhat in between the gas and the liquid and also depend on how removed one is from the critical point. Dense gasses have the solubilizing power of liquids and the mobility of gasses as depicted in Table 21.1.3. There are quite a few empirical correlations and theoretical models, which are primarily extensions of corresponding low-pressure liquid and gas counter parts. Similarly, of the classical experimental methods can be used for measurement of transport properties of supercritical fluids. A rather brief overview of the methods applicable for supercritical fluids will be presented since specialized reviews in the area give a good account of the state of the art.30,42 For engineering purposes, one can use applicable property estimation methods available in flowsheet simulators43 such as ASPEN PLUS, PROII, HYSIM, and CHEMCAD. These methods are discussed in a text classical in the field.44 Table 21.1.3. Transport properties of gases, liquids, and supercritical fluids State
Condition o
Property Density, g/cm3 -3
Diffusivity, cm2/s Viscosity, g/cm-s
Gas
1 atm, 25 C
0.6-2×10
1-4×10-1
1-3×10-4
Liquid
1 atm, 25oC
0.6-1.6
0.2-2×10-5
0.2-3×10-2
SC Fluid
Tc, Pc
0.2-0.5
0.5-4×10-3
1-3×10-4
SC Fluid
Tc, 4Pc
0.4-0.9
0.1-1×10-3
3-9×10-4
21.1.2.2.1 Viscosity Both the capillary viscometer (providing about 0.7% accuracy), the theory of which is based on the Hagen-Poiseuille equation and the oscillating disc viscometer (providing about 0.2% accuracy) are applicable to experimental determination of viscosity at high pressures and temperatures.
1432
Aydin K. Sunol and Sermin G. Sunol
Most of the theoretically based estimation methods for viscosity of dense gases rely on modified Enskog theory. Corresponding states based methods are also popular but should be used with care due to their empiric nature. 21.1.2.2.2 Diffusivity The experimental techniques used to measure the diffusion are based upon chromatographic methods,19 NMR (for self-diffusion), and photon correlation spectroscopy.45 The so called chromatographic method for the measurement of diffusion coefficients, strictly speaking, is not a chromatographic method since no adsorption/desorption or retention due to partition in two phases are involved in the method. The experimental system used is however a chromatograph. Diffusion occurs in an empty, inert column on which the fluid phase is not supposed to be adsorbed. The fluid, in which the solute diffuses flows, continuously through the empty column and the solute, which is introduced into the column at one end, is detected at the other end as effluent concentration. The theory of diffusion in flowing fluids is first given by Taylor46 and Aris.47 According to Aris, a sharp band of solute, which is allowed to dissolve in a solvent flowing laminarly in an empty tube, can be described in the limit of a long column as a Gaussian distribution, the variance of which, σ2, in length units is: σ(x) = 2Deff t 2
where: t Deff
time of migration of the peak effective diffusion coefficient given by,
Deff = D12 +
r 2u 2 48D12
where: u r D12
average solvent velocity inner radius of the tube binary diffusion coefficient.
The first term describes the longitudinal diffusion in the axial direction. The second term is called the Taylor diffusion coefficient and describes band broadening due to the parabolic flow profile and therefore radial diffusion. The height equivalent to a theoretical plate, H, is a measure of the relative peak broadening and is defined as H = σ(x) / L 2
where: L
length of the tubing.
Combining the above equations yields H=
2D12 r 2u + u 24D12
Bulk diffusivity can be calculated from the above equation.
21.1 Supercritical solvents
1433
The estimation of low pressure diffusivity is based on the corresponding states theory. The dense gas diffusion coefficient estimation is based on the Enskog theory. The binary diffusion coefficient Dvij at high pressures as modeled by the Dawson-Khoury-Kobayashi correlation, is next given as a representative model. For a binary system, the equations are:
( )
Dvij ρ vm = 1 + a1ρ vrm + a 2 ρ vrm
2
( )
+ a 3 ρ vrm
ρ vrm = Vcm / Vmv , ρ vm = 1 / Vmv , Vcm =
3
Dv (p = 0)ρ (p = 1 atm) m ij
y iVci* + y jVcj* yi + y j
where: ρm ρrm a1,a2,a3 Vcm Vm Vci yi
molar density reduced molar density constants critical molar volume molar volume critical volume of component i mole fraction of component i
Dvij (p=0) is the low pressure binary diffusion coefficient obtained from the ChapmanEnskog-Wilke-Lee model. The molar volume and the molar density are calculated with an equation of state model. 21.1.2.2.3 Thermal conductivity Transient hot-wire and co-axial cylinder methods are typically applied for measurement of the thermal conductivity of supercritical fluids.48 The theory of thermal conductivity is based on the Enskog theory and a typical model for both vapor and liquid thermal conductivity is given by Chung-Lee-Starling. This model makes use of the PRWS (Peng-Robinson with the Wong-Sandler mixing rules) to provide a flexible and predictive equation of state. The main equation for the Chung-Lee-Starling thermal conductivity model is: λ=
. η(p = 0) 312 M
f1 + f 2
where: f1 = fcn(ρ rm , ω , p r , κ) f 2 = fcn(Tc , M, Vcm , ρ rm , ω , p r , κ ) ψ = fcn(Cv , ω , Tr )
η(p=0) can be calculated by the low pressure Chung-Lee-Starling model. The parameter pr is the reduced dipole moment given by: p r = 4152 .
(V
p Tc1/ 2
cm
)
1434
Aydin K. Sunol and Sermin G. Sunol
where: λ η M ρrm ω p pr Vcm Tc, Tr κ
thermal conductivity viscosity molecular mass reduced molar density accentric factor dipole moment reduced dipole moment critical molar volume critical and reduced temperature polar parameter
For low pressures, f1 is reduced to 1.0 and f2 is reduced to zero. This gives the Chung-Lee-Starling expression for thermal conductivity of low pressure gases. The molar density, ρrm, can be calculated using and equation of state model (for example, the Peng-Robinson-Wong-Sandler equation of state) where the mixing rule for b is obtained as follows. The second virial coefficient must depend quadratically on the mole fraction: B(T ) = ∑ ∑ x i x j B ij i
j
with: B ij =
(B
ii
+ B ij
)
2
(1− k ) ij
The relationships between the equation of state at low pressure and the virial coefficient are: B =b −
a a ; B ii = b i − i RT RT
Wong and Sandler has shown that the following mixing rule does satisfy the second virial coefficient equation:
∑∑x x B i
b= 1−
j
i
AmE (p = ∞) ΛRT
j
ij
− ∑ x i B ii i
where: B kij b a AEm
the second virial coefficient the interaction coefficient of the molecules i and j function of Tci and Pci and xi function of Tci, Pci, xi and kij Helmholtz free energy
The pressure correction to the thermal conductivity for a pure component or mixture at low pressure is given by:
(
λv = fcn λv (p = 0), ρ rm , y i , M i ,Tci ,Vci , Z ci
)
21.1 Supercritical solvents
1435
where: ρ rm = ∑ y i i
Vci Vmv
where: λv ρrm yi Mi Tci Vci Zci
thermal conductivity reduced molar density for mixtures mole fraction of component i molecular weight of component i critical temperature of component i critical volume of component i critical compressibility of component i
The vapor molar volume, Vvm can be obtained from the equation of state models as described above. 21.1.2.2.4 Surface tension Stability of phase boundaries depends on the surface tension. Surface tension in a supercritical fluid system is of major importance for drying, surfactant efficacy, and extraction. The surface tension of a gas increases with pressure and approaches zero at the critical point while the surface tension of liquid decreases with pressure resulting in dissolution of supercritical components in the liquid phase. The methods useful in correlating surface tension include Macleod-Sugden correlation and corresponding states theory.21 21.1.2.3 Entrainer (co-solvent effects) of supercritical solvents Entrainers, modifiers, and co-solvents are basically mixed solvent systems and provide another dimension to supercritical fluid extraction. The entrainers enhance the solubility of the low volatile substance in the solvent, provide selective solubility in multi-solute instances, and enhance the sensitivity of the solubility and selectivity to temperature, pressure, and composition. The entrainers may be reactive and are also useful as slurrying media. Table 21.1.4 shows representative data on the effect of entrainers on vapor liquid systems that has been systematically studied by Brunner’s group. Kurnik and Reid50 as well as Johnston’s group51 present data for dense gas-solid systems. Table 21.1.4. The effect of entrainers on separation factor (Adapted from Brunner21) Entrainer Hexadecanol(1) - octadecane(2) - entrainer - nitrous oxide (120 bar, 70°C, y0 = 1.5 wt%, a0 = 2, y(1)+(2) = 2 wt%) acetone methanol dichloromethane benzene methyl acetate hexane cyclohexane
α
3 2.4 2.1 2 2 1.9 1.7
Entrainer
α
Octadecane(1) - salicylic acid phenal ester(2) entrainer - nitrous oxide (120 bar, 70 °C, y0 = 1.5 wt%, a0 = 2, y(1)+(2) = 1.7 wt%) methanol 1.65 dichloromethane 1.5 benzene 1.45 cyclohexane 1.4 acetone 1.4 hexane 1.3 methyl acetate 1.2
1436
Aydin K. Sunol and Sermin G. Sunol
α
Entrainer Hexadecanol(1) - octadecane(2) - entrainer ethane (120 bar, 70°C, y0 = 15 wt%, a0 = 2.5, y(1)+(2) = 30 wt%) methanol acetone hexane (120 bar, 90°C, y(1)+(2) = 20 wt%) methyl acetate benzene dichloromethane
2.2 1.6 1.6
Entrainer
α
Hexadecanol(1) - octadecane(2) - entrainer carbon dioxide (120 bar, 70 °C, y0 = 0.4 wt%, a0 = 2.5, y(1)+(2) = 1 wt%) methanol acetone methyl acetate hexane
3.4 1.8 1.6 1.4
1.8 1.2 1.1
Separation factor, α = (y2/x2)/(y1/x2)
Although entrainers provide the aforementioned added advantages, they also bring forth more complex process flowsheets. The separation of extract from entrainer and entrainer from the supercritical component are not as easy and as sharp as in the instances without entrainer. The selection of entrainer is based on thermodynamic, environmental, and economic considerations. The solute-entrainer and entrainer-supercritical separations are the key from the processing perspective.52 21.1.2.4 Reaction rate implication in supercritical solvents Reactions in supercritical media utilize high pressures. Therefore, the effect of pressure on reaction equilibrium as well as reaction rate plays an important role in supercritical phase reactions.53 The kinetics of the reaction can be explained in terms of the transition-state theory. According to the theory, the reaction occurs via a transition state species M* and the generic elementary reaction can be written as: aA + bB + ............ ↔ M* → Products The effect of pressure on the rate constant is given as: ∂ ln K x* ∂ ln k x = ∂P T ∂P
∆V * ∂ ln κ ∂ ln κ + + = RT ∂P T T ∂P T
where: kx P T K*x κT ∆V* R
rate constant in mole fraction units pressure temperature mole fraction based equilibrium constant for reaction involving reactants and transition state isothermal compressibility activation volume (difference between partial molar volumes of activated complex and reactants), ∆V* = V* - aVA - bVB - ... universal gas constant
The rate constant in the above equation is expressed in terms of pressure independent units (mole fraction). If the rate constant is expressed in terms of pressure dependent units (such as concentration), the relevant equation is:
21.1 Supercritical solvents
1437
∆V * ∂ ln κ ∂ ln k + + κ T (1 − a − b −L) =− RT ∂P T ∂P T If the volume of activation is positive, the reaction is hindered by pressure. However, for high negative values of the volume of activation, the pressure enhances the rate of the reaction. Therefore, supercritical fluids that exhibit very high negative activation volumes for certain reactions will improve the rate of the reaction. The volume of reaction, rather than activation, is crucial in determining the effect of pressure on the equilibrium constant. ∆V ∂ ln K x =− f RT ∂P T , x where: ∆Vf
reaction volume (difference between partial molar volumes of products and reactants)
If the equilibrium constant is expressed in terms of pressure dependent units (such as concentration), the relevant equation is: ∆Vr ∂ ln K + κT =− RT ∂P T , x
∑v
i
where: vi
stoichiometric coefficient
As the above equation implies, supercritical fluids that exhibit very high negative activation volumes for certain reactions will improve the equilibrium conversion of the reaction. 21.1.2.5 Sorption behavior of supercritical solvents Both adsorption from a supercritical fluid to an adsorbent and desorption from an adsorbent find applications in supercritical fluid processing.54,55 The extrapolation of classical sorption theory to supercritical conditions has merits. The supercritical conditions are believed to necessitate monolayer coverage and density dependent isotherms. Considerable success has been observed by the authors in working with an equation of state based upon the Toth isoterm.56 It is also important to note that the retrograde behavior observed for vapor-liquid phase equilibrium is experimentally observed and predicted for sorptive systems. 21.1.2.6 Swelling with supercritical solvents As the pressure of the gas is increased, the solubility of the supercritical gas in the solid polymer increases resulting in swelling, a phenomena that could be advantageous in certain applications while its deleterious impact should be minimized, if not totally eliminated, in other instances. The sorption of supercritical solvent and the resulting swelling could be very high, for example around 30% and 20% respectively for carbon dioxide in polymethylmethacrylate (PMMA). In such instances, the experimental information could be summarized using polymer equation of states such as Sanchez-Lacombe where a single mixture fitting parameter is used.6
1438
Aydin K. Sunol and Sermin G. Sunol
Swelling can be advantageous in that it enables permeation and diffusion of the supercritical fluid into the polymer network. Fragrances, dyes, or medicinal substances loaded in the supercritical fluid can readily impregnate into the polymer and load the polymer with the aforementioned additives. Upon release of the pressure, only the supercritical fluid (i.e., carbon dioxide) flashes off. This avenue has led to a plethora of controlled (timed) release products. Naturally, swelling may be undesirable in many instances as well. For example, swelling of organic polymer based membranes decreased selectivity. Other possible deleterious effects could include malfunctions due to solubilization and swelling of sealants such as gaskets or o-rings. 21.1.2.7 Surfactants and micro-emulsions Most highly polar and ionic species are not amenable to processing with desirable solvents such as carbon dioxide or any other solvent such as water that has a higher critical temperature well above the decomposition temperature of many solutes. In such instances, the combination of the unique properties of supercritical fluids with those of micro-emulsions can be used to increase the range of applications of supercritical fluids.57 The resulting thermodynamically stable systems generally contain water, a surfactant and a supercritical fluid (as opposed to a non-polar liquid in liquid micro-emulsions). The possible supercritical fluids that could be used in these systems include carbon dioxide, ethylene, ethane, propane, propylene, n-butane, and n-pentane while many ionic and non-ionic surfactants can be used. The major difference between the liquid based emulsions and the supercritical ones is the effect of pressure. The pressure affects the miscibility gaps as well as the microstructure of the micro-emulsion phase. The incorporation of the micro-emulsion phase creates interesting potential advantages for reactions as well as separations. Isolation of components from fermentation broths and garment cleaning appear to be two of the more competitive applications of these systems. 21.1.3 SEPARATION WITH SUPERCRITICAL SOLVENTS Supercritical fluids are effective at much lower temperatures than distillation, and their application in separation avoids degradation and decomposition of heat-labile compounds. Attractiveness of supercritical extraction processes are due to the sensitivity of responses to process variables, promise of complete and versatile regeneration of solvents, energy savings, enhanced solute volatilities, solvent selectivities, favorable transport properties for solvents, and state governed effectiveness of solvents which enables the use of low cost, non-toxic, environmentally acceptable solvents. The impact of inherent characteristics of supercritical fluids on separations is summarized in Table 21.1.5. Table 21.1.5 The characteristics and challenges with supercritical separations Inherent characteristics of systems @ supercritical conditions
Resulting promise
Enhanced solubility
Effective at lower temperatures
Favorable transport properties
High mass transfer rates especially in porous media
Challenges to be met Recovery of substances inadvertently extracted
21.1 Supercritical solvents
Inherent characteristics of systems @ supercritical conditions
1439
Resulting promise
Challenges to be met
Effective recovery of solvents, and Sensitivity of responses to process innovative fractionation possibili- Control and optimization variables ties High pressure
High capital cost and need for recovery of mechanical energy
Expansion of solvent spate to enviDominance of physical characteris- ronmentally benign, inexpensive, tic on solvency over chemical and non flammable solvents and their mixtures
Figure 21.1.12. A generic process flow sheet.
There have been many useful attempts made to classify separation technologies. Supercritical fluids are applicable with both intra-phase and inter-phase separations. Due to the ease and flexibility in which a new phase can be formed for regeneration of the solvent, inter-phase is the more common. Furthermore, material solubility and swelling problems, particularly with organic-component based membranes, limit inter-phase separations. This is due to the enhanced solubility of these components in supercritical solvents. Generic steps involved in a typical chemical process are shown in Figure 21.1.12, with each separation sub task type identified. Environmental processes are usually dominated by feed preparation tasks while biochemical processes utilize all of the four subtasks. Blending is considered a finishing task while splitting is considered a feed preparation task. The objective criteria used for the separation may vary according to the subtask. For example, in feed preparation, technical feasibility such as removal of the fines is more appropriate while the usual economic objectives subject to technical feasibility constraints become the goal for the purification and finishing stages. The feed preparation task involves removal of insoluble components and at times non-condensable compounds. The technology selection and the scheduling of tasks are con-
1440
Aydin K. Sunol and Sermin G. Sunol
Figure 21.1.13. Supercritical leaching and options for supercritical solvent regeneration.
centration and particle size distribution dependent. These processes are usually physical/mechanical in nature. The feed preparation may be carried out in a unit designed particularly for the purpose or the function may be tied to another function within a single unit. For example, the fines, either nucleated through temperature and/or pressure perturbation, or particles suspended in the supercritical fluid can be readily separated through cyclones. Concentration (or isolation) tasks are usually necessary with streams where the product concentration is very low (dilute). Adsorption and extraction are typical technologies utilized here. Energy considerations dominate the technology selection subject to product contamination and stability constraints. Thus, the base technology depends on the availability of the Mass Separating Agent (e.g., adsorbent or solvent) and its efficient recycle. Supercritical extraction of organic compounds from aqueous streams is a typical example of isolation step. The candidate technologies for purification are many. Distillation, the work-horse of the chemical processes, leads the pack. Most of the synthesis effort to date has concentrated on the product purification step. This step is often the last step for liquid products especially in the chemical and petrochemical industries. The biochemical industry utilizes membrane and chromatographic processes more than the other industries due to the thermal stability and purity requirements. In the electronic industry, membrane processes are more prevalent due the ultra-purities necessary. Supercritical fractionation of alcohol water systems with the aid of a dense gas is an example of a purification step. Crystallization, drying, and blending are typical operations necessary to polish (finish) the product to its final form. Particle nucleation from a supercritical fluid solution is an example of finishing. Each supercritical separation technique is identified on this generic process.
21.1 Supercritical solvents
1441
21.1.3.1 Leaching - generic application Leaching is usually incorrectly referred to as extraction of a solid substrate, and is commercially the most significant application of supercritical fluids. A generic flowsheet, shown in Figure 21.1.13, illustrates the basic processing concepts. A heat labile natural substrate is usually contacted with a flowing supercritical fluid in a semi-continuous mode. The supercritical fluid may optionally be doped with an entrainer. Due to the stationary (batch) nature of the leaching stage, two or more parallel vessels loaded with the substrate may be utilized. The material leached is recovered and fractionated through pressure reduction, temperature perturbation, adsorption, membranes or through absorption. During the initial stages of leaching, the solute loading is high and may approach the equilibrium solubility limit, while at the latter stages kinetic and mass transfer limitations are responsible from relatively lower the concentration of the solute. Both coffee decaffeination59 and hops extraction60 are popular commercial successes of this application. Further, typical applications are given in Table 21.1.5. Table 21.1.6. Leaching applications of supercritical fluids Purpose
Supercritical medium
Substrate
61
Food processing Food purification
Coffee decaffeination Cholesterol removal from egg yolk
CO2 Essential oils
62
Leaching of natural products
CO2
Oil from soybean, sunflower, spearmint, Camomile, thyme, rosemary sage, chervil
Leaching of natural products
CO2, N2O, CHF3, SF6
Fungal oil
Leaching of natural products
Methanol, ethanol, acetone
Liquid from hazelnut seed coat
Petrochemical Obtaining liquids from solid fuels63
Toluene, water, acetone, methLiquid from coal, lignite anol, THF, tetralin, ethanol
Cleaning and upgrading solid fuels64
Toluene, methanol, ethanol, Sulfur, nitrogen, oxygen removal water, acetone, tetralin, CO2 Chemical
65
Oil from contaminated glass grinds, organic additives from extrusion-molded ceramic parts
Cleaning of parts
CO2
Cleaning of parts
CO2, fluorocarbon solvents Biochemical
Vitamin, pharmaceutical compounds
CO2
Natural plants Environmental
Extraction of soils
Particles from precision devices
66
CO2 + modifier
67
Polycyclic-aromatic hydrocarbons, pesticides
1442
Aydin K. Sunol and Sermin G. Sunol
Purpose
Supercritical medium
Extraction of soils
Nitrous oxide
Extraction of soils
Water
Substrate Aromatic amines, pesticides Polycyclic-aromatic hydrocarbons
6
Polymer processing
Siloxane oligomer, cyclic trimer, organic compounds from plastic, carbon tetrachloride from polyisoprene
CO2
Removal of trace compounds
21.1.3.2 Extraction - generic applications Liquid streams containing close-boiling and heat-labile component can be fractionated with the aid of supercritical fluids or valuable components of aqueous streams could be isolated and concentrated using supercritical solvents. Entrainers could also be added to the solvent to enhance the selectivity, particularly in fractionation instances. The main reason for using entrainers for concentration purposes is to enhance solubility. One can also use membranes along with supercritical fractionator.68 Typical applications of this process technology are given in Table 21.1.7 and Table 21.1.8 while several review papers and texts provide useful design information.69-72 Table 21.1.7. Extraction applications of supercritical fluids Purpose
Supercritical medium
Substrate
Food processing Oil purification
Food purification
CO2
Flavor extraction
CO2
Beer, wine and fruit flavor concentrates
Fractionation of food
CO2
Oil, milk fat
Cholesterol removal from milk fat
Forest products CO2, SO2, N2O, H2O, Extraction of lignocellulosic methylamine, ethanol, materials and biomass ethylene
Cellulose or ligno-cellulosic materials from wood or wood pulp
Biochemical Medicinal compound
CO2
Fermentation broth Petrochemical
Purifying liquid fuels
CO2
Waste water extraction73
CO2+entrainer
Purification of essential oils
CO2
Deasphaltation, demetallization Environmental Aqueous toxic wastes, phenol, metabolic wastes Essential oils Terpenes from orange, citrus and bergamot peel oil
21.1 Supercritical solvents
1443
Table 21.1.8. Fractionation applications of supercritical fluids Purpose
Supercritical medium
Substrate Essential oils
Fractionation74
CO2
Glycerol, fatty acids, glycerides Chemical
Fractionation
CO2
Xylenes, glycerides, alkanes, aromatics
Fractionation
CO2 + entrainer
Methyl esters, fullerenes Petrochemical
Fractionation
Toluene, pentane, propane,
Petroleum pitch, asphalt
Polymer processing CO2
Polydimethylsiloxane, fluoropolymers, polypropylene
The methods useful in design of extraction systems are the same as those traditional methods applicable to liquid and gaseous systems and include Kremser, McCabe Thiele, Panchon Savarit (including Janecke), and more rigorous rate-based as well as equilibrium based methods. Due to the highly non-linear nature of the physical-chemical properties in the critical region, great care has to be taken in representing both the properties and the solution algorithms. 21.1.3.3 Crystallization generic applications Particles nucleated from supercritiFigure 21.1.14. The pressure reduction path for nucleation of partical fluids have unique particle size, cles from supercritical fluids. shape, and particle size distributions.75 The means of achieving nucleation and growth are many, and include pressure reduction, temperature perturbation, and addition of anti-solvents. Time profiling and staging have profound effect on the particle characteristics. The effect of pressure reduction on the resulting nucleated material for different solute concentrations16 is illustrated in Figure 21.1.14. 21.1.3.4 Sorption - generic applications The sorption applications21 include regeneration of porous beds,76 preparative scale supercritical chromatography,21 simulated moving beds,21 thermal swing schemes,77 and adsorption/desorptions cycles. Although initial applications of supercritical fluids in this domain were on regeneration of porous beds, more recent emphasis on fractionation best reflects
1444
Aydin K. Sunol and Sermin G. Sunol
where the true potential of this technique lies. Typical examples are provided in Table 21.1.9. Table 21.1.9 Sorption applications of supercritical fluids21,72 Purpose
Supercritical medium
Substrate
Environmental Soil remediation
CO2 + entrainers
Soil
Regeneration of adsorbents
CO2
Polychlorinated biphenyls from adsorbents Petrochemical Benzene, toluene, ethyl acetate from activated carbon
Regeneration of adsorbents
CO2
Regeneration of catalysts
CO2, THF, hexane, Hydrocarbon, coke removal from catalysts ethylbenzene, benzene
Chromatographic fractionation
CO2
Coal tar, mineral oil distillate 78
Essential oils Chromatographic fractionation
CO2 + alcohol
Black pepper, clove extract
Polymer processing Chromatographic fractionation
CO2, n-pentane+methStyrene oligomer anol Biochemical
Chromatographic fractionation
CO2 + alcohol
Prostaglandis
Chromatographic fractionation
CO2 + alcohol
Vitamins, polyunsaturated fatty acids
Chromatographic chiral separation
CO2
Optical isomers
In synthesis of supercritical sorption processes and their operating policies, the synergy between the characteristics supercritical fluids and the sorption needs to captured. Naturally, the domain requires good distributed process models and solvers in addition to physical property models. 21.1.4 REACTIONS IN SUPERCRITICAL SOLVENTS As stated in the previous sections, supercritical solvents are widely used for separation, especially for extraction of thermally labile compounds. Although the unique properties of supercritical fluids make them attractive as a reaction medium as well, the use of supercritical solvents as reaction medium is becoming more and more popular only in the recent years. The first review on reactions in supercritical fluids was presented by Subramaniam and McHugh,79 whereas a more recent and through review was given by Savage et al.80 The motivations for using supercritical solvents in chemical reactions are many.81,82 In this section, the motivations will be stated and explained with examples from each type of reaction. Emphasis in this chapter will be given to more recent examples, which were not included in the previous review papers.
21.1 Supercritical solvents
1445
Reasons for carrying out reactions in supercritical fluids: There are several reasons for carrying out reactions in the supercritical phase. Naturally, some of the reasons are coupled. Nevertheless, they, in general, relate to control, favorable mass transfer and kinetic considerations. • Enhanced reaction rate The effect of pressure on the reaction rate and equilibrium constant at high pressures is described in Section 21.1.2.4. As can be perceived from this section, supercritical fluids that exhibit very high negative activation volumes for certain reactions will improve the rate and equilibrium conversion of the reaction. • Homogenization Reactions that otherwise would be carried out in more than one phase (heterogeneous reactions) can be transformed to homogeneous ones with the aid of supercritical fluids, so that inter-phase transport limitations are eliminated. This is realized due to enhanced solubilities of the reaction components in the supercritical fluids. Typical examples are reactions in water (supercritical water can solubilize organic compounds), homogeneous catalytic reactions, and reactions of organometallic compounds. Homogenizing one compound more than the others in a system may also affect relative rates in complex reactions and enhance the selectivity. • Enhanced mass transfer In many instances, reaction rates are limited by diffusion in the liquid phase. The rate of these reactions can be increased if the reaction is carried out in the supercritical phase. Typical examples are enzyme catalyzed reactions as well as some very fast reactions such as certain free radical reactions. Selectivity considerations usually dominate in complex reactions. If some steps of the complex reaction are controlled by diffusion, changing the diffusivity changes the relative rates of the reaction steps and affects the selectivity. • Ease of down-stream separation Another reason for using supercritical fluids as the reaction medium is to fractionate products, to purify the products or to remove unreacted reactants from the product stream. Supercritical fluids can be used as either a solvent or anti-solvent in these instances. • Increased catalyst activity Some heterogeneous catalytic reactions are carried out in the supercritical phase, in order to increase catalyst activity and life through in-situ regeneration of surfaces with tuning of operation conditions. For example, supercritical fluids are capable of dissolving carbon that may otherwise be deposited on the catalyst in the absence of the supercritical solvent. • Tunable reaction rates through dielectric constant Some properties of supercritical fluids can be monitored (manipulated) continuously by adjusting the density of the fluid. Dielectric constant is such a property and the solvent’s dielectric constant can influence the rate of the reaction. 21.1.4.1 Homogenous reactions in supercritical solvents - examples Homogeneous reactions carried out in supercritical fluids can be either catalytic or non-catalytic. The objective of carrying out both catalytic and non-catalytic reactions in supercritical fluids is to increase the overall rate of the reaction by eliminating the inter-phase transport effects.
1446
Aydin K. Sunol and Sermin G. Sunol
21.1.4.1.1 Homogeneous reactions catalyzed by organometallic compounds Homogeneous catalysts have advantages over heterogeneous catalysts such as possibility of carrying out the reaction at milder conditions, higher activity, and selectivity, ease of spectroscopic monitoring, and controlled and tunable reaction sites. Organic reactants and products are not soluble in water while most catalytic materials are soluble in water. Therefore, homogenization of organic systems utilized environmentally undesirable organic solvents. An alternative to environmentally unacceptable organic solvents is a supercritical solvent that has added advantages over organic solvents such as increased reaction rate, higher selectivity and easy separation of reactants and products as well as of the catalyst after the reaction. Since the properties of supercritical solvents can be adjusted by manipulating the operating conditions, reaction rate and selectivity are better tunable in reactions carried out in supercritical solvents. Carbon dioxide is the supercritical solvent that is most commonly used in homogeneous catalytic reactions. In addition to being environmentally acceptable (nontoxic, nonflammable), inexpensive, and available in large quantities, carbon dioxide does not participate in most reactions. It also has an ambient critical temperature. Although, supercritical carbon dioxide is more effective in dissolution of non-polar, nonionic and low molecular mass compounds, addition of co-solvents enhances the solubility of many otherwise insoluble compounds in supercritical carbon dioxide. A recent review by Noyori et al.83 discusses homogeneous catalytic reactions under supercritical conditions. When homogeneous reactions are carried out under supercritical conditions, gas/liquid interfacial transport is eliminated, which is an advantage for reactions such as hydrogenation, where diffusion of gas into the liquid may be limiting the reaction rate. In asymmetric hydrogenation reactions, hydrogen and the supercritical solvent are miscible and this results in better enantioselectivity. In Diels-Alder reactions, the advantage of the supercritical solvent is the higher selectivity obtained rather than increased rate of the reaction due to the solvent. Most of the oxidation reactions are carried out in supercritical water. Some heterogeneously catalyzed reactions are also carried out in supercritical carbon dioxide. Recently, homogeneously catalyzed reactions carried out in supercritical carbon dioxide have been reported. Examples of homogeneous catalytic reactions carried out under supercritical conditions are summarized in Table 21.1.10. Table 21.1.10. Homogenous reactions in supercritical carbon dioxide catalyzed by organametallic compounds Reaction
Catalyst Isomerization
1-hexene to 2-hexene
Iron catalyst
Hydrogenation CO2 to formic acid
Ruthenium(II) phosphine complex
Asymmetric hydrogenation of tiglic acid
Ruthenium catalyst
Asymmetric hydrogenation of enamides
Cationic rhodium complex
Cyclopropene
Manganese catalyst
21.1 Supercritical solvents
1447
Reaction
Catalyst Hydroformylation
1-Octene
Rhodium catalyst
Propylene
Cobalt carbonyl Olefin metathesis
Ring opening metathesis polymerization
Ruthenium catalyst
Ring closing metathesis of dienes to cyclic olefins
Ruthenium catalyst
Diels-Alder reactions Synthesis of 2-pyrones
Nickel catalyst
Synthesis of cyclopentones
Cobalt catalyst
Cyclotrimerization of alkynes to substituted benzene derivaCobalt catalyst tives Oxidation Alkene epoxidation
Molybdenum catalyst
2,3-Dimethylbutene epoxidation
Molybdenum catalyst
Cyclooctene epoxidation
Molybdenum catalyst
Cyclohexene epoxidation
Molybdenum catalyst
21.1.4.1.2 Homogeneous reactions of supercritical water Homogeneous reactions carried out in supercritical fluids are reactions in supercritical water, organo-metallic reactions and Diels Alder reactions. Reactions in supercritical water are well studied84 and will be described in the following section. Despite the higher temperature and pressures required in supercritical water applications, this solvent possesses unique properties that make it attractive as a reaction medium. Supercritical water has a lower dielectric constant as compared to liquid water, and the dielectric constant of supercritical water changes significantly with the density. Also, the effect of hydrogen bonding is less pronounced at supercritical conditions, one consequence of which is the high solubility of organics in supercritical water. The reactions in the supercritical water medium are carried out in a single phase, which implies high reactant concentration and negligible inter-phase mass transfer resistance. Also, the ion dissociation constant of water is higher in the critical region and is lower as supercritical conditions are accessed. These properties also vary continuously in the supercritical region, so that they can be tuned during the reaction by changing the temperature and/or pressure. Examples of the homogeneous reactions are given in Table 21.1.11. In some of the reactions discussed in this category, acid or base catalysts were used to enhance the rates.84 Table 21.1.11. Homogeneous reactions of supercritical water Reaction
Catalyst
C-C Bond formation Phenol and p-cresol alkylation
None
1448
Aydin K. Sunol and Sermin G. Sunol
Reaction
Catalyst
Diels-Alder cycloaddition
None/NaOH
Ring opening of 2,5-dimethylfuran
Acid Hydration/dehydration
Conversion of tert-butyl alcohol to isobutylene
None/H2SO4/NaOH
Dehydration of cyclohexanol, 2-methyl cyclohexanol, 2-phenylethanol
Acid
Hydrolysis Esters to carboxylic acids and alcohols
Autocatalytic
Nitriles to amides and then to acid
Autocatalytic
Butyronitrile
Autocatalytic
Polyethyleneteraphthalate and polyurethane
None
Diaryl ether to hydroxyarene
None
Triglycerides into fatty acid
None Decomposition
Cellulose and glucose decomposition
None
Nitrobenzene
None
4-Nitroaniline
None
4-Nitrotoluene
None Oxidation
Phenols
None
Ethanol
None
2-Propanol
None
2-Butanol
None
Chlorinated hydrocarbons
None
21.1.4.1.3 Homogeneous non-catalytic reactions in supercritical solvents The use of supercritical fluids as reaction media for organometallic species has also been investigated.85 Reactions include photochemical replacement of carbon monoxide with N2 and H2 in metal carbonyls, where the reaction medium is supercritical xenon. Also, photochemical activation of C-H bonds by organometallic complexes in supercritical carbon dioxide has also been investigated. More recent studies on photochemical reactions also include laser flash photolysis of metal carbonyls in supercritical carbon dioxide and ethane86 and laser flash photolysis of the hydrogen abstraction reaction of triplet benxophenone87 in supercritical ethane and CHF3. 21.1.4.2 Heterogeneous reactions in supercritical solvents - examples Heterogeneous reactions in supercritical fluids can be catalytic or non-catalytic. Catalytic heterogeneous reactions are reactions carried out on solid catalysts and are of great impor-
21.1 Supercritical solvents
1449
tance in the chemical process industries. As described in the next section, the advantages of carrying out these reactions in a supercritical medium include enhanced inter-phase and intra-particle mass and heat transfer and in-situ regeneration of catalyst. Catalytic supercritical water oxidation will also be discussed. Other heterogeneous reactions that will be described are fuels processing and treatment of biomass. 21.1.4.2.1 Heterogeneous catalytic reactions in supercritical solvents Obviously, a solid catalyzed reaction takes place only on the active sites of the porous catalyst with the implication of some mass and heat transport steps prior to and after the reaction. The first step is the diffusion of the reactants through the film surrounding the catalyst particle to the external surface of the catalyst, followed by diffusion of the reactants into the catalyst pore to the active site in the pores. These steps are limited by the diffusivity and viscosity of the reactants. In the case of a supercritical fluid phase reaction, the diffusivity is higher than the liquid diffusivity, viscosity is less than the liquid viscosity and therefore, the rate of transfer to the active site will be higher. After the adsorption, reaction and desorption steps, the products have to diffuse out of the pore, and again through the film surrounding the particle into the bulk fluid. Rates of these steps can be accelerated utilizing a supercritical medium for the reaction. Heat transfer effects are also important in a solid catalyzed reaction. Higher thermal conductivity of supercritical fluids is an advantage as well.88 For two-phase reactions (typically hydrogenation and oxidation reactions), the reaction steps include the diffusion of the gas reactant to and through the gas-liquid interface and then into the bulk liquid. This mass transfer limitation is also eliminated if the reaction is carried out in a supercritical medium where the reaction takes place in a single phase.88 Supercritical fluids bring other benefits to the solid catalyzed reaction rate besides eliminating or minimizing mass and heat transfer resistance. Supercritical solvents have the ability to regenerate the catalyst during the course of the reaction, which increases the catalyst life and activity, since undesirable deposits on the catalyst, such as carbon deposits, are soluble in the supercritical fluids. The rate of the intrinsic reaction is increased in supercritical fluids and tuning the properties of the supercritical medium can control the selectivity.88 Supercritical fluids may also bring opportunities in downstream separation of the reactants and products. Examples of solid catalyzed reactions in supercritical fluids are given in Table 21.1.12.88 Table 21.1.12. Solid catalyzed reactions in supercritical solvents Reaction
Supercritical medium
Catalyst
Hydrogenation Fats and oils
Propane, CO2
Supported platinum, palladium catalysts
Acetophenone, cyclohexene
CO2
Palladium on polysiloxane
Fischer-Tropsch synthesis
n-hexane, n-pentane, propane
Fe, Ru, Co/Al2O3, SiO2
Oxidation Toluene
CO2
Co/Al2O3
Propene
SC reactant
CaI2, CuI, Cu2/MgO or Al2O3
Isobutane
SC reactant
SiO2,TiO3, Pd/carbon
1450
Aydin K. Sunol and Sermin G. Sunol
Reaction
Supercritical medium
Catalyst
Cracking Heptane
CO2
Zeolite Isomerization
1-Hexene
CO2, and co-solvents
Pt/Al2O3
Xylene
SC reactant
Solid acid catalyst Alkylation
Benzene, ethylene, isopentane, CO2 or SC reactant isobutene, isobutane Mesitylene, propene, propan-2-ol
Zeolite
CO2 Disproportionation
Toluene to p-xylene, benzene
SC reactant
Zeolite
Ethylbenzene to benzene and diethylbenzene
Butane, pentane
Zeolite
Catalytic Supercritical Water Oxidation is an important class of solid catalyzed reactions that utilize advantageous solution properties of supercritical water (dielectric constant, electrolytic conductance, dissociation constant, hydrogen bonding) as well as the superior transport properties of the supercritical medium (viscosity, heat capacity, diffusion coefficient, density). The most commonly encountered oxidation reactions carried out in supercritical water are oxidation of alcohols, acetic acid, ammonia, benzene, benzoic acid, butanol, chlorophenol, dichlorobenzene, phenol, 2-propanol, (catalyzed by metal oxide catalysts such as CuO/ZnO, TiO2, MnO2, KMnO4, V2O5, Cr2O3) and 2,4-dichlorophenol, MEK, and pyridine (catalyzed by supported noble metal catalysts such as supported platinum).89 21.1.4.2.2 Heterogeneous non-catalytic reactions in supercritical solvents Use of the supercritical fluids as the reaction medium in synfuel processing is one of the earliest applications in the field. The advantage of the supercritical fluid as the reaction medium are again three-fold. During thermal degradation of fuels (oil-shale, coal), primary pyrolysis products usually undergo secondary reactions yielding to repolymerization (coking) or cracking into gas phase. Both reactions decrease the yield of the desired product (oil). To overcome this problem, a dense (supercritical) hydrogen donor (tetralin), or non-hydrogen donor (toluene), or an inorganic (water) medium is used.63 Also, a supercritical medium provides ease of transport in the porous coal matrix. Finally, downstream processing (separation of the products) becomes an easier task when a supercritical medium is used. Kershaw90 reviews the use of supercritical fluids in coal processing, while Sunol discusses the mechanism.91 The forest product applications in this category include biomass conversion92 and delignification for pulping purposes.93
21.1 Supercritical solvents
1451
21.1.4.3 Biochemical reactions - examples Due to their tunable properties, supercritical solvents provide a useful medium for enzyme catalyzed reactions.94 The mechanism of enzyme-catalyzed reactions is similar to the mechanism described for solid catalyzed reactions. External as well as internal transport effects may limit the reaction rate. Utilizing supercritical fluids enhances external transport rate due to an increase in the diffusivity and therefore mass transfer coefficient. Internal transport rate depends on the fluid medium as well as the morphology of the enzyme. Supercritical fluids can alter both. Water is known to be essential for the enzyme activity. Small amounts of water enhance enzyme activity, however excess water hinders the rate of some enzyme catalyzed reactions. The active site concentration on enzymes, hence the enzyme activity is found to be higher in the presence of hydrophobic supercritical fluids (ethane, ethylene) as compared to hydrophilic supercritical carbon dioxide. The effect of pressure on enzyme catalyzed reactions can be explained in terms of the transition theory. Supercritical fluids that exhibit very high negative activation volumes for certain reactions are expected to improve the rate of the reaction. Although, supercritical carbon dioxide has the advantage of being non-toxic and abundant, it is practically immiscible with water. Therefore, supercritical fluids used as the reaction medium in enzyme catalyzed reactions include fluoroform, sulfur hexafluoride and ethane, while lipases are the enzymes utilized in such reactions.95 21.1.4.4 Polymerization reactions - examples Supercritical carbon dioxide is a promising green alternative to traditional solvents in polymer synthesis due to gas-like transport properties and liquid-like solubility. Supercritical carbon dioxide can be removed easily from the polymer solution by depressurization during drying of the polymer. Supercritical carbon dioxide provides easy separation of the polymer from the unreacted monomers and catalysts. Finally, supercritical carbon dioxide also exhibits Lewis acid-base interactions with electron donating functional groups of polymer chains.96 Examples of homogeneous and heterogeneous polymerization reactions carried out in supercritical carbon dioxide are given in Table 21.1.13. Table 21.1.13. Polymerization reactions in supercritical fluids Polymerization mechanism
Substrate
Homogeneous, free radical/cationic polymerization
Amorphous fluoropolymers
Precipitation, free radical polymerization
Vinyl polymer, semicrystalline fluoropolymers
Dispersion, free radical polymerization
Polyvinyl acetate and ethylene vinyl acetate copolymer
Dispersion, cationic polymerization
Isobutylene polymer
Homogeneous/precipitation, cationic polymerization
Vinyl ether polymer
Transition metal catalyzed, ring opening methathesis Norbornene polymer, polycarbonate polymerization
Butane, pentane, and propane are also used as the reaction medium in polymer synthesis.97 Furthermore, some polymerization reactions (such as polyethylene synthesis) are carried out under supercritical conditions of the monomer.
1452
Aydin K. Sunol and Sermin G. Sunol
Figure 21.1.15. A flexible pilot plant for supercritical fluid aided materials processing.
21.1.4.5 Materials processing with supercritical solvents The material field related applications of supercritical fluids are rapidly developing with exciting innovative developments continually emerging. The recent increase of activity in the supercritical field at large is partially due to these new horizons. Supercritical fluids are used to make highly porous material, aids in making uniform porous or non-porous films, to deposit solubilized components by diffusion and permeation into porous media resulting in composites with tailored characteristics or functionalized surfaces, is used to make particles with desired shape and particle size distribution, is used to encapsulate particles for time release applications, is used to fractionate macromolecules to desired dispersity and molecular weight distributions, and soon may be part of our household for cleaning/washing. What makes supercritical fluids so attractive in this domain is their sensitivity to a large number of processing variables in a region where transition from a single or multiphase system into another is rather simple through a variety of paths. One can start with a homogeneous phase and use pressure, temperature, mass separating agents, other external fields such as electromagnetic or irradiation, to nucleate and grow, or react or fractionate, to form new material with unique performance characteristics. In the homogenization step, supercritical fluids are used to solubilize. If solubilization in the supercritical fluid is not possible, the supercritical fluid can be used to induce phase separation as an anti-solvent in a subsequent step. The supercritical fluids are effective in heterogeneous environments as well. They penetrate into porous environment loaded with additives or, used as a pure supercritical fluid to clean, dry (extract), coat, impregnate and process (e.g. extrude) a low viscosity solution. A flexible pilot plant that addresses all the materials processing demands excluding extrusion is shown in Figure 21.1.15 while the modes for encapsulation and aerogel/impregnation are expanded upon Figures 21.1.16 and 21.1.17 respectively.
21.1 Supercritical solvents
1453
Figure 21.1.16. Particle coating section of the plant.
Figure 21.1.17. Supercritical drying section of the plant.
21.1.4.6 Particle synthesis - generic application The routes to particle synthesis via supercritical fluids basically follow two paths, Rapid Expansion of Supercritical Solution (RESS) and Supercritical Anti-Solvent (SAS).6,98 The basic processing steps are outlined in Figure 21.18 and Figure 21.19, respectively. RESS involves homogenization of the particles raw material in the supercritical fluid followed by rapid expansion of the solution through an expansion device such as a nozzle. Depending on the nozzle design, time-temperature and time-pressure profiles, and whether one uses
1454
Figure 21.1.18. Homogeneous supercritical solution path to materials synthesis.
Aydin K. Sunol and Sermin G. Sunol
Figure 21.1.19. Anti-solvent path to materials synthesis.
entrainers or not, the particle size, shape, and particle size distribution changes. The primary particle creation mechanism is nucleation and growth. Usually the amount solubilized in the SCF is not very high for particle synthesis. For many applications such as pharmaceutical and energetic materials, the particle base material is not soluble in dense gasses. In those instances, these substances are solubilized in a liquid solvent and in a subsequent stage the supercritical anti-solvent is used for synthesis of particles. 21.1.4.7 Encapsulation - generic application Micro-encapsulation of drug-polymer systems using the RESS (Rapid Expansion of Supercritical Fluid Solutions) techniques have been initiated with limited success due to poor understanding of the complex phenomena involved in co-nucleation of components. Not only do the particles have to be nucleated with the desired particle size and shape but also encapsulate the material simultaneously in an uniform fashion. An alternative approach99 depicted in Figure 21.1.17 involves a sequential method where synthesized particles are coated with polymeric thin films by simultaneous nucleation of polymeric material out of a supercritical fluid, encapsulating the particles fluidized in the supercritical fluid, followed by further polymerization and binding of the encapsulating material on the particle surface. The method involves a recirculatory system that includes dissolution of the polymer in the supercritical solvent and coating the particles through a temperature swing operation in the fluidized bed that contains the particles. The particulate material coated with the tailor polymeric material possesses unique timed-release characteristics, improved stability, and often-enhanced behavior. 21.1.4.8 Spraying and coating - generic application The RESS and SAS approaches can be applied to coating and spraying applications if the inhomogenization step is exercised on a surface (a base material). 21.1.4.9 Extrusion - generic application The addition of a supercritical fluid to a Newtonian or non-Newtonian fluid reduces the viscosity of the fluid improving its processibility and end materials morphology. Supercritical fluids can aid in food and polymer extrusion. 21.1.4.10 Perfusion (impregnation) - generic application Excellent transport characteristics, solubilization power, and sensitivity to process variables all contribute to the success of the methodology. The generic scheme as shown in Fig-
21.1 Supercritical solvents
1455
Figure 21.1.20. Supercritical fluid impregnation processes. Figure 21.1.21. Sol-gel drying methods.
ure 21.1.20 involves homogenization, impregnation, deposition, and optional post treatment steps such as curing reactions. The mechanism of transport in the porous matrix is permeation and diffusion while the primary mechanisms for deposition are pressure reduction, temperature swing, sorption, and reaction (i.e., polymerization). This process is a simplified version of impregnation process.100 The impregnating solvent does not contain any material to be deposited and the pressure release causes disintegration of the impregnated material.6,72 21.1.4.11 Parts cleaning - generic application In essence, the part cleaning process is basically extraction/leaching with or without surfactants. The basic steps are shown in Figure 21.1.13. Drying in the absence of capillary forces and solvent residue free substrate makes the technology attractive. The technology is comprehensive covered in a monograph65 and there are many existing commercial applications. 21.1.4.12 Drying - generic application A pilot plant used in drying is shown in Figures 21.1.15 and 21.1.17. The supercritical drying routes are particularly attractive in their ability to eliminate or at least minimize the capillary effect that cause non-uniformities in films as well as shrinkage and collapsing in pore structure. These supercritical avenues permit successful creation of highly porous structures such as foams, aerogels, coatings, and films. The solvent can be removed from the wet gel using different methods. These methods and resulting gels are shown in Figure 21.1.21. If solvent is evaporated slowly from the gel, a xerogel is obtained. During evaporation, large capillary forces are exerted as the liquid-vapor interface moves through the gel. These forces cause shrinkage of the pores within the gel. Removal of the solvent (alcohol) from the gel under supercritical conditions results in the formation of the aerogel. Since this drying procedure eliminates the liquid-vapor interface, aerogels are formed in the absence of capillary forces. Aerogels retain the morphology of the original alcogel. There are several methods developed for removing the solvent from the gel under supercritical conditions. The first one is the one suggested in the pioneering work by Kistler,101 in which the solvent is brought to supercritical conditions in an autoclave and evacuated under these conditions. In order to pressurize the autoclave to a pressure above the critical value for the alcohol, more alcohol is added to the autoclave. Supercritical conditions of the solvent are reached by supplying heat to the autoclave. After the pressure
1456
Aydin K. Sunol and Sermin G. Sunol
reaches a preselected value, which is above the critical, the temperature is raised at constant pressure. Once the temperature is also above its critical value, the supercritical fluid is vented out of the autoclave at constant temperature.102 The P-T behavior for this procedure is shown as route A-B (1) in Figure 21.1.22. Van Lierop and co-workers later imFigure 21.1.22. Pressure-temperature paths for super- proved this method. In this improved vercritical drying. sion of the method, any boiling of the solvent was completely suppressed. This was done by introducing pressure of an inert gas into the autoclave prior to any heat treatment and furthermore, the autoclave was kept closed during heating until the temperature reached the predetermined value above the critical. The P-T behavior for this process is shown as route A’-B (2) in Figure 21.1.22. In another method, developed by Jacobucci104 and co-workers, alcohol is removed by supercritical CO2 extraction in a semi-continuous system. Solvent exchange is another method utilized to dry alcogels. The liquid in the pores (excess water of the hydrolysis) of the wet-gel is first displaced by alcohol and then alcohol is displaced by liquid CO2 at about 16-20oC and 100 bar. After solvent exchange, the temperature of the system is increased to 40oC and supercritical carbon dioxide is vented out of the system.105 In yet another method, the system is pressurized with carbon dioxide prior to drying. Through partial solvent exchange during heating, drying of the gel at temperatures much lower than the critical temperature of the solvent (alcohol) can be achieved. Also, gels can be dried at a wide range of temperature and pressure combinations where optimum tunable conditions (resulting in optimum pore and surface characteristics) can be determined for a variety of objectives and cases.106 REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12
Hannay, and Hogarth, Proc. Roy. Soc. A, 29, 324 (1879). J. Timmermans, J. Chim. Phys., 20, 491 (1923). G.M. Schneider, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 91-115. G.M. Schneider, in Water-A Comprehensive Treatise, vol. 2, F. Frank Ed., Plenum Press, New York, 1973, Chp. 6. E.U. Franck, in Physical Chemistry: Advanced Treatise, H. Eyring, D. Henderson, and W. Jost, Eds., Vol. 1, Academic Press, New York, 1971 M.A. McHugh, V.J. Krukonis, Supercritical Fluid Extraction, Principles and Practice, Butterworth-Heinemann, Boston, 1994. C. J. Peters, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J. M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 117-146. J. DeSwaan Arons, and T.W. deLoss, in Models for Thermodynamic and Phase Equilibria Calculations, S.I. Sandler, Ed., Marcel Dekker, New York, 1994, pp. 363-505. P.H. Van Konynenburg, and R.L. Scott, Philos. Trans. R. Soc., 298, 495 (1980). J.S. Rowlinson, F.L. Swinton, Liquids and Liquid Mixtures, Butterworths, London, 1982. M.B. King, Phase Equilibria in Mixtures, Pergamon Press, Oxford, 1969. J.C. Rainwater, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 57-162.
21.1 Supercritical solvents
13 14 15 16 17 18 19
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
50
1457
M.P. Ekart, J.F. Brenneke, and C.A. Eckert, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 163-192. R. J. Sadus, High Pressure Phase Behaviour of Multicomponent Fluid Mixtures, Elsevier, Amsterdam, 1992. W.B. Street, in Chemical Engineering at Supercritical Fluid Conditions, M.E. Paulatis, J.M.L. Penninger, R.D. Gray, and P. Davidson, Eds., Ann Arbor Science, Ann Arbor, MI, 1983. E. Kiran, and W. Zhuang, in Supercritical Fluids, Extraction and Pollution Prevention, M.A. Abraham, and A.K. Sunol, Eds., American Chemical Society, Washington, DC, 1997, pp. 2-36. Y. Tsekhankaya, M.B. Jomtev, and E.V. Mashkina, Z. Fiz. Chim., 38, 2166 (1964). E.H. Chimowitz and K.J. Pennisi, AICHE J., 32, 1665 (1986). S.G. Sunol, B.Mierau, I. Serifoglu, and A.K. Sunol, in Supercritical Fluids, Extraction and Pollution Prevention, M.A. Abraham, and A.K. Sunol Eds., American Chemical Society, Washington, DC, 1997, pp.188-206. T.J. Bruno, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 293-324. G. Brunner, Gas Extraction, Springer, New York, 1994. J.W. Gibbs, Trans. Conn. Acad. III, 108 (1876), 343 (1878). R.C. Reid, and B.L. Beegle, AIChE J., 23, (1977). M.L. Michelsen, and R.A. Heidemann, AIChE J., 27, 521 (1981). R.A. Heidemann, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 39-64. J.W. Tester, and M. Modell, Thermodynamics and Its Applications, Prentice Hall PTR, New Jersey, 1997. A.Z. Panagiotopoulos, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 411-437. K.S. Shing, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 227-244. J.M.H. Sengers, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levelt Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 3-38. J.V. Sengers, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levelt Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 231-272. S.I. Sandler, H. Orbey, and B.-I. Lee, in Models for Thermodynamic and Phase Equilibria Calculations, S.I. Sandler, Ed., Marcel Dekker, New York, 1994, pp. 87-186. D.Y. Peng, and D.B. Robinson, Ind. Eng. Chem. Fundam., 15, 59 (1976). G.S. Soave, Chem. Eng. Sci., 27, 1197 (1972). J.M. Prausnitz, R.N. Lichtenthaler, E.G. de Azevedo, 3rd ed., Prentice Hall PTR, New Jersey, 1999. J.J. Martin, Ind. Eng. Chem. Fundam., 18, 81 (1979). A.Z. Panagiotopoulos and R.C. Reid, Fluid Phase Equilibria, 29, 525 (1986). D.S.H. Wong and S.L. Sandler, AICHE J., 38, 671 (1992). H. Orbey, and S.I. Sandler, Modeling of Vapor-Liquid Equilibria, Cambridge University Press, Cambridge, 1998. I.C. Sanchez, and C. Panayiotou, in Models for Thermodynamic and Phase Equilibria Calculations, S.I. Sandler, Ed., Marcel Dekker, New York, 1994, pp. 187-285. I.C. Sanchez, and R.H. Lacombe, J. Phys. Chem., 80, 2352, 2568 (1976). W.G. Chapman, G. Jackson, and K.E. Gubbins, Molec. Phys., 65, 1057 (1988). V. Vesovic, and W.A. Wakeham, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 245-289. W.D. Seider, J.D. Seader, and D.R. Lewin, Process Design Principles, Wiley, New York, 1999. R.C. Reid, J.M. Prausnitz, and B Poiling, The Properties of Gases and Liquids, 4th ed., McGraw Hill, New York (1987). H. Saad and E. Gulari, Ber. Bunsenges. Phys. Chem., 88, 834 (1984). G. Taylor, Proc. Roy. Soc. A, 219, 186 (1953). R. Aris, Proc. Roy. Soc. A, 235, 67 (1956). C.A.N. deCastro, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 335-363. S. Peter, A. Blaha-Schnabel, H. Schiemann, and E. Weidner, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 731-738 R.T. Kurnik, and R.C. Reid, J. Fluid Phase Equilib., 8, 93 (1982).
1458
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
Aydin K. Sunol and Sermin G. Sunol
J.M. Dobbs, J.M. Wong, and K.P. Johnston, Ind. Eng. Chem. Res., 26, 56 (1987) A.K. Sunol, B. Hagh, and S. Chen, in Supercritical Fluid Technology, J.M.L. Penniger, M. Radosz, M.A. McHugh, and V.J. Krukonis, Eds., Elsevier, Amsterdam, 1985, pp. 451-464. C.D. Hubbard and R. van Eldik, in Chemistry Under Extreme or Non-Classical Conditions, R. van Eldik, and C.D. Hubbard, Eds., Wiley, New York, 1997, pp. 53-102. C.-S. Tan, and D.-C. Liou, Ind. Eng. Chem. Res., 29, 1412 (1990). C.-S. Tan, and D.-C. Liou, Ind. Eng. Chem. Res., 27, 988 (1988). U. Akman, A.K. Sunol, AIChE J., 37, 215 (1991). E.J. Beckman, J.L. Fulton, and R.D. Smith, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 405-449. J.L. Bravo, J.R. Fair, J.L. Humphrey, C.L. Martin, A.F. Seibert, and S. Joshi, Fluid Mixture Separation Technologies for Cost Reduction and Process Improvement, Noyes Pub., New Jersey, 1986. E. Lack, and H. Seidlitz, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 101-139. D.S. Gardner, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 84-100. N. Sanders, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 34-49. D.A. Moyler, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 140-183. R.R. Maddox, J. Gibson, and D.F. Williams, CEP, June (1979). A. Wilhelm, and K. Hedden, in Supercritical Fluid Technology, J.M.L. Penniger, M. Radosz, M.A. McHugh, and V.J. Krukonis, Eds., Elsevier, Amsterdam, 1985, pp. 357-375. J. McHardy and S. P. Sawan Eds. Supercritical Fluid Cleaning, Noyes Pub., New Jersey, 1998 U. Nguyen, D.A. Evans, and G. Frakman, in Supercritical Fluid Processing of Food and Biomaterials, H. Rizvi, Ed., Blackie Academic & Professional, London, 1994, pp. 103-113. A. Akgerman, G. Madras, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 669-695. J.-H. Hsu, and C.-S. Tan, in., Supercritical Fluid Processing of Food and Biomaterials, H. Rizvi, Ed. Blackie Academic & Professional, London, 1994, pp. 114-122. M.B. King, and O. Catchpole, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 184-231. R. Eggers, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 232-260. G. Vetter, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 261-298. E. Stahl, K.W. Quirin, D. Gerard, Dense Gases for Extraction and Refining, Springer-Verlag, Berlin, 1988. H. Schmieder, N. Dahmen, J. Schon, and G.Wiegand, in Chemistry Under Extreme or Non-Classical Conditions, R. van Eldik, and C.D. Hubbard, Eds., Wiley, New York, 1997, pp. 273-316. Y. Ikushima, N. Saito, K. Hatakeda, and S. Ito, in Supercritical Fluid Processing of Food and Biomaterials, H. Rizvi, Ed., Blackie Academic & Professional, London, 1994, pp. 244-254. M.H.M. Caralp, A.A. Clifford, and S.E. Coleby, in Extraction of Natural Products Using Near-Critical Solvents, M.B. King, and T.R. Bott, Eds., Blackie Academic & Professional, Glasgow, 1993, pp. 50-83. A. Akgerman, in Supercritical Fluids, Extraction and Pollution Prevention, M.A. Abraham, and A.K. Sunol Eds., American Chemical Society, Washington, DC, 1997, pp. 208-231. B.Mierau, and A.K. Sunol, in High Pressure Chemical Engineering, Ph. R. von Ruhr, and Ch. Trepp Eds., Elsevier, 1996, Amsterdam, pp. 321-326. I. Flament, U. Keller, and L. Wunsche, in Supercritical Fluid Processing of Food and Biomaterials, H. Rizvi, Ed., Blackie Academic & Professional, London, 1994, pp. 62-74. B. Subramanian, and M.A. McHugh, , Ind. Eng. Chem. Res., 37, 4203 (1998). P.E. Savage, S. Gopalan, T.I. Mizan, C.J. Martino, and E.E. Brock, AIChE J., 41, 1723 (1995). A.A. Clifford, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 449-479. B.C. Wu, S.C. Paspek, M.T. Klein, and C. LaMarka, in Supercritical Fluid Technology, T.J. Bruno, and J.F. Ely, Eds., CRC Press, London, 1991, pp. 511-524. G.P. Jessop, T. Ikariya, and R. Noyori, Chem. Rev., 99, 474 (1999). P.E. Savage, Chem. Rev., 99, 603 (1999).
21.2 Ionic liquids
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102
103 104 105 106
1459
M. Poliakoff, M.W. George, and S.M. Howdle, in Chemistry Under Extreme or Non-Classical Conditions, R. van Eldik, and C.D. Hubbard, Eds., Wiley, New York, 1997, pp. 189-218. Q. Ji, E.M. Eyring, R. van Eldik, K.P. Johnston, S.R. Goates, M.L. Lee, J. Phys. Chem., 99, 13461 (1995). C.B. Roberts, J.F. Brennecke, and J.E. Chateauneuf, AIChE J., 41, 1306 (1995). A. Baiker, Chem. Rev., 99, 443 (1999). Z.Y. Ding, M.A. Frisch, L. Li, and E.F. Gloyna, Ind. Eng. Chem. Res., 35, 3257 (1996). J.R. Kershaw, J. Supercrit. Fluids, 2, 35 (1989). A.K. Sunol, and G.H. Beyer, Ind. Eng. Chem. Res., 29, 842 (1990). J.R. Vick Roy, and A.O. Converse, in Supercritical Fluid Technology, J.M.L. Penniger, M. Radosz, M.A. McHugh, and V.J. Krukonis, Eds., Elsevier, Amsterdam, 1985, pp. 397-414. K.A. Sunol, US Patent 5,041,192. K. Nakamura, in Supercritical Fluid Processing of Food and Biomaterials, S.S,H. Rizvi, Ed., Blackie Academic & Professional, London, 1994, pp. 54-61. A.J. Mesiano, E.J. Beckman, and A.J. Russell, Chem. Rev., 99, 623 (1999). J.L. Kendall, D.A. Canelas, J.L. Young, and J.M. DeSimone, Chem. Rev., 99, 543 (1999). G. Sirinivasan, J.R. Elliot, Ind. Eng. Chem. Res., 31, 1414 (1992). P.G. Debedenetti, in Supercritical Fluids, Fundamentals for Application, E. Kiran, and J.M.H. Levent Sengers, Eds., Kluver Academic Publishers, Dordrecht, 1994, pp. 719-729. A.K. Sunol, J. Kosky, M. Murphy, E. Hansen, J. Jones, B. Mierau, S. G. Sunol, in Proceedings of the 5th International Symposium on Supercritical Fluids, Tome 1, ISAFS , Nice, 1998. K.A. Sunol, US Patent 4,992,308. S.S. Kistler, J. Physical Chemistry, 36 (1932) 52. W.J. Schmitt, R.A. Grieger-Block, T.W. Chapman, in Chemical Engineering at Supercritical Fluid Conditions, M.E. Paulatis, J.M.L. Penninger, R.D. Gray, and P. Davidson, Eds., Ann Arbor Science, Ann Arbor, MI, 1983, p.445-460. J.G. Van Lierop, A. Huitzing, W.C.P.M. Meerman and C.A.M. Mulder, J. Non-Crystalline Solids, 82, 265 (1986). R.J. Ayen, P.A. Iaobucci, Reviews in Chemical Engineering, 5, 157 (1988). B. Rangarajan, C.R. Lira, J. Supercritical Fluids, 4, 1 (1991). S.G. Sunol, A.K. Sunol, O.Keskin, O. Guney, Innovations in Supercritical Fluids Science and Technology, K. W. Hutchenson and N. R. Foster Eds., American Chemical Society, Washington, DC, 1995, pp. 258-268.
21.2 IONIC LIQUIDS D.W. Rooney, K.R. Seddon School of Chemistry, The Queen’s University of Belfast Belfast, Northern Ireland
21.2.1 INTRODUCTION The first question one would ask is “What is an ionic liquid?” Ionic liquids can be described, in the crudest terms, as room temperature molten salts. The term “ionic liquid” could therefore be applied to all molten salt systems such as cryolite (Na3AlF6) used in aluminum production or even molten table salt (NaCl). However the use of the word “molten” conjures up images of high temperature processes which are highly corrosive and difficult to design. In contrast to this, the ionic liquids discussed here are generally benign solvents which can be applied to a significant number of industrial processes leading to enhanced yields, greater recyclability and processes with an overall reduced environmental impact. Therefore an ionic liquid is normally described as a molten salts which is fluid at room temperature, or close to room temperature (salts melting below 100oC are often considered in this category). One only has to look at recent patent publications to discover that these solvents are finding
1460
D.W. Rooney, K.R. Seddon
application in commercial sectors as diverse as the nuclear industry,1,2 pharmaceuticals and fine chemicals,3,4 as well as in mainstream petrochemical processes.5-7 This contrasts with the unique perspective of Takahashi et al. who state that the two major applications are batteries and electrolytes, clearly unaware of the modern literature.8 In a number of indicated cases, these processes have been taken through the development process to a point of industrial commercialization and represent first generation ionic liquid processes, principally based on chloroaluminate(III) ionic liquids which are currently ready for industrial uptake. Following this line, second generation ionic liquid processes based on other, more benign, ionic liquids are currently under investigation and development in a variety of laboratories around the world. Many of these processes utilize the ability of a range of ionic liquids to selectively immobilize transition metal catalysts for liquid-liquid two-phase catalysis while permitting easy, often trivial, extraction of products.9-14 So why have these solvents not been used to rapidly replace the current volatile organics currently found in industry? There are a number of answers to this, but the most significant reason is economics. By looking through any chemical catalogue, it is obvious that 1-methylimidizole (a precursor for the manufacture of ionic liquids) is considerably more expensive than standard solvents. By the time this compound is processed to the final ionic liquid product, its cost will have increased to many times that of normal solvents. This is not helped by the fact that at present there is a very limited market for these compounds, keeping retail prices high. It is envisaged that if ionic liquid technology does become widely accepted, then the cost of production will decrease rapidly. Other quaternary ammonium salts and those that are based around pyridinium are cheaper alternatives (principally due to the scale of manufacture), but economics is not the only problem. As yet, there is a significant deficit of raw physical property data for engineers to use when designing new processes or retrofitting old plants. For the efficient design of any new industrial process incorporating ionic liquid technology, a complete understanding of the behavior of the solvent during operation is necessary. Physical properties such as viscosity, density, heat capacity and surface tension are all important during these early design stages. Others like electrochemical windows and electrical conductivities, will be important for more specific applications. With the advent of computers, chemical engineers have been able to use powerful process simulation software packages to estimate how a particular process will behave under certain operating conditions. These software packages, like the manual calculations which preceded them, predict the physical properties of organic solvents by using a number of empirical and semi-empirical equations which are available. Unfortunately these equations where developed for molecular compounds and tend to require the critical temperature and pressure data, information which does not apply to ionic liquids. In addition, other techniques for predicting physical properties like surface tension involve using group contriFigure 21.2.1. Showing the relationship between lattice bution methods, but again these fail to acenergy26 and melting point27 for the Group 1 halide count for organic salts and therefore cannot salts.
21.2 Ionic liquids
1461
be used. This lack of predictive power and the overall general lack of physical property data in the literature will inevitably slow the transfer of ionic liquid technology into industrial processes. 21.2.2 FUNDAMENTAL PRINCIPLES OF THE FORMATION OF ROOM TEMPERATURE IONIC LIQUIDS A number of detailed reviews and articles have been published recently on the theory and development behind ionic liquid technology.15-25 However in order to develop a greater understanding and appreciation of these solvents, we will summarize the main points here. Therefore the purpose of this section is to give an introduction to the nature and properties of room-temperature ionic liquids, with particular emphasis being placed upon their potential as solvents for industrially relevant catalytic reactions, and (more generally) for clean technology. 21.2.2.1 Development of ionic liquids To begin with, the melting point of a salt is related to its lattice energy. In fact, if one was to plot the lattice energy of a series of salts, for instance the Group 1 halides, against the melting points in Kelvin, then one can see that there is reasonable linearity between the melting point and the lattice energy (see Figure 21.2.1). Although not particularly accurate, this simple approach will give an estimated melting point of the Group 1 salts if the lattice energy is known. The deviations from this treatment are frequent and are usually due to other forms of bonding within the structure, as shall be explained later. The first theoretical treatment of lattice energy began with Born and Landé, and was then further developed by Kapustinskii28 into the what is know as the “Kapustinskii equation” (eqn. [21.2.1]): U=
287.2vZ + Z − r0
0.345 1 − r0
[21.2.1]
where: U v r0 Z+, Z-
lattice energy number of ions per molecule sum of ionic radii charge of the ionic species
Therefore by increasing the value of r0, i.e., by using larger anionic and cationic components in the salt, it is possible to lower this energy and therefore reduce the melting point. This effect is shown Table 21.2.1. Table 21.2.1. Melting points27 of various inorganic salts, melting points given in ºC Anion
Lithium
Sodium
Potassium
Rubidium
Caesium
Fluoride
842
988
846
775
682
Chloride
614
801
776
718
645
Bromide
550
747
734
693
636
Iodide
450
661
686
647
626
From this one can clearly see that as we increase both the size of the anion and the cation, the melting point decreases. From the Kapustinskii equation one must also note that by
1462
D.W. Rooney, K.R. Seddon
increasing the ionic charge will tend to increase the lattice energy of the crystal. However the effect on melting point is complicated by the fact that according to Fajans’ rules an increasing charge also results in increasing covalency particularly for small cations and large anions. The effect of size can even be observed in the melting points of the lithium halide series, where the small size of the lithium ion leads to a greater covalent character in the lithium halide Figure 21.2.2. Showing the relationship between the bond, reducing the melting point when comsize of similar anions containing halides and melting pared to the other Group 1 salts. point of various sodium salts. For a given cation, and singly charged cations and ions, equation [21.2.1] demonstrates that the lattice energy will become only be a function of r0, and since r0 is the sum of r+ and r-, then the lattice energy is only dependant on anion size. The radius for simple anions can be found in a number of different sources.27,29 However for more complex anions it becomes more difficult to assign unique values to the radii. Yatsimirskii30 has shown that it is possible to determine the apparent values of the ionic radii of these ions indirectly from the lattice energy of the compounds containing them. These values are generally referred to as the thermochemical radii and a number are shown in Table 21.2.2. In most cases the fact that ions such as [CNS]- and [CH3COO]-, are markedly non-spherical makes these radii of limited use. However for the case of tetrahedral ions, the symmetry is sufficiently high enough for comparison purposes. If one were to plot the melting points a number of sodium salts against the opposing anions’ thermochemical radius a clear relationship is observed (see Figure 21.2.2) Table 21.2.2. Anionic and thermochemical radii Name
Anion
r, pm
31
[FeCl4]
358
Tetrachloroborate31
[BCl4]-
310
Tetrachloroaluminate31
[AlCl4]-
Tetrachlorogallate31 Trifluoromethylsulfonate32
Tetrachloroferrate
Hexafluoroarsenate
32
Hexafluorophosphate Periodate
33
Perchlorate
-
[ClO3]
201
Thiocyanate33
[CNS]-
195
295
Borate33
[BiO4]-
191
[GaCl4]-
289
Bromate33
[BrO3]-
191
[CF3SO3]-
267
Nitrate33
[NO3]-
188
259
-
[PF6]
253
-
[IO4] 33
Permanganate
249 -
[ClO4] 33
Tetrafluoroborate
236 -
[MnO4] 33
Trinitrophenoxide
33
r, pm
Chlorate
[AsF6]
-
[BF4]
[C6H2(NO2)3O]
-
33
Anion -
-
32
Name
33
Cyanide 33
Cyanate
33
Formate
33
229
33
222
[IO3] 33
Acetate Nitrite
33
Amide
181
-
Iodate
240
[CN]
-
181
[CNO]
-
[HCOO]
160 -
160 -
[CH3CO2]
156
-
156
-
129
[NO2] [NH2]
21.2 Ionic liquids
1463
Table 21.2.3. Some melting points for various [C2-mim] salts Melting point, oC
Salt [C2-mim]Cl35 [C2-mim]Br [C2-mim]I
84
36,37
81
36,37
79-81 38
[C2-mim][PF6]
62
39
[C2-mim][NO2]
55 40
[C2-mim][GaCl4]
47
39
38
[C2-mim][NO3] Figure 21.2.3. The aromatic heterocyclic N-butylpyridinium, 1-ethyl methylimidazolium cations, and general structural information for [NCn-py]+ and [Cn-mim]+ based ionic liquids.
35
[C2-mim][AlCl4]
7
[C2-mim][(CF3SO2)2N]
38
-15
It should be noted that if one plotted the melting points and radii of the planar anions in Figure 21.2.2, they would not fall on the trend line. Since our desired goal is to produce a salt which melts at or around room temperature, we extrapolate that we would require an anion with a radius of at least 400 pm. The number of such large anions is limited, and they also have a relatively high degree of covalence associated with the structure. However, as clear from above, it is also possible to reduce the lattice energy by increasing the size of the cation. By moving to organic salts the effect can be demonstrated more easily. For example the organic salt [EtNH3][NO3] was shown in 1914 to have a melting point of 12oC and was hence the first room temperature ionic liquid.34 This concept can be taken further and applied to other organic systems, such as, the cations “butylpyridinium” and “1-ethyl 3-methylimidazolium” ([NC4-py]+ [C2-mim]+), (see Figure 21.2.3). For the purpose of clarity we will simply refer to the length of the alkyl chain by the number of carbon atoms in that chain, hence ethyl is represented by C2, butyl by C4, etc. In the case of the [C2-mim]+ cation it can be observed that it has the lowest possible symmetry making it even more difficult for a crystal to form. Given this argument one would expect the melting point of salts based around this cation to be low. Indeed Wilkes et al.35 have shown that the melting point for [C2-mim]Cl is 84oC significantly lower than any of the Group 1 halides. The melting points of a number of [C2-mim]+ salts have been reported over recent years, and some of these are summarized in Table 21.2.3. In order to develop a greater understanding of these salts, the nature of the [C2-mim]+ cation and how it interacts with various anions has been explored in both solution41 and solid42 phases and via theoretical studies.43 It is now possible to produce a similar plot to Figure 21.2.2 showing the effect of anion on the melting point of the [C2-mim]+ salts using the data in Table 21.2.3. The [AlCl4]- anomaly can be partly explained by the fact that in the ionic liquid an equilibrium existing between the [AlCl4]- anion and the larger [Al2Cl7]- anion, is expected to be much higher than for the [GaCl4]- ionic liquid. In addition results from our own research
1464
D.W. Rooney, K.R. Seddon
Figure 21.2.4. Relationship between the size of similar anions containing halides and the melting point of their [C2-mim] salts.
Figure 21.2.5. Phase diagram for the [Cn-mim]Cl ionic liquids.44
Figure 21.2.6. Melting point phase diagram for [Cn-mim][PF6] ionic liquids as a function of alkyl chain length n showing the melting transitions from crystalline (closed square) and glassy (open square) materials and the clearing transition (circle) of the liquid crystalline (LC) terms.
Figure 21.2.7. Melting point phase diagram for [Cn-mim][BF4] ionic liquids as a function of alkyl chain length n showing the melting transitions from crystalline (closed square) and glassy (open square) materials and the clearing transition (circle) of the liquid crystalline terms.
show that the melting point for the [FeCl4]- ionic liquid does appear to follow the given trend. In order to expand the number of potential ionic liquids the possibility also exists to increase the length of the alkyl group thereby further decreasing the lattice energy. For the [Cn-mim] chlorides the result is a range of salts all of which have low melting points (some of which are below room temperature) as shown in Figure 21.2.5, though often at the cost of increased viscosity. One can observe that there is a maximum chain length allowed before other forms of bonding begin to dominate and the melting point increases. The melting points indicated below the 0oC line are in fact glass transition temperatures rather than true melting points. As stated above, one point to note is that the viscosity of these room temperature “chloride” liquids is very high and they would therefore have to be used at relatively high temperatures if required in a process. Figures 21.2.6 and 21.2.7 illustrate the significant variation in melting point which can be induced by simply changing the anion from [Cn-mim][Cl] to the [Cn-mim][PF6] and
21.2 Ionic liquids
1465
Table 21.2.4 Selected melting points of the Group 1 chloride mixtures48
Table 21.2.5. The melting points of selected tetrachloroaluminate(III) salts48 Mole ratio
m.p., °C
LiCl-AlCl3
35:65
80
368
LiCl-AlCl3
50:50
132
25:75
551
NaCl-AlCl3
39:61
108
LiCl-CsCl
60:40
355
NaCl-AlCl3
50:50
151
NaCl-KCl
50:50
658
RbCl-AlCl3
30:70
148
CsCl-KCl
35:65
610
RbCl-AlCl3
50:50
336
System
Mole ratio
m.p., °C
LiCl-LiF
70:30
496
LiCl-LiI
35:65
LiCl-NaCl
System
[Cn-mim][BF4] ionic liquids.45-47 This procedure has produced a lower melting point salt for example the [C4-mim][PF6] has a melting point of 5oC whereas the [C4-mim][Cl] has a melting point of 80oC. These lower melting point liquids with the shorter alkyl chains lead to a much more fluid and easily managed liquid. An interesting feature of these phase diagrams is the appearance of liquid crystalline phases with the longer alkyl chains, and this is confirmed when their optical textures are examined.35 The implication of the existence of these stable phases has still to be explored in terms of stereochemical control of reactions. From these figures it can be seen that those salts with short alkyl chains (n=2-10) are isotropic ionic liquids at room temperature and exhibit a wide liquid range, whereas the longer chain analogues are low melting mesomorphic crystalline solids which display an enantiotropic smectic A mesophase. The thermal range of the mesophase increases with increasing chain length and in the case of the longest chain salt prepared, [C18-mim][BF4], the mesophase range is ca. 150oC. 21.2.2.2 Binary ionic liquid systems In Table 21.2.1 it was shown that the melting points of the Group 1 salts are significantly above room temperature, and far too high to form a generic medium for reactive chemistry. However by both increasing the size of the cation and the anion it has become possible to produce salts that are liquid at room temperature. It is well known that mixing together different salts deforms the crystal structure, leading to a lower lattice energy, and hence a lower melting point. At certain concentrations, referred to as eutectic points, the melting point has reached its minimum. This effect on melting point, obtained by combining the Group 1 chlorides into various mixtures can be seen in Table 21.2.4. The melting points of the simple tetrachloroaluminate(III) salts of both the sodium and [C2-mim] cations have been shown to be significantly lower than their respective chloride salts, indeed [C2-mim][AlCl4] was shown to be a liquid at room temperature. Such salts are produced by combining equimolar quantities of either NaCl or [C2-mim]Cl with AlCl3. However, as can be seen from Table 21.2.5, the 50:50 mole ratio does not usually correspond to the lowest melting point. For the inorganic salts we now have melting points which are in the maximum range of some of the high-boiling organic solvents (e.g., 1,4-dichlorobenzene, b.p. 174oC). Hence given that the 50:50 mole ratio does not correspond to the lowest melting point of the inorganic salts, one would expect the same to be true for the organic cations. Figures 21.2.8 and
1466
D.W. Rooney, K.R. Seddon
21.2.9 show the phase diagrams of [C2-mim]Cl when combined with varying quantities of AlCl3 and GaCl3. Thus a large number of ambient temperature ionic liquids can be formed from a mixture of solid aluminium(III) chloride and solid 1-ethyl-3-methylimidazolium chloride. Mixing these two salts together results in an exothermic reaction and generation of a clear, colorless liquid, and an ambient temperature molten salt with low Figure 21.2.8. Phase diagram for the [C2-mim]Cl-AlCl3 system.35 viscosity. We can see that depending on the proportions of the two components it is possible to obtain eutectics with melting points as low as -96°C.35 These liquids are also thermally stable in excess of 200oC allowing for a tremendous liquidus range. When the molar proportions are equal, the system is a neutral stoichiometric salt 1-ethyl-3-methylimidazolium tetrachloroaluminate(III) chloride [C2-mim][AlCl4] which melts at its congruent melting point of about 7°C. The lowest melting point is achieved when the molar Figure 21.2.9. Phase diagram for the [C2-mim]Cl-GaCl3 system.40 ratio of the system is 1:2 [C2-mim]Cl:AlCl3. Qualitatively these phase diagrams are similar to that reported by Hurley and Weir for the 1-ethylpyridinium bromide-AlCl3 binaries49 in that they all show a characteristic “W” shape. It has been shown that if the butyl and ethyl groups are exchanged for a generic linear alkyl function then a series of cations can be generated. These can then be balanced against any one of a number of opposing anions to produce a final salt. Given that the range of available anions and cations has expanded enormously in the past decade it is possible to produce a number of combinations. Indeed, it is our best estimate that, if binary and ternary mixtures are included (and there are very good practical and economic reasons for doing that), there are approximately one trillion (1018) accessible room temperature ionic liquids.50 21.2.3 CATALYSIS IN IONIC LIQUIDS The wide range of reactions that have been undertaken in low temperature ionic liquid solvents is quite remarkable. It is limited simply by ones imagination. The specific and tuneable solvent properties of ionic liquids are a key feature for their use as solvents and have
21.2 Ionic liquids
1467
been utilized, especially in combination with the catalytic properties of the chloroaluminate(III) ionic liquids, to develop a range of synthetically important catalytic reactions. Currently a number of these are being investigated as economically and environmentally viable alternatives to existing industrial processes. A number of reactions are summarized below; for a more detailed discussion in this area readers are recommended to the recent review by Holbrey and Seddon,25 and to a forthcoming book in the NATO ARW series. 21.2.3.1 Reactions involving first generation chloroaluminate(III) ionic liquids The chloroaluminate ionic liquid mixtures are governed by the following primary equilibrium, equation:51 2[AlCl4]- ↔ [Al2Cl7]- + Cl-; k ≈ 10-16
[21.2.2]
This is an acid base equilibrium under the Franklin definition. The [Al2Cl7]- species is the “acid” and the Cl- is the base. Note that this is an aprotic equilibrium. Therefore if the mole ratio of [C2-mim]Cl:AlCl3 is greater than, less than, or exactly equal to 50:50, the solvent behavior can be described as Franklin basic, Franklin acidic, or neutral. Considering that aluminum trichloride is a very important commercial catalyst with over 25,000 tonnes produced annually in the USA alone, such liquids containing aluminum trichloride and allowing for differing levels of acidity have been extensively studied as first generation ionic catalytic solvents in a wide variety of synthetic and catalytic processes. Ionic liquids could therefore be used as substitutes for conventional solid or suspended sources of aluminum(III) chloride. As liquid phase catalysts, they allow for tremendous control of reactor inventories and can be cleaned and recycled with ease. Therefore, ionic liquids, in ideal cases, have no waste associated with them, whereas the supported aluminum(III) chloride catalysts will require large (and annually rising) waste disposal costs. An industrial example of the use of chloroaluminate ionic liquids in alkene catalysts is the recent development of the IFP Difasol process52 which is widely used industrially for alkene dimerization (typically propene and butanes). It was observed by Chauvin and coworkers53-60 that chloroaluminate(III) ionic liquids would be good solvents for the nickel catalyst used in the reaction, and discovered that by using a ternary ionic liquid system ([C4-mim]Cl-AlCl3-EtAlCl2) it was possible to form the active catalyst from a NiCl2L2 precursor and that, the ionic liquid solvent stabilized the active nickel species. Overall they found that the nickel catalyst remains selectively dissolved in the ionic liquid phase to the extent that over 250 kg of propene could be dimerized per 1 g of nickel catalyst. In addition the product was insoluble in the ionic liquid, which made product recovery facile. Other studies have shown that a wide range of acidic chloroaluminate(III) and alkylchloroaluminate(III) ionic liquids can be used to catalyze the dimerization and oligomerization of olefins.61,62 In the reaction, the olefinic feedstock may be mixed with, or simply bubbled through, the ionic liquid catalyst to produce oligomeric products. A significant outcome of this reaction is that the product has a low solubility in the ionic liquid and separates as a less dense organic phase which is readily removed. In addition to the above reactions it was shown that isobutene can be polymerized in an acidic ionic liquid to polyisobutene with a higher molecular weight than is formed using
1468
D.W. Rooney, K.R. Seddon
other polymerization processes. Polyisobutene, traditionally prepared by the Cosden process, is a valuable lubricant, and also a route to higher value-added materials. In general it was observed that the catalytic activity of the ionic liquids increases towards higher degrees of polymerization from short-chain oligomers as the alkylchain length of the 1-alkyl-3-methylimidazolium or N-alkylpyridinium cation is increased.63 The ionic liquid process has a number of significant advantages over the industrial Cosden process. This system uses a supported or liquid phase aluminum(III) chloride catalyst.63 Using the ionic liquid process, the polymer forms a separate layer, which is substantially free of catalyst and ionic liquid solvent. This effect greatly enhances the degree of control available to reduce undesirable secondary reactions (i.e., isomerization) without requiring alkali quenching of the reaction. In addition Ziegler-Natta polymerization reactions have also shown some success when carried out in ionic liquids.64 The most common production methods for this form of polymerization involve the use of triethylaluminium catalysts at ca. 100°C and 100 atmospheres pressure. Advances have been developed through the use of organometallic transition metal catalysts, typically nickel or titanium. Given the solvent characteristics of ionic liquids it should be possible to effectively immobilize such catalysts in an ionic liquid solvent. Indeed, Carlin and Wilkes64 have reported the Ziegler-Natta polymerization of ethene in an ionic liquid solvent. In these reactions an acidic [C2-mim]Cl-AlCl3 ionic liquid solvent was used to support dichlorobis(η5-cyclopentadienyl)titanium(IV) with an alkylchloroaluminium(III) co-catalyst. Electrophilic substitution65 and other reactions of naphthalenes (alkylation, acylation, condensation and migration in acidic ionic liquids66,67 have been reported. Anthracene undergoes photochemical [4+4] cycloaddition reactions68,69 in acidic chloroaluminate(III) ionic liquids. One interesting study included a one-pot synthesis of anthraquinone from benzene giving a 94% yield. In general a much wider range of redox products are formed than occur in conventional solvents; the strong Brønsted acidity of the ionic liquid induces protonation of anthracene, by residual traces of HCl, to form an anthracenium species which couples readily via photochemically driven electron transfer mechanisms. Both the Friedel-Crafts alkylations and acylations are of great importance to the fine chemical and pharmaceutical industries. Typically, these reactions are run in an inert solvent with suspended or dissolved aluminum(III) chloride as a catalyst, and may take six hours and go only to 80% completion giving a mixture of isomeric products. In addition, there are a number of problems, especially with misnamed “catalytic” Friedel-Crafts acylation reactions, which are actually stoichiometric, consuming 1 mole of AlCl3 per mole of reactant. Annually, massive quantities of aluminum(III) chloride are consumed in these reactions causing a number environmental problems through waste disposal. Both alkylation and acylation reactions under Friedel-Crafts conditions have been demonstrated using chloroaluminate(III) ionic liquids as both solvent and catalysts.66-77 Here it has been shown that reaction rates are much faster with total reagent conversion and often with surprising specificity to a single product. Boon et al. have reported the alkylation of benzene with a wide number of alkyl halides in acidic chloroaluminate(III) ionic liquids73 and general organic reactions in low melting chloroaluminate ionic liquids have also been described,75,78 which include chlorinations and nitrations in acidic ionic liquids.72,73,76
21.2 Ionic liquids
1469
One specific example is the alkylation of benzene with chloroethane which gives a mixture of mono- to hexa-substituted products. The ionic liquid solvent/catalyst activates the reaction and the alkylation can be performed even at temperatures as low as -20°C in the ionic liquid solvent. Again it was shown that the products have a low solubility in the ionic liquid leading to facile separation. Many acylation reactions have been demonstrated in acidic chloroaluminate(III) ionic liquids liquids.70,74 However as described above these processes are essentially non-catalytic in aluminum(III) chloride which necessitates destroying the ionic liquid catalyst by quenching with water to extract the products. However, regioselectivity and reaction rates observed from acylation reactions in ionic liquids are equal to the best published results. The Friedel-Crafts acylation of benzene has been shown to be promoted by acidic chloroaluminate(III) ionic liquids.74 It was observed that the acylated products of these reactions have high selectivities to a single isomer for example toluene, chlorobenzene and anisole are acylated in the 4-position with 98% specificity. Naphthalene is acylated in the 1-position which is the thermodynamically unfavored product. In addition to benzene and other simple aromatic rings, a range of organic and organometallic substrates (e.g., ferrocene)79,80 have been acylated in acidic chloroaluminate(III) ionic liquids. 21.2.3.2 Reactions in neutral or second generation ionic liquids Acidic [C2-mim]Cl-AlCl3 mixtures have been shown above to be very useful catalytic solvents for a number of industrially relevant reactions. However, reactions catalyzed by aluminum trichloride, whether used in an ionic liquid or as a solid phase reactions have one significant disadvantage, they are air and moisture sensitive. Using such liquids within an industrial process will therefore require that reactants are kept as dry as possible, adding to the overall process cost. We have seen how the activity and properties of the liquid can be readily controlled by both changing the anion or the cations present. By switching to neutral ionic liquids containing for example [BF4]-, [PF6]- and [SbF6]- anions, the reactive polymerization and oligomerization reactions of olefins catalyzed by acidic anions is not observed and more controlled, specific reactions can be catalyzed. Since these ionic liquids cannot support the existence of reactive Lewis acid conjugate anions (such as [Al2Cl7]- i.e. there is no analogous mechanism to support 2[BF4]- ↔ [B2F7]+ F-), they are much less reactive when used as solvents. Under these conditions, many conventional transition metal catalysts can be utilized. Again modification of the ionic liquid solvents allows the potential to immobilize catalysts, stabilizing the active species and to optimize reactant/product solubilities and to permit facile extraction of products. In general it has been found that charged, especially cationic transition metal complexes are most effectively “immobilized” in the ionic liquid solvents. In addition ionic liquids can immobilize less complex catalysts than two-phase aqueous-organic systems, where expensive, unstable, synthetically challenging ligands are often required. Hydrogenation of olefins catalyzed by transition metal complexes dissolved in ionic liquid solvents have been reported using rhodium-,81 and ruthenium- and cobalt-containing catalysts.82 In these studies it was shown that hydrogenation rates where up to five times higher than the comparable reactions when carried out in propanone. The solubilities of the alkene reagents, TOFs, and product distributions where all strongly influenced by the nature of the anion in the ionic liquid solvent.
1470
D.W. Rooney, K.R. Seddon
Chauvin et al. reported the asymmetric hydrogenation of acetamidocinnamic acid83 to (S)-phenylalanine with a cationic chiral rhodium catalyst in [C4-mim][SbF6] ionic liquid, more recently the 2-arylacrylic acid has been produced with a reasonable 64% yield84 using a chiral ruthenium catalysts in [C4-mim][BF4] ionic liquids. Palladium catalysts85 immobilized in an ionic liquid-polymer gel membrane86 containing either [C2-mim][CF3SO3] or [C2-mim][BF4] have also been reported as catalysts for heterogeneous hydrogenation reactions. Fuller et al have also reported the hydroformylation86 of pent-1-ene in [C4-mim][PF6] using a rhodium catalysts showing both a high catalytic activity and product separation as a second organic phase. However it was observed that a small quantity of the neutral catalyst leached into the organic phase. Other commercially important reactions include the hydrodimerization of 1,3-butadiene to octa-2,7-dien-1-ol87,88 carried out using palladium catalysts in [C4-mim][BF4]. The catalyst precursor [Pd(mim)2Cl2] was prepared in situ from an imidazolium tetrachloropalladate(II) salt, [C4-mim]2[PdCl4], dissolved in the ionic liquid solvent. The reaction proceeds in a liquid-liquid two-phase system, where the products separate from the catalytic reaction mixture as a separate layer on cooling. Ionic liquids have been demonstrated as effective solvents for Diels-Alder reactions89-91 where they have shown significant rate enhancements as well as high yields and selectivities when compared with the best results obtained in conventional solvents. To date, the biggest developments in Diels-Alder chemistry have come through reactions in Li[ClO4]-Et2O, where the high electrolyte concentrations are cited as beneficial through “salt-effects” and the high internal pressure of the solvent. However the use of such mixtures of perchlorate salts with organic molecules could cause a number of hazards when used on an industrial scale. Hence there is considerable potential for ionic liquids in this area. The stability of the neutral ionic liquids allows them to be used in environments unsuitable to the tetrachloroaluminate(III) based ionic liquids. As such they offer considerable advantages allowing them to be used with “wet” process streams. The use of the fluorinated anions as shown in Figures 21.2.6 and 21.2.7, i.e., [PF6] and [BF4], introduces the concept of hydrophilic ionic liquids which are partially immiscible with water. In fact all [PF6] liquids are found to be immiscible with water and all [BF4] liquids with chain lengths greater than C4 will form two separate phases with water at sufficient concentrations. This property has prompted investigation into the application of these solvents as extraction solvents for a number of materials. Rogers et al.92 has studied the partition of benzene with water and has reFigure 21.2.10. The phase preference of the three forms cently studied the relationship between pH and extraction efficiency.93 Figure 21.2.10 ilof thymol blue in [C4-mim][PF6].93
21.2 Ionic liquids
1471
lustrates the qualitative partitioning of thymol blue in its three forms between aqueous (top) and [C4-mim][PF6] (bottom) phases, as aqueous phase pH is changed from very acidic, to very basic. At low pH the thymol blue exists in its red form as a neutral zwitterion which prefers the ionic liquid phase. As the pH is increased via the addition of NaOH, the yellow monoanion forms with some detectable increase in concentration in the aqueous phase. The blue dianion, above pH = 10, partitions quantitatively to the aqueous phase. The same result was also obtained by bubbling CO2 and NH3 through the ionic liquid phase. Lye94 has also studied the use of ionic liquids in the extraction of erythromycin-A for the Rhodococcus R312 catalysed biotransformation of 1,3-dicyanobenzene (1,3-DCB) in a liquid-liquid two-phase system. Here it was found that the ionic solvent was less harmful to the system than conventional molecular solvents with the specific activity of the biocatalyst in the water-[C4-mim][PF6] system being almost an order of magnitude greater than in the water-toluene system. In addition, recent reports have effectively shown the potential for combining ionic liquid-and supercritical CO2 systems for product extraction and separation95 of naphthalene. Here it was found that the CO2 was highly soluble in [C4-mim][PF6] reaching a mole fraction of 0.6 at 8 MPa and after separation there was no detectable ionic liquid in the extract. Another recent example is a paper by Armstrong et al.,96 where ionic liquids ([C4-mim][PF6] and [C4-mim][BF4]) where used as stationary phases for gas chromatography. From these studies it was concluded that the ionic liquids act as nonpolar stationary phases when separating nonpolar analytes, however they are highly interactive and retentive when used to separate molecules with somewhat acidic or basic functional groups.
Figure 21.2.11. The electrochemical windows of a number of ionic liquids.97
1472
D.W. Rooney, K.R. Seddon
21.2.4 ELECTROCHEMICAL APPLICATIONS Since ionic liquid where first developed as an adjunct to the U.S. Naval program for use as battery electrolytes it seems only fair to begin this discussion by firstly looking at the electrochemical properties. One of the most important electrochemical properties of a solvent is its electrochemical window, the range of potentials over which a solvent system (including the solvent and any supporting electrolyte) is inert. The cathodic limit results from reduction of the solvent system; the anodic limit from its oxidation. Electrochemical windows are somewhat arbitrary as they are dependent upon the amount of background current which can be tolerated. Figure 21.2.11 shows the electrochemical windows of a range of ionic liquids as well as more commonly used solvents such as water and acetonitrile. The electrochemical windows shown in Figure 21.2.11 were determined using cyclic voltammetry at a glassy carbon disc electrode and are referenced against the Ag(I)/Ag couple (versus Ag(I)/Ag) and converted to the equivalent SCE potential. The reference electrode consisted of a silver wire immersed in a 0.1 M solution of AgNO3 dissolved in [C6-mim][NO3]. In chloroaluminate ionic liquids, the most common reference electrode is an aluminum wire immersed in neat acidic ionic liquid (versus Al(III)/Al). It can be seen from Figure 21.2.11 that the [C4-mim] cation has a cathodic limit of approximately -2 V versus SCE and that this value is essentially the same for all of the [Cn-mim] cations. Given that the deposition potentials for many metals will fall positive of this potential, it becomes possible to use ionic liquids as electrolytes for metal plating and other similar processes. The broad electrochemical windows (in some cases, over 4 V) indicate that a variety of organic and inorganic electrochemical oxidations and reduction should be possible in ionic liquids. A wide variety of metals can be electrodeposited from room-temperature ionic liquids. The electrodeposition mechanisms of transition metals, lanthanides and Group 13 metals have been investigated98-103 as have cobalt,104-107 iron,108 manganese,109 tin,110 gold,111,112 silver,113-116 palladium,117,118 mercury,119 cerium,120 and lead,121 and the actinides.122 In addition, ionic liquids have been used extensively to study the electrochemistry and spectrochemistry of metal-chloride complexes.123-131 Chloroaluminate ionic liquids have been used in the electrodeposition of aluminum and aluminum-transition metal alloys.132 Transition metal-aluminum alloys are valued for their corrosion resistance and magnetic properties. A convenient method for creating thin alloy films is through the electrodeposition of two or more metals. The electrodeposition of aluminum and aluminum alloys from aqueous solutions is complicated by the fact that dihydrogen is evolved before aluminum is Table 21.2.6. Diffusion coefficients and deposited. However, these materials can be Stokes-Einstein products of a number of electrodeposited from aprotic non-aqueous species studied in the 40-60 mol% solvents or high temperature molten salts. [C2-mim]Cl-AlCl3 ionic liquids at 40ºC The electrodeposition of aluminum alloys -7 2 -1 -1 -10 -2 -1 such as Cr-Al,133 Mn-Al,134,135 Ni-Al136 and Do 10 cm s ηDo T , 10 g cm s K Ti-Al137 have been demonstrated from inorCo(II) 6.6 2.6 ganic chloroaluminate molten salts, mainly Cu(I) 2 0.7 Franklin acidic (AlCl3-rich) AlCl3 - NaCl. One unfortunate drawback to the use of alNi(II) 10 3.5 kali-chloride based chloroaluminates such Zn(II) 6.7 2.3 as AlCl3-NaCl for the electroplating of
21.2 Ionic liquids
1473
these alloys is the substantial vapor pressure of Al2Cl6 associated with the acidic region of these melts. This is of course not the case when working with room temperature ionic liquid systems. Electrochemical and spectroscopic studies of cobalt,138 copper,139 and nickel,140 have been carried out in the AlCl3-butylpyridinium chloride molten salt system. The direct current and pulsed current electrodeposition of Ni-Al alloys has also been shown in acidic AlCl3-butylpyridinium chloride ionic liquids. This particular alloy has also been shown to be successful in AlCl3-[C2-mim]Cl as have Co-Al and Cu-Al.141 Electrochemical techniques can also be used to calculate the diffusion coefficients of metal ions. Table 21.2.6 shows the calculated diffusion coefficients and stokes-Einstein products of cobalt(II),104 copper(I),142 nickel(II)143 and zinc(II)144 in the 40-60 mol% [C2-mim]Cl-AlCl3 ionic liquid. 21.2.4.1 Electrosynthesis Electrosynthesis provides an attractive alternative to conventional methods used for performing synthetic chemistry. It can effect the clean, complete conversion of starting material to product without using hazardous or toxic experimental conditions. Ionic liquids possess many advantages over conventional solvents typically used in electrochemical experiments. Their polar nature allows them to dissolve large concentrations of a wide variety of organic and inorganic compounds. They possess no vapor pressure, are stable on heating and because they are completely ionized the need for a supporting electrolyte is eliminated. Such benefits are not associated with many of the non-aqueous systems employed in electrochemical applications. Electrochemical studies in ionic liquids are common;145-148 however the application of this technology to molecular synthesis remains largely unexplored. Some of the examples to date deal with electro-polymerizations. Osteryoung et al. prepared polyaniline on platinum and glassy carbon electrodes by anodic oxidation of the monomer in imidazolium based chloroaluminate ionic liquids.149 Elsewhere the formation of polyfluorene was achieved by Janiszewska and Osteryoung.150 Actual transformation of an organic moiety was achieved electrochemically in 1-methyl-3-butylimidazolium chloroaluminate systems. The reactions studied dealt with the reduction of aromatic ketones.151 Reductions of perylene and phenazine have also been achieved electrochemically in basic mixtures of aluminum trichloride and 1-ethyl-3-methylimidazolium chloride.152,153 The use of a neat ionic liquid 1-ethyl-3-methylimidazolium hydrogen dichloride [C2-mim][HCl2] was also investigated with regard to the reduction of phenazine.152 Most of the work in electrosynthesis has been carried out in air- and moisture-sensitive ionic liquids, those containing chloroaluminates and [HCl2]-; little investigation into the electrosynthetic possibilities presented by air and moisture stable ionic liquids has been attempted.154 21.2.5 PHYSICAL CHARACTERIZATION 21.2.5.1 Viscosity Viscosity is probably the most important physical property for initially determining the “processability” of a solvent. It is necessary for all calculations involving fluid flow, (pumping, mixing, etc.) as well as in the estimation of heat transfer and diffusion coefficients. Information on the change in viscosity as a function of temperature, solvent or reagent addition is required for the efficient and design of any handling equipment. Ideally one would like the viscosity of a fluid to be as low as possible allowing for the fluid to be
1474
D.W. Rooney, K.R. Seddon
pumped easily. In addition it is desired for the fluid to have only small changes in viscosity through the normal operating temperature range. In general, the published data on the viscosity of ionic liquids is scarce. Most of this published literature on ionic liquids viscosity deals with the first generation ionic liquids. The viscosity of any fluid is highly dependent on both the measuring technique used and the purity of the samples. Given this difficulty the reported values in the literature are often neither comparable or reproducible. Once again, the first studies of the viscosity where related to the development of ionic liquids as nonaqueous battery electrolytes. Hussey et al.155 reported the viscosity of several N-alkylpyridinium chloroaluminate salts over a temperature range of 25 to 75oC where they observed an increase in viscosity with alkyl chain length and concluded that the temperature dependence of the viscosity could be fitted to the Arrhenius type equation, i.e., η = η 0 exp(E η / RT )
[21.2.3]
where: Eη R T
is the energy of activation for viscous flow is the gas constant is the absolute temperature
This conclusion was supported by the linearity of the ln η vs. 1/T plots for the studied temperature range. The above equation can be modified to give; ln η = Aη + B η / T
[21.2.4]
where: Aη Bη
= ln η0 = Eη/R
The fitted parameters, along with the values of Eη, appear in Table 21.2.7. Table 21.2.7. Parameters for Arrhenius equations for viscosity Melt
Mol ratio
- Aη
Eη
[NC1-py]Cl -AlCl3
1:2
4.801
4.639
[NC2-py]Cl -AlCl3
1:2
4.355
4.276
[NC3-py]Cl -AlCl3
1:2
4.734
4.523
[NC4-py]Cl -AlCl3
1:2
5.314
4.950
[NC2-py]Br -AlCl3
1:2
5.575
5.136
The viscosities of the 1,3-dialkylimidazoilium aluminium chloride156 and 1-methyl-3-ethylimidazolium aluminium bromide157 ionic liquids have also been reported for different compositions and temperatures. For both the chloroaluminate and bromoaluminate ionic liquids the temperature dependence was found not to have an Arrhenious type curve, with non-linear plots of ln η vs. 1/T. In these studies the temperature range used was wider than that of the N-alkylpyridinium. This non-Arrhenius behavior is characteristic of glass forming melts. Here the three parameter Vogel-Tammann-Fulcher (VFT) equation:
21.2 Ionic liquids
1475
ln η = K η / (T − T0 ) +
1
2
lnT + ln Aη
[21.2.5]
where: Aη T0 Kη
is a scaling factor is the “ideal transition temperature” and is a constant characteristic of the material
can be used for such fluids, with Kη being the formal analogue of the Arrhenius activation energy. The parameter T0 has been given a quasi-theoretical significance by the free volume theory of Cohen and Turnbull. According to this theory, T0 represents the temperature at which the free volume of the liquid disappears and liquid transport becomes impossible.155,157 Alternatively, it has been interpreted by Adams and Gibbs as the temperature at which the configuration entropy of the supercooled liquid becomes zero.155,157 The fitted parameters for the VFT equation are displayed in Table 21.2.8 for the 1,3-dialkylimidazolium chloride-aluminium chloride systems. Table 21.2.8. Fitted parameters for the VFT equation for some AlCl3 systems156 Melt [C1-mim]Cl- AlCl3
[C3-mim]Cl- AlCl3
[C4-mim]Cl- AlCl3
[C4-bim]Cl- AlCl3
Mol ratio
T0, K
Kη, K
- ln Aη
1:2
109.1
1024
5.444
2:1
194.8
811
4.739
1:1
141.5
879
5.157
1:2
160.0
608
4.351
2:1
185.0
958
5.208
1:1
184.5
484
3.819
1:2
154.8
666
4.537
2:1
165.4
1261
5.793
1:1
156.5
792
4.811
1:2
150.7
750
4.765
These systems show a strong dependence of the composition for the basic range, with an increase in viscosity as the aluminum trichloride mole fraction decreases. In the acidic range the viscosity seems almost independent of the composition, nevertheless showing a slight decrease with increasing aluminum trichloride mole fraction. However it was also observed that increased viscosities where associated with longer alkyl chain lengths. In the study of the 1-methyl-3-ethylimidazolium bromide - aluminum bromide systems, the VFT parameters for the different compositions were fitted as a function of X (AlBr3 mole fraction) to cubic polynomial equations of the form F (x ) = c 0 + c 1 X + c 2 X 2 + c 3 X 3
[21.2.6]
in neutral and basic melts and to second order equations of the form F (x) = c 0 + c 1 X −1 + c 2 X −2
[21.2.7]
in acidic melts. The values generated for these constants are displayed in Table 21.2.9.
1476
D.W. Rooney, K.R. Seddon
Table 21.2.9. Parameters for 1-ethyl-3-methylimidazolium bromide/aluminum bromide systems157 Melt composition Neutral and basic melts 0.35 ≤ X ≤ 0.50
Acidic melts 0.50 < X ≤ 0.75
VFT Parameter
c0
c1
c2
2.1405 × 10
-1.5128 × 10
Kη
-1.4147 × 104
1.1709 × 105
-3.0743 × 105
2.6769 × 105
ln Aη
-1.3581 × 101
1.8522 × 102
-5.1772 × 102
4.7652 × 102
T0
-3.8691 × 102
7.6383 × 102
-2.6176 × 102
-
Kη
6.6201 × 10
-7.8145 × 10
3
-2.5214 × 10
-
ln Aη
2.5515 × 10
-2.2656 × 10
1
7.3019
3 1
4
3.9396 × 10
c3
T0
3
-3.4596 × 104
4
3
-
In this case, the viscosity showed a minimum at X =0.50, with the viscosity rising as X is either increased or decreased; displaying an irregular increase in the acidic range. This result is somewhat different for the chloroaluminate analogue discussed above. It is important to note that in all the studies described above, different types of capillary viscometers were used for measuring the viscosity. This technique generates a kinematic viscosity, ν, that has to be multiplied by the density, ρ, of the melt to obtain the absolute viscosity, η. Other viscosity studies in ionic liquids were carried out by Grätzel et al.158 as part of a research program for the application of ionic liquids as solvents for a dye-sensitized nanocrystalline solar cell. A series of imidazolium based salts with various alkyl substituents and different anions were characterized in terms of viscosity in order to establish a correlation between chemical structure and physical properties. The results are summarized in Table 21.2.10. Table 21.2.10. Viscosity at 20oC, cps. TfO-: triflate; NfO-: nonaflate; Tf2N-: bis(tryfil)amide; TA-: trifluoroacetate; HB-: heptafluorobutanoate; AcO-: acetate158 Imidazolium cation
TfO-
NfO-
1-Methyl
3-Methyl
45
1-Butyl
90
373
1-i-Butyl
1-Ethyl
74
3-Ethyl
34
35
105
162
52
73
182
54
53
1-Butyl
3-Methyl
AcO-
248
323
1-Ethyl-2-methyl 3-Methyl 1-Ethyl-5-methyl
HB-
83
1-CF3CH2 3-Ethyl
TA-
44
1-Ethyl
1-MeOEt
Tf2N-
35
43
48
89
88 51
37 36
21.2 Ionic liquids
1477
According to this study, the viscosity of ionic liquids is mainly controlled by hydrogen bonding, van der Waals forces, molecular weight, and mobility. In respect to the cation structure, alkyl lengthening or fluorination make the salt more viscous. Also the reduction of freedom of rotation (from butyl to isobutyl) increased the viscosity. Unexpectedly, methylation at the C2(H) of the imidazolium ring increased viscosity even if it suppressed the position most likely to hydrogen bond with the anion.158 In respect to the anion, from TfO- to NfO- and from TA- to HB-, the increase of van der Waals attraction dominates over the H-bonding decrease due to the better charge delocalization. However, from TfO- to Tf2N-, the almost complete suppression of H-bonding seems to influence more than the increase in Van der Waals attractions smaller anion weight of TfO-. On the other hand, comparing TA- and AcO-, the H-bonding strength of AcO- is more important than its smaller size. The salts with lower viscosity are TA-, with minimal anion weight and moderate basicity, and Tf2N-, with minimal basicity and moderate anion weight.158 Table 21.2.11. Viscosity of some ionic liquids at 25oC. For the binary mixtures the ratio means the proportion of AlX3 to ImX or TMSuX Cation [C2-py]
[C2-mim]
[C4-mim]
Trimethylsulfonium
η, cps at 25oC
Anion Br/AlCl3 (ratio 2:1)
22.5155
Br/AlBr3 (ratio 2:1)
~50159
Cl/AlCl3 (ratio 2:1)
13.5156
Br/AlBr3 (ratio 2:1)
31157
BF4
37.7 (22oC)160
Triflate (CF3SO3)
43161
Mesylate (CH3SO3)
160161
BF4
233 (30oC)162
PF6
312 (30oC)162
Cl/AlCl3 (ratio 2)
39.3163
Br/AlCl3 (ratio 2)
54.9163
Br/AlBr3 (ratio 2)
138163
Clearly, many structural parameters affect the viscosity of ionic liquids and a more exhaustive study is needed to rationalize the different trends and to establish a correlation model for prediction. Finally, there are some scattered data on viscosity of other ionic liquids which are summarized in Table 21.2.11, together with the most relevant data from the studies cited above. As it was mentioned above, care must be taken when comparing or using these data as different measuring techniques can yield different viscosity results and furthermore, our recent investigations (to be published shortly) have shown that the presence of traces of water or other impurities like Cl- can have a dramatic effect on viscosity.
1478
D.W. Rooney, K.R. Seddon
Table 21.2.12. Densities of the [C4-mim][PF6] and [C4-mim][BF4] ionic liquids when compared against other solvents at room temperature Solvent
ρ, kg m-3
trichloromethane
1483
164
1372
dichloromethane
1327
[C4-mim][BF4]
164
1208
water
1000
diethyl ether
714
toluene
867
2-propanol
786
benzene
877
[C4-mim][PF6]
21.2.5.2 Density Information on solvent density is also another important physical property. This is a particularly used in fluid flow calculations and for the design of liquid/liquid two phase mixer-settler units. Given that many of the ionic liquids have a “heavy” anion it would be expected that their density would be relatively high when compared to normal industrial solvents. However it can be seen from the Table 21.2.12 that the density of the [C4-mim][PF6] and [C4-mim][BF4] ionic liquids fall within the region of the chlorinated solvents. Overall the density of ionic liquids is somewhat easier to model than the viscosity. In general the change of density with temperature has been fitted to linear equations of the form ρ =a + b ×T
[21.2.8]
where a and b are constants.155-157 The fitted parameters for N-alkylpyridinium are displayed in Table 21.2.13. The bromide containing melts showed a much higher density than the chloride ones which is to be expected. It has also been shown that the lengthening of the alkyl chain smoothly decreases the density.155,156 Table 21.2.13. Fitted parameters for the density of N-alkylpyridinium155 Mol ratio
a
-b × 103
Temp range, oC
[NC1-py]Cl -AlCl3
1:2
1.4625
0.87103
25-75
[NC2-py]Cl -AlCl3
1:2
1.4307
0.91446
25-67
[NC3-py]Cl -AlCl3
1:2
1.3973
0.8847
26-75
[NC4-py]Cl -AlCl3
1:2
1.3680
0.86042
31-76
[NC2-py]Br -AlCl3
1:2
1.5473
0.93059
25-32
Melt
The variation of density with different concentrations of aluminum trichloride in 1,3-dialkylimidazolium chloride- aluminum chloride melts is not however a smooth function.156 In the study carried out by Wilkes et al. on the density of 1-methyl-3-ethylimidazolium bromide aluminum bromide systems, the change of the parameters a and b with composition, X, were successfully fitted to polynomials of third order. The constants obtained from this study are shown in Table 21.2.14. Table 21.2.14. Calculated constants for third order polynomial c0 Neutral and basic melts 0.35 ≤ X ≤ 0.50
a b
c1
1.0588
c2
3.8594 -3
1.2907× 10
c3
-2.8945 -2
-1.0448× 10
0.00 -2
1.1519× 10
0.00
21.2 Ionic liquids
1479
c0 Acidic melts 0.5 0< X ≤ 0.75
c1
c2 1
c3 1
a
-5.8410
37.009× 10
-56.829× 10
30.474× 101
b
1.7946× 10-2
-9.0089× 10-2
1.4121× 10-1
-7.4099× 10-2
Grätzel et al.158 has also measured the densities for the different ionic liquids with the results are shown in Table 21.2.15. Table 21.2.15. Density at 22oC or indicated temperature, g cm-3. TfO-: triflate; NfO-: nonaflate; Tf2N-: bis(tryfil)amide; TA-: trifluoroacetate; HB-: heptafluorobutanoate; AcO-: acetate158 Imidazolium cation
TfO-
NfO-
1-Me 1-Et 3-Me
1-Bu
1.390 (20)
1.290
1.520 1.473
(18)
1-Et-2-Me 1-Et-5-Me
1-Et
1.364
HB-
1.285
1.450
1.429
(21)
1.209
1.333
(20)
1.496 1.656(20) 1.452(21)
1.330
1-Bu
1.427
(18)
(19)
1.404
1.250 1.183(23)
1.495(21)
3-Me 3-Me
(19)
1.428
1-CF3CH2 3-Et
TA-
1.559
1-i-Bu 1-MeOEt
Tf2N-
1.334
1.470 1.432(23)
3-Et
As with viscosity, there are some scattered data on density of other ionic liquids which is summarized in Table 21.2.16, together with the most relevant data from the studies cited above. Table 21.2.16. Density of some ionic liquids at 25oC. For the binary mixtures the ratio means the proportion of AlX3 to ImX or TMSuX Cation 1-ethylpyridinium
1-ethyl-3-methylimidazolium
1-butyl-3-methylimidazolium
ρ, g cm-3 at 25oC
Anion Br/AlCl3 (ratio 2)
1.52155
Br/AlBr3 (ratio 2)
2.20159
Cl/AlCl3 (ratio 2)
1.39156
Br/AlBr3 (ratio 2)
2.22157
BF4
1.24 (22oC)160
Triflate (CF3SO3)
1.38161
Mesylate (CH3SO3)
1.24161
BF4
1.17 (30oC)162
PF6
1.37 (30oC)162
1480
D.W. Rooney, K.R. Seddon
Cation 1-butyl-3-ethylimidazolium 1-dodecyl-3-ethylimidazolium
Trimethylsulfonium
ρ, g cm-3 at 25oC
Anion Triflate (CF3SO3)
1.27161
Mesylate (CH3SO3)
1.14161
Triflate (CF3SO3)
1.10161
Cl/AlCl3 (ratio 2)
1.40163
Br/AlCl3 (ratio 2)
1.59163
Br/AlBr3(ratio 2)
2.40163
21.2.6 SUMMARY In summary ionic liquids have a proven ability to enhance current industrial chemistry. Within current industrial processes using conventional solvents, selectivities, TOF and reaction rates are effectively uncontrolled: however by using ionic liquid media for the catalysis, it is possible to have a profound influence on all these factors. Tailoring of the ionic liquid by a combination of subtle (i.e., changing cation substitution patterns) and gross (anion type) modifications can permit very precise tuning of reactions. Ionic liquids have a proven ability to be used as effective solvents and catalysts for clean chemical reactions; as replacements for volatile organic and dipolar aprotic solvents (i.e. DMF, DMSO) and solid acid catalysts in reactions whether being used at the laboratory or industrial scale. Many of their physical properties are in the same region as organic solvents and as such could easily be incorporated into any future processes. Properties such as negligible vapor pressure and high density/variable interfacial tension offer considerable advantages in the design of unit operations such as evaporators, mixer-settlers and reactors. It is envisaged that as more information is gathered on their physical properties it will be possible to accurately design and optimize entire plants around these novel solvent systems. In a time where considerable emphasis is being placed on clean reactions and processes with minimal waste and efficient product extraction ionic liquids have emerged as a novel and exciting alternative to modern production methods. They are truly designer solvents. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
W. Pitner, D. Rooney, K. Seddon, R. Thied, World Patent, WO9941752. W. Pitner, M. Fields, D. Rooney, K. Seddon, R. Thied, World Patent, WO9914160. P. Davey, M. Earle, C. Newman, K. Seddon, World Patent, WO9919288. C. Lok, M. Earle, J. Hamill, G. Roberts, C. Adams, K. Seddon, World Patent, WO9807680. C. Greco, F. Sherif, L. Shyu, U.S. Patent, US5824832. W. Keim, P. Wasserscheid, World Patent, WO9847616. O. Hodgson, M. Morgan, B. Ellis, A. Abdul-Sada, M. Atkins, K. Seddon, U.S. Patent, US5994602. S. Takahashi, N. Koura, S. Kohara, M.L. Saboungi, L.A. Curtis, Plasmas and ions, 2, (1999). A.Carmichael, M. Earle, J. Holbrey, P. McCormac, K. Seddon, Organic letters, 1, 7, 997-1000, (1999). C. DeBellefon, E. Pollet, P. Grenouillet, J. Mol. Cat. A-Chemical, 145, (1999). W. Chen, L. Xu, C. Chatterton, J. Xiao, Chem, Com., 13, (1999). L. Simon, J. Dupont, R. deSouza, Applied Catalysis A-General, 175, (1998). S. Silva, P. Suarez, R. deSouza, J. Dupont. Polymer Bulletin, 40, 4-5, (1998). Y. Chauvin, H. Olivier, C.N. Wyrvalski, L.C. Simon, R.F. De Souza, J. Catalysis, 165, 2, (1997). Y. Chauvin. and H. Olivier-Bourbigou, Chemtech, 25, (1995). Y. Chauvin, Actualite Chimique, (1996). M. Freemantle, Chem. Eng. News, 76, (30th March 1998). M. Freemantle, Chem. Eng. News, 76, (24th August 1998). M. Freemantle, Chem. Eng. News, 77, (4th January 1999). C.L. Hussey, Adv. Molten Salt Chem., 5, (1983).
21.2 Ionic liquids
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
52 53 54
55 56 57 58 59 60 61 62 63 64 65
1481
C.L. Hussey, Pure Appl. Chem., 60, 1, (1988). K.R. Seddon, Kinetics and Catalysis, 37, 693, (1996). K.R. Seddon, J. Chemical Tech. Biotech., 68, (1997). K.R. Seddon, Molten Salt Forum: Proceedings of the 5th International Conference on Molten Salt Chemistry and Technology, (1998). J. D. Holbury and K.R. Seddon, Clean products and processes, 1, (1999) C.L. Huheey, Inorganic chemistry, Principles of structure and reactivity, Harper and Row, (1972) CRC handbook, 53rd ed, CRC Press, (1972). A.F. Kapustinskii, Z. Physik. Chem. (Leipzig), B22, 257, (1933). J. Emsley, The Elements, Oxford, (1989). K.B. Yatsimirskii, Izvest, Akad. Nauk SSSR, Otdel, Khim. Nauk., 453, (1947). R.D. Shannon, Acta Crystallogr., Sect. A: Found. Crystallogr., 32, 751 (1976). A. B. McEwen, H. L. Ngo, K. LeCompte, J. L. Goldman, J. Elect. Chem. Soc., 146, 5, (1999). T.C. Waddington, Adv. Inorg. Chem. Radiochem., 1, (1959). P. Walden, Bull. Acad. Imper. Sci. (St. Petersburg), 1800, (1914). A. Fannin, D. Floreani, L. King, J. Landers, B. Piersma, D. Stech, R. Vaughn, J. Wilkes, J. Williams, J. Phys. Chem, 88, (1984). A.Elaiwi, P. Hitchcock, K. Seddon, N. Srinivasan, Y. Tan, T. Welton and J. Zora, J. Chem. Soc. Dalton Trans, 3467, (1995). B. Chan, N. Chang and M. Grimmel, Aust. J. Chem., 30, (1977). A. McEwen, H. Ngo, K. LeCompte, J. Goldman, J. Electrochem. Soc, 146, 5, (1999). J. Wilkes and M. Zaworotko, J. Chem. Soc., Chem. Com., 965, (1992). S. Wicelinski, R. Gale, Themochemica Acta, 126, (1988). S. Tait and R. Osteryoung, Inorg, Chem., 23, (1984). A. Abdul-Sala, A. greenway, P. Hitchcock, T. Mohammed, K. Seddon and J. Zora, J. Chem. Soc., Chem. Commun., 1753, (1986) . K. Dieter, C. Dymek, N. Heimer, J. Rovang, Jwilkes, J. Am. Chem. Soc., 10, 2711, (1988). T. Bradley, PhD thesis, Queens University of Belfast, Unpublished results, (2000). C.J. Bowlas, D.W. Bruce and K.R. Seddon, K.R, J. Chem. Soc., Chem. Commun., (1996). C.M. Gordon, J.D. Holbrey, A. Kennedy and K.R. Seddon, J. Mater. Chem., 8, (1998). J.D. Holbrey and K.R. Seddon, J. Chem. Soc., Dalton Trans., (1999). V. Posypaiko, E. Alekseeva, Phase Equilibria in binary halides, IFI/Plenum, (1987). F. Hurley, T. Weir, J. Electrochem. Soc., 98, 203, (1951). K. Seddon, in The International George Papatheodorou Symposium: Proceedings, Eds. S. Boghosian, V. Dracopoulos, C.G. Kontoyannis and G.A. Voyiatzis, (1999). R.A. Osteryoung, in Molten Salt Chemistry: An Introduction and Selected Applications (Eds. Mamantov, G. and Marassi, R.), NATO ASI Series C: Mathematical and Physical Sciences, 202, 329, (1987). D. Commereuc., Y. Chauvin, G. Leger, and J. Gaillard, Revue de L’Institut Francais du Petrole, 37, 639, (1982). Y. Chauvin, D. Commereuc, A. Hirschauer, F. Hugues, and L. Saussine, French Patent, FR 2,611,700. Y. Chauvin, F. DiMarco, H.Olivier, and B. Gilbert, Ninth International Symposium on Molten Salts, C.L. Hussey, D.S. Newman, G. Mamantov and Y. Ito (Eds.), The Electrochemical Society, Inc., San Francisco, California. USA, 617, (1994). Y. Chauvin, S. Einloft and H. Olivier, Industrial & Engineering Chemistry Research, 34, 1149, (1995). Y. Chauvin, B. Gilbert and I. Guibard, J. Chem. Soc., Chem. Commun., 1715, (1990). Y. Chauvin, B. Gilbert and I. Guibard, in Seventh International Symposium on Molten Salts, C.L. Hussey, S.N. Flengas, J.S. Wilkes and Y. Ito (Eds.), The Electrochemical Society, Montreal, Canada, 822, (1990). Y. Chauvin, H. Olivier, C.N. Wyrvalski, L.C. Simon, R. De Souza and J. Dupont, J. Catal., 165, 275, (1997). S. Einloft, F.K. Dietrich, R.F. de Souza and J. Dupont, Polyhedron, 15, 3257, (1996). H. Olivier, Y. Chauvin, and A. Hirschauer, Abst. Papers Am. Chem. Soc., 203, 75, (1992). A.K.Abdul-Sada, P.W.Ambler, P.K.G.Hodgson, K.R.Seddon, N.J.Steward, World Patent, WO 9521871. P.W. Ambler, P.K.G. Hodgson and N.J. Stewart, Butene Polymers, European Patent Application, EP/0558187A0558181, (1996). K. Weissermel, and H.J. Arpe, Industrial Organic Chemistry, VCH, Weinheim, 3rd edition, (1997). R.T. Carlin and J.S. Wilkes, J. Mol. Catal., 63, 125, (1990). L.M. Skrzynecki-Cooke and S.W. Lander,. Abst. Papers Am. Chem. Soc., 193, 72, (1987).
1482
66 67 68 69
70 71
72
73 74 75
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
D.W. Rooney, K.R. Seddon
E. Ota, in Joint (Sixth) International Symposium on Molten Salts, G. Mamantov, M. Blander, C.L. Hussey, C. Mamantov, M.L. Saboungi and J.S. Wilkes (Eds.), The Electrochemical Society, Inc., 1002, (1987). E. Ota, J. Electrochem. Soc., 134, C512, (1987). G. Hondrogiannis, C.W. Lee, R.M. Pagni and G. Mamantov, J. Am. Chem. Soc., 115, 9828, (1993). R..M. Pagni, G. Mamantov, C.W. Lee and G. Hondrogannis, in Ninth International Symposium on Molten Salts, C.L. Hussey, D.S. Newman, G. Mamantov and Y. Ito (Eds.), The Electrochemical Society, Inc., San Francisco, California. USA, 638, (1994). C.J.Adams, M.J.Earle, G.Roberts, K.R.Seddon, Chem. Commun., 207, (1998). J.A. Boon, S.W. Lander, J.A. Levisky, J.L. Pflug, L.M. Skrynecki-Cooke and J.S. Wilkes, in Joint (Sixth) International Symposium on Molten Salts, G. Mamantov, M. Blander, C.L. Hussey, C. Mamantov, M.L. Saboungi and J.S. Wilkes (Eds.), The Electrochemical Society, Inc., 979, (1987). J.A. Boon, S.W. Lander, J.A. Levisky, J.L. Pflug, L.M. Skrynecki-Cooke and J.S. Wilkes: Catalysis and reactivity of electrophilic reactions in room-temperature chloroaluminate molten-salts, J. Electrochem. Soc., 134, 510, (1987). J.A. Boon, J.A. Levisky, J.L. Pflug and J.S. Wilkes, J. Org. Chem., 51, 480, (1986). M.B. Jones, D.E. Bartak and D.C. Stanley, Abst. Papers Am. Chem. Soc., 190, 58, (1985). J.A. Levisky, J.L. Pflug and J.S. Wilkes, in The Fourth International Symposium on Molten Salts, M. Blander, D.S. Newman, M.L. Saboungi, G. Mamantov and K. Johnson (Eds.), The Electrochemical Society, Inc., San Francisco, 174, (1984). B.J. Piersma and M. Merchant, in Seventh International Symposium on Molten Salts, C.L. Hussey, S.N. Flengas, J.S. Wilkes and Y. Ito (Eds.), The Electrochemical Society, Montreal, Canada, 805, (1990). J.S. Wilkes, in Molten salt chemistry: An introduction and selected applications, G. Mamantov and R. Marassi (Eds.), D. Reichel Publishing Company, Dordrecht, 405-416, (1987). J.A. Levisky, J.L. Pflug and J.S. Wilkes, J. Electrochem. Soc., 130, C127, (1983) P.J. Dyson, M.C. Grossel, N. Srinivasan, T. Vine, T. Welton, D.J. Williams, A.J.P. White and T Zigras,: J. Chem. Soc., J. Chem. Soc., Dalton Trans., 3465, (1997). J.K.D. Surette, L. Green and R.D. Singer, J. Chem. Soc., Chem. Commun, 2753, (1996). P.A.Z. Suarez, J.E.L. Dullius, S. Einloft, R.F. De Souza and J. Dupont, Polyhedron, 15, 1217, (1996). P.A.Z. Suarez, J.E.L. Dullius, S. Einloft, R.F. De Souza and J. Dupont, Inorg. Chim. Acta, 255, 207, (1997). Y. Chauvin, L. Mussmann and H. Olivier, Angew. Chem. Int. Ed. Engl., 2698, (1995). A.L. Monteiro, F.K. Zinn, R.F. De Souza and J. Dupont, Tetrahedron Asymmetry, 8, 177, (1997). R.T. Carlin and J. Fuller, J. Chem. Soc., Chem. Commun, 1345, (1997). J. Fuller, A.C. Breda and R.T. Carlin, J. Electrochem. Soc., 144, (1997). J.E.L. Dullius, P.A.Z. Suarez, S. Einloft, R.F. de Souza, J. Dupont, J. Fischer and A. DeCian, Organometallics, 17, 815, (1998). S.M. Silva, P.A.Z. Suarez, R.F. De Souza, Polymer Bulletin, 40, 401, (1998). D.A. Jaeger and C.E. Tucker, Tetrahedron Lett., 30, 1785, (1989). M.J. Earle, P.B. McCormac and K.R. Seddon, Green Chem., 1, (1999) T. Fischer, T. Sethi, T. Welton and J. Woolf , Tet. Lett., 40, 793, (1999). J.G. Huddleston, H.D. Willauer, R.P. Swatloski, A.E. Visser, R.D. Rogers. Chem Commun., 1765, (1998). A.E. Visser, R.P. Swatloski and R.D. Rogers, Green Chemistry, 2, 1, (2000). S.G.Cull, J.D. Holbrey, V. Vargas-Mora, K.R. Seddon and G.J. Lye, Biotechnol. Bioeng., (in press) L. A. Blanchard, D. Hancu, E. J. Beckman, J. F. Brennecke, Nature, 399, 6th May, 28, (1999). D. Armstrong, L. He, Y, Liu, Analytical chemistry, 71, 17, (1999) W.R. Pitner, Unpublished Results, Queens University of Belfast, 2000 Y. Liu, P. Y. Chen, I. W. Sun, C. L.Hussey. J. Electrochem. Soc., 144, 7, 1997. Y.F. Lin, I.W. Sun, Electrochimica Acta, 44, 16, (1999). Y.F. Lin, I.W. Sun, J. Electrochem. Soc., 146, 3, (1999). W. J. Gau, I.W. Sun, J. Electrochem. Soc., 143, 1, (1996). E.G.S. Jeng, I.W. Sun, J. Electrochem. Soc., 145, 4, (1998). W.J. Gau, I.W. Sun, J. Electrochem. Soc., 143, 3, (1996). J.A. Mitchell, W. R. Pitner, C. L. Hussey, and G. R. Stafford, Journal of the Electrochemical Society, 143, 11, (1996). M.R. Ali, A. Nishikata, T. Tsuru, Electrochimica Acta, 42, 12, (1997). R.T. Carlin, P.C. Trulove, H.C. DeLong, J. Electrochem. Soc., 143, 9, (1996). R.T. Carlin, H.C. DeLong, J. Fuller and P.C. Trulove, J. Electrochem. Soc., 145, 5, (1998). S. Pye, J. Winnick, P.A. Kohl, J. Electrochem. Soc, 144, 6, (1997). H.C. Delong, J.A. Mitchell, P.C. Trulove, High Temperature Material Processes, 2, 4, (1998).
21.2 Ionic liquids
1483
110 G.P. Ling, N. Koura, Denki Kagaku, 65, 2, (1997). 111 X.H. Xu, C.L. Hussey, J. Electrochem. Soc, 139, 11, (1992). 112 M. Hasan, I.V. Kozhevnikov, M.R.H. Siddiqui, A. Steiner, N. Winterton, Inorganic Chemistry, 38, 25, (1999). 113 X.H. Xu, C.L. Hussey, J. Electrochem. Soc, 139, 5, (1992). 114 F. Endres, W. Freyland, B. Gilbert, Berichte Der Bunsen-Gesellschaft-Physical Chemistry Chemical Physics, 101, 7, (1997). 115 C.A. Zell, F. Endres, W. Freyland, Physical Chemistry Chemical Physics, 1, 4, (1999). 116 F. Endres, W. Freyland, Journal Of Physical Chemistry B, 102, 50, (1998). 117 I.W. Sun, C.L. Hussey, J. Electroanalytical Chem., 274, 1-2, (1989). 118 H.C. Delong, J.S. Wilkes, R.T. Carlin, J. Electrochem. Soc, 141, 4, (1994). 119 X.H. Xu, C.L. Hussey, J. Electrochem. Soc, 140, 5, (1993). 120 F.M. Lin, C.L. Hussey, J. Electrochem. Soc, 140, 11, (1993). 121 C.L. Hussey, X.H. Xu, J. Electrochem. Soc, 138, 7, (1991). 122 D. Costa, W.H. Smith, Abstracts Of Papers Of The American Chemical Society, 216, 2, (1998). 123 S.K.D. Strubinger, I.W. Sun, W.E. Cleland, C.L. Hussey, Inorganic Chemistry, 29, 5, (1990). 124 P.A. Barnard, C.L. Hussey, J. Electrochem. Soc., 137, 3, (1990). 125 S.K.D. Strubinger, I.W. Sun, W.E. Cleland, C.L. Hussey, Inorganic Chemistry, 29, 21, (1990). 126 D. Appleby, P.B. Hitchcock, K.R. Seddon, J.E. Turp, J.A. Zora, C.L. Hussey, J.R. Sanders, T.A. Ryan, Dalton Trans., 6, (1990). 127 S.K.D Strubinger, C.L. Hussey, W.E. Cleland, Inorganic Chemistry, 30, 22, (1991). 128 C.L. Hussey, P.A. Barnard, I.W. Sun, D. Appleby, P.B. Hitchcock, K.R. Seddon, T. Welton, J.A. Zora, J. Electrochem. Soc., 138, 9, (1991). 129 R. Quigley, P.A. Barnard, C.L. Hussey, K.R. Seddon, Inorganic Chemistry, 31, 7, (1992). 130 C.L. Hussey, R. Quigley, K.R. Seddon, Inorganic Chemistry, 34, 1, (1995). 131 R.I. Crisp, C.L. Hussey, K.R. Seddon, Polyhedron, 14, 19, (1995). 132 Q. Liao, W.R. Pitner, G. Stewart, C.L. Hussey, and G.R. Stafford, J. Electrochem. Soc., 144, 3, (1997). 133 T.P. Moffat, J. Electrochem. Soc., 141, (1994) 134 G. R. Stafford, ibid., 136, (1989) 135 G. R. Stafford, B. Grushko, and R. D. Mc Michale, J. Alloys Compd., 200, (1993) 136 T. P. Moffat, J. Electrochem. Soc., 141, (1994). 137 G. R. Stafford, ibid., 945, (1994) 138 C. L. Hussey and T. M. Laher, ibid., 20, (1981) 139 C. L. Hussey, L. A. King and R. A. Carpio, J. Electrochem. Soc., 126, (1979) 140 R. J. Gale, B. Gilbert and R. A. Osteryoung, Inorg. Chem., 18, (1979) 141 W. R. Pitner , PhD Thesis, University of Mississippi, (1997) 142 B. J. Tierney, W. R. Pitner, C. L. Hussey, and G. R. Stafford. J. Electrochem. Soc., 145, (1998). 143 W. R. Pitner, C. L. Hussey, and G. R. Stafford, J. Electrochem. Soc., 143, 1, (1996). 144 W. R. Pitner and C. L. Hussey, J. Electrochem. Soc., 144, 9, (1997). 145 R.T. Carlin, P.C. Truelove, Electrochimica Acta, 37, 14, (1992). 146 G.T. Cheek, R.A. Osteryoung, J. Electrochem. Soc., 129, 11, (1982). 147 M. Lipsztajn, R.A. Osteryoung, Inorganic Chemistry, 24, 5, (1985). 148 C.L. Hussey, L.A. King, The Second International Symposium on Molten Salts, Pittsburgh, USA, The Electrochemical Society, Inc., (1978). 149 J. S. Tang, R.A. Osteryoung, Synthetic Metals, 45, 1, (1991). 150 L. Janiszewska, R.A. Osteryoung, J. Electrochem. Soc., 135, 1, (1988). 151 G.T. Cheek, R.B. Herzog, The Fourth International Symposium on Molten Salts, San Francisco, The Electrochemical Society, Inc., (1984). 152 J.E. Coffield, G. Mamantov, J. Electrochem. Soc., 138, 9, (1991). 153 J.E. Coffield, G. Mamantov, J. Electrochem. Soc., 139, 2, (1992). 154 F. Fuller, R.T. Carlin, R.A. Osteryoung, J. Electrochem. Soc., 144, 11, (1997). 155 R.A. Carpio, A.K. Lowell, R.E. Lindstrom, J.C. Nardi and C.L. Hussey, J.Electrochem.Soc, 126; (1979). 156 A.A. Fannin, D.A. Floreani, L.A. King, J.S. Landers; B.J. Piersma, D.J. Stech, R.L. Vaughn J.S. Wilkes and J.L. Williams, J.Phys.Chem; 88, (1984). 157 J.R. Sanders, E.H. Ward and C.L. Hussey, J.Electrochem.Soc.; 133; (1986). 158 P. Bônhote, A. Dias, N. Papageorgiou, K. Kalyaanaasundaram and Grätzel, Inorg.Chem.; 35, (1996). 159 J. Robinson, R.C. Bugle, H.L. Chum, D. Koran, and R.A. Osteryoung, J.Am.Chem.Soc, 101, (1979). 160 J. Fuller, R.T. Carlin and R.A. Osteryoung, J. Electrochem.Soc; 144, 11, (1997).
1484
Victor Cherginets
161 E.I. Cooper, E.J.M. O’Sullivan, Proceedings of the 8th international symposium on molten salts; 92, 16, (1992). 162 P.A.Z. Suarez, S. Einloft, J.E.L. Dullius, R.F. de Souza, and J. Dupont, J.Chem.Phys., 95, (1998). 163 M. Ma and K.E. Johnsom, Can. J. Chem., 73, (1995). 164 M.Torres, Unpublished results, Queens University of Belfast, (2000).
21.3 OXIDE SOLUBILITIES IN IONIC MELTS Victor Cherginets Institute for Single Crystals, Kharkov, Ukraine
21.3.1 METHODS USED FOR SOLUBILITY ESTIMATIONS IN IONIC MELTS Processes of the dissolution of metal oxides in ionic melts are accompanied by interactions between ions of dissolved substance with ions of the melt (solvation). The superimposition of the mentioned processes results in the formation of metal complexes with the melt anions and cation complexes with oxide-ions. Therewithal, the definite part of the oxide passes into the solution without dissociation as uncharged particles. Thus, in saturated solution of oxide the following equilibria take place: MeOs = MeO l
[21.3.1]
MeO l = Me 2+ + O 2 −
[21.3.2]
Me 2+ + iX k − = MeX i2 − ik
[21.3.3]
O 2 − + nKt m+ = Kt n O nm − 2
[21.3.4]
where: s and l Me Ktm+ Xk-
subscripts denoting solid and dissolved oxide, respectively, the designation of two-valent metal the melt cation (such as Cs+, K+, Na+, Ba2+, Ca2+, etc.), the melt anion (Cl-, Br- , I-, SO42-, PO3-, etc.)
Since at the dissolution of any oxide in melts a degree of interaction “Ktm+ - O2-” should be the same and the complexation with melt anions for cations of oxide may be assumed as closed,1 therefore, it can be believed that oxide solubilities depend mainly on the degree of interactions [21.3.1] and [21.3.2], latter may be considered as an acid-base interactions as proposed by Lux.2 The constant of [21.3.2] K MeO = where: a and m
a Me 2 + aO 2 − a MeO
≈
m Me 2 + mO 2 − m MeO
[21.3.5]
activities and molarities of the particles noted in the subscripts
may be considered as a measure of acidic properties of the cation if reaction [21.3.2] is homogeneous. However, a majority of oxides possess limited solubilities in molten salts and the excess of the oxide should precipitate. In this case the fixation and removing oxide ions
21.3 Oxide solubilities in ionic melts
1485
from melt took place not only because of the cation acidity but owing to formation of new phase - the precipitate of the oxide. Therefore, works in which the estimations of cation acidities were based on measurements of changes in oxide concentration before and after cation addition3 or by interactions between, e.g., carbonates with Lux acids4,5 contained distorted results. The latter works4,5 contain the additional error - insoluble metal carbonates MeCO3 after interaction with acid K2Cr2O74 or NaPO35 were transformed into insoluble chromates or phosphates - all reactions were heterogeneous. The above may be also referred to works of Slama6,7 where cation acidities were estimated on the base of reactions Me 2+ + NO 3− = MeO ↓ + NO 2+
[21.3.6]
NO +2 + NO 3− = 2NO 2 ↑ +
[21.3.7]
1
2
O2
there was no evidence of homogeneity of reaction [21.3.6]. The mentioned method seems to have no wide usage for studying behavior of oxides in molten salts, solubility studies by isothermal saturation and potentiometric titration methods are more precise and informative. From the listed above scheme of equilibria [21.3.1-21.3.4] it follows that the fixation of oxide ions by metal cations is in common case a heterogeneous process resulting in the formation of a new phase and cation acidity may have no connection with the completeness of interaction Me2+ - O2- in molten salts. From eq. [21.3.1] it is clear that molecular oxide concentration in the saturated solution is not dependent on the constituent ion concentrations but connected with the precipitate properties, mainly with the molar surface of the deposit, σ. Since the concentration of the non-dissociated oxide in the saturated solution, as a rule, is hardly determined, for practical purposes the solubility product, the magnitudes of PMeO (used below pP≡-logP), PMeO = a Me 2 + aO 2 − ≈ m Me 2 + mO 2 −
[21.3.8]
are usually employed for the description of the saturated solutions. The known values of PMeO and KMeO give possibility to estimate the concentration of the non-dissociated oxide, sMeO, in the saturated solution: s MeO =
PMeO K MeO
[21.3.9]
It should be noted, however, that there was no reliable method for determining dissociation constants and sMeO. Let us consider two generally accepted methods for oxide solubility determinations. 21.3.1.1 Isothermal saturation method This method is simple enough and includes some modifications: 1. placing baked sample of the oxide in the melt-solvent and regular tests of the metal concentration in the melt up to the saturation. Known test routines are either radiochemical8 or complexonometric9 analysis of cooled samples of the saturated solutions;
1486
Victor Cherginets
2. the addition of known weights of oxide to the melt up to the saturation detected by a potentiometric technique.10,11 The sum of molecular and ionic form concentrations in the saturated solution, ΣsMeO, is obtained as the main result of such investigations, this magnitude may be expressed by the following equation:
∑s
MeO
= PMeO +
PMeO K MeO
[21.3.10]
Since ΣsMeO is the main result of this method, there exists a possibility to determine the dissociation constant according to routine 2 if an oxygen electrode was preliminary calibrated by known additions of the completely dissociated Lux base. But there is no information about such studies. Studies connected with the analysis of cooled samples allow to obtain thermal dependencies of oxide solubilities by analysis of the saturated melt heated to the definite temperature. Main disadvantages of isothermal saturation method are: • the inclusion of suspended particles of oxide in the sample for analysis and following overvaluing of the results; • the magnitude determined is the concentration but not its logarithm; range of oxides available for this routine is essentially narrowed and this method is unsuitable for studies of practically insoluble substances. The presence of oxygen impurities in the melt studied leads to lowering the calculated solubilities, especially, if oxide solubility is of the same order as initial oxide concentration since oxide admixtures suppress the oxide dissolution. 21.3.1.2 Potentiometric titration method This method is more frequently used than the previously described one. It may be employed for oxide solubility studies in a wide range of concentrations of saturated solutions. This method allows to determine solubility products of oxides and in some cases (the existence of the non-saturated solution region) dissociation constants may be calculated. As compared with isothermal saturation a potentiometric method possesses the following advantages: • e.m.f. values are linearly dependent on logarithm of O2- concentration; it is possible to measure the latter even in solutions with extremely low potential-determining ion concentrations; • simplicity of e.m.f. measurements and good reproducibility of the experimental values; • possibility to made studies in situ, i.e., without any effect on the process studied. Main limitations of the potentiometric titration are due to the potential of liquid junction, however, the latter is negligible in molten salts,10 and the absence of reliable and generally accepted methods for calculation of activity coefficients for oxide ions and cations. As for determination of solubility and dissociation parameters it should be noted that this method allows to determine values KMeO only if the potentiometric curve contains the non-saturated solution region. Studies in the wide temperature range are impeded as compared with the isothermal saturation method.
21.3 Oxide solubilities in ionic melts
1487
21.3.2 OXYGEN-CONTAINING MELTS Frederics and Temple studied CuO, MgO, PbO and ZnO solubilities in molten equimolar mixture KNO3-NaNO3 in the range 290-320oC.12 The solubilities were determined by isothermal saturation method with the potentiometric control of solubilities. Solubility products were 2.24×10-13 for CuO, 2.2×10-13 for ZnO, 2.16×10-14 for MgO and 4.34×10-12 for PbO. The increase of the temperature to 320oC led to the increase of the solubilities by 6-10 times.12 Individual Lux acids derived from B2O3 (borax) and P2O5 (sodium metaphosphate) are interesting since they are often used as acidic components of different fluxes. Delimarsky and Andreeva10 determined PbO solubility in molten NaPO3 at 720oC by isothermal saturation method with the potentiometric control of saturation. The concentration cell with oxygen electrodes Pt(O2) was used. PbO solubility was estimated as 31.6 mol%. Nan and Delimarsky investigated solubilities of acidic metal oxides (MoO3, WO3, TiO2) in molten borax at 900oC by a similar method.13 The control of saturation was performed by a gravimetric analysis. Solubilities of the oxides deviated from those predicted by Shreder’s equation, because of chemical interactions between the substances dissolved and the solvent. Oxide solubilities increased with the reduction of melting point of oxide. Solubilities were 66.1 mol% for MoO3, 63.2 mol% for WO3 and 21.2 mol% for TiO3. 21.3.3 HALIDE MELTS 21.3.3.1 The eutectic mixture KCl-LiCl (0.41:0.59) Delarue14,15 performed qualitative studies on oxide solubilities in the melt at t~500oC by visual method. The experiments consisted of addition of potassium hydroxide, KOH, to the metal chloride solution in KCl-LiCl. If there were no formation of the oxide deposit then the oxide was considered soluble, in other cases, oxide was referred to slightly soluble or insoluble depending on the amount of oxide deposited. CdO, PbO, BaO, CaO and Ag2O belonged to the group of soluble oxides, CoO, NiO, ZnO were examples of slightly soluble oxides and MgO, BeO and Al2O3 were insoluble. Copper (II) oxide was unstable in molten chlorides since its dissolution was accompanied by the reduction of CuII to CuI. This transformation has been demonstrated16 to be substantial even under the partial pressure of chlorine equal to 1 atm. The similar behavior is typical of other transition cations in the highest degree of oxidation, e.g., TlIII, FeIII, and AuIII. The addition of corresponding chlorides to the saturated solutions formed by non-oxidizing cations resulted in dissolution of latter due to reactions similar to: 2CuCl 2 + ZnO↓ → 2CuCl + ZnCl 2 +
1
2
O2 ↑
[21.3.11]
An important feature of oxide solubilities14,15 was the essential difference of solubility values of “powdered” (added as powder) and prepared in situ oxide solubilities. Solubilities of latter samples have been found to be considerably greater (CaO, CdO), although there was no explanation to this observation. Quantitative characteristics of oxides solubilities in KCl-LiCl were not presented.14,15 Laitinen and Bhitia17 determined solubility products for certain oxides at 450oC in order to evaluate possibility to use corresponding metal-oxide electrodes for oxoacidity measurements in molten salts. Oxide solubilities were as the following (mole/kg): NiO 3.3×10-4, BiOCl - 6.8×10-4, PdO - 9.4×10-3, PtO - 3.32×10-2, Cu2O - 3.8×10-2.
1488
Victor Cherginets
The important fact found was the relatively high PtO solubility, therefore, platinum gas oxygen electrode, sometimes considered as metal-oxide one, cannot be used for measurement of pO in strongly acidic media because of complete dissolution of the oxide film over its surface. Pt|PtO electrode was attempted for pO index measurements in buffer solutions Ni2+/NiO.18 Shapoval et al.19-21 studied saturated solutions of different oxides by polarographic technique. A degree of interaction between oxide dissolved and the melt has been estimated taking into account potential and polarogram shifts and their deviations from theoretical ones. Rate of cation reduction has been found to be limited by the acid-base dissociation stage. Stability constants of some oxides have been found to be (in mol%): CoO (9.9±1.9)×10, NiO - (6.8±0.8)×103, PbO - (3.4±1.3)×103, Bi2O3 - (8.7±3.5)×103. Cadmium, copper (I) and silver oxides have been shown to completely dissociate under the experimental conditions (400-600oC) this was in a good agreement with Delarue’s results.14,15 Stabilities of iron (II) and (III) oxides in the chloride melt at 470oC have been investigated.22,23 Oxide precipitation by carbonate CO32- ≡ O2- has been shown by potentiometric and diffractometric data to result in formation of FeO from FeII solutions. FeIII precipitation led to formation of solid solutions LiFeO2-LiyFe1-yO. Solubility products in molarity scale have been determined as FeO-10-5.4, Fe3O4-10-36.3, Fe2O3-10-29.16. Cations Fe3+ have been found to oxidize chloride melt with chlorine evolution, that is in good agreement with the results of other studies.14,15 Carbonate ion as oxide donor had its dissociation to O2- essentially incomplete and, therefore, obtained solubility products gave systematic error.20,21 Cherginets and Rebrova24,25 studied oxide solubilities in this melt at 700oC. Solubility products of CoO (pP=4.43±0.11) NiO (pP=5.34±0.2) and MgO (pP=5.87±0.05) were higher than those in molten eutectic KCl-NaCl. The shift of the solubilities in molar fraction scale was ~3.36 log units and allowed to estimate solubilities of other MeO oxides on the base of the known values in KCl-NaCl. It has been shown that acidic properties of Pb2+ and Cd2+ cations were suppressed by those of Li+, therefore, these cations and Ba2+ and Sr2+ did not change oxoacidic properties of the molten KCl-LiCl. 21.3.3.2 Molten KCl-NaCl (0.50:0.50) One of the first works devoted to oxide solubilities determination in this melt has been performed by Naumann and Reinhardt.8 CaO, SrO and BaO solubilities have been determined by isothermal saturation technique with isotopic control of the saturation in temperature range from melting point to 900oC. KCl-NaCl and individual molten chlorides, KCl and NaCl, have been used as solvents. Oxide solubilities have been stated to increase in the sequence CaO<SrO
21.3 Oxide solubilities in ionic melts
1489
Li2O, and 1.2×10-2 for BaO, these values showed that all studied cations had considerable acidic properties.26,27 Ovsyannikova and Rybkin3 developed the cation acidity scale in molten KCl-NaCl at o 700 C on the basis of e.m.f. shifts after addition of metal sulfates. Acidic properties of the main subgroup elements decreased with the increase of their atomic numbers and there was no similar relationships for side groups and transition metals. Quantitative data elucidation according to acid-base equilibrium [21.3.2] in this melt is affected by additional reaction: SO 42 − = SO 3 ↑ + O 2 −
[21.3.12]
caused by additions to chloride melts of the corresponding sulfates. The use of sulfates may cause SO3 formation from highly acidic cation solutions. In particular, Na+ is the most acidic cation of KCl-NaCl melt, therefore, additions of more basic oxides than Na2O should result in oxide ion exchange: Me 2I O + 2Na + → 2Me + + Na 2 O
[21.3.13]
which is shifted to the right. Similar considerations have been made for K+ - Na+ - O2- system in molten KCl-NaCl.28 From [21.3.13] it follows that any solution of cations in Na-based melts cannot create oxide ion concentration exceeding that of equimolar addition of corresponding salt of the most acidic cation of the melt (Na+). However, the oxide ion concentration in BaSO4 solution has been shown3 to be 10 times higher than in K2SO4 and Na2SO4 (∆E=0.11 V). Similar deviations for other cations are smaller, e.g., Cs - 0.49 V, Sr - 0.30 V, Rb - 0.1 V. A potentiometric study of ZnO, MgO, NiO and SrO solubilities at 700oC have been conducted29 by direct and reverse titrations of cation by KOH. These measurements have resulted in a set of characteristics corresponding (by the calculation formula) to Figure 21.3.1. Alkaline oxide solubilities in KCl-NaCl: 111,28-32 at solubility products and dissocia1000K; 2;29 3;8 4;9 5;33,38 6;40,41 7;42 at 700oC. tion constants. The averaging or another procedure of data treatment were not included. Oxide solubilities have been found to increase in sequence: MgO
1490
Victor Cherginets
Thermal dependence of ZrO2 solubility (molar fraction, N) in molten KCl-NaCl was evaluated in the temperature range of 973-1174 K:39 N=(-5.7±3.1)×10-6 + (7±3)×10-9×T. ZrO2 solubility at t~700oC was negligible. Barbin et al.40,41 determined thermal dependence of Li2O solubility in equimolar mixture KCl-NaCl by isothermal saturation technique in temperature range of 973-1073 K: N=0.107-5.221/T, in this range the solubility varied from 0.52 to 0.86 mol%. In earlier work by Kaneko and Kojima,42 the solubility at 973 K was lower (0.31 mol%). The solubility of lithium oxide was close to that for BaO. Solubility products of some oxides in molten KCl-NaCl in molarity scale are presented in Table 21.3.1 and plotted in Figure 21.3.1 (data obtained by same authors are connected by lines). Table 21.3.1. Oxide solubilities in molten KCl-NaCl at 700oC (-log P, molarities) pP11,28-32
pP29
MgO
9.00±0.15
8.46
CaO
~6.29
Oxide
SrO
3.00
BaO
2.31±0.05
NiO
11.2
ZnO Cu2O
pP8
pP9
pP33,38 9.27±0.06
6.62
8.36
4.36±0.06
5.84
7.60
3.08±0.40
4.22
7.05
2.30±0.15
8.32
9.03±0.06
6.18
6.93±0.20
5.4
4.17±0.30
MnO
6.78±0.05
CoO
7.89±0.03
CdO
5.00±0.03
PbO
5.12±0.05
Two groups of values are presented: obtained from isothermal saturation and potentiometric results. The precision of results from isothermal saturation is worse. The potentiometric results should give lower values than isothermal saturation because it includes molecular oxide concentration. Figure 21.3.1 shows that results have opposite trend to [21.3.10]. The following explanation of the above discrepancy is proposed.43,44 Let us consider the chemical potential, µ, of oxide in its saturated solution: 0 µ MeO,s = µ MeO , l + RT ln s MeO, l
[21.3.14]
From this equation it follows that the oxide solubility should be constant. But every precipitate possesses a definite surface, hence, the effect of latter should be taken into account, too: 0 µ MeO,s + σS = µ MeO , l + RT ln s MeO, l
where:
σ S
the surface energy molar surface square of the precipitate
[21.3.15]
21.3 Oxide solubilities in ionic melts
1491
Since for the same oxide σ is constant, the increase of the precipitate square should result in the increase of precipitate solubility. Similar considerations were included in well-known Ostvald-Freundlich equation.45 For substances having 1:1 dissociation, a similar equation can be written in the following form: RT s 1 RT P1 σ 1 1 = = − ln ln M s2 M P2 d r2 r1
[21.3.16]
where: s1, s2 P1, P2 M d
the solubility of crystals with radii r1 and r2, respectively corresponding solubility products molecular weight density
From this equation, it follows that the increase of crystal size reduces its solubility. Oxide particles deposited from more concentrated cation solution, should possess larger sizes due to the so-called “deposit ageing”, than those obtained from mere diluted solutions. Therefore, results of Delimarsky et al.29 obtained from 0.01 mole/kg solutions should be higher that those obtained in other works11,33,38 using 0.05 mole/kg solutions. In the first case solubility was greater approximately by half-order of the magnitude than in second one. Oxide formed from more concentrated solution should have less surface and, hence, less solubility. Data obtained in works11,33,38 practically coincide although we33,38 have used NaOH as titrant similarly to other study.29 Sedimentational titration results in formation of fine dispersed oxide which immediately begins to age because of the recrystallization, i.e., the growth of larger crystals and disappearance of smaller ones, this process leads to the surface energy, σS, decrease. Continuous holding at high temperature and high solution concentration favors this process. In particular, samples used for isothermal saturation technique studies exposed to high temperatures for a long time, and tablets obtained are held in the contact with the melt for some hours. High concentration solutions favor transfer of substance from small to large crystals (diffusion). The differences between data in references 8 and 9 may be explained by different conditions of the powder calcination. During the annealing of the oxide sample recrystallization processes occur, which lead to the reduction of surface square of the powder and the decrease of solubility. The differences in data for KCl-NaCl are caused by the differences in crystal sizes of solid oxide being in equilibrium state with the saturated solution. 21.3.3.3 Other chloride-based melts ZrO2 solubilities in molten mixtures KCl-KPO3 at 800oC have been investigated.46 ZrIV concentration in pure KPO3 was 1.34 wt%, and in the equimolar mixture KCl-KPO3 concentration was 3.25 wt%. A reason for solubility increase is in the depolymerization of PO3- which may be schematically described by the following equation:
(PO 3 ) n
n−
= nPO 3−
[21.3.17]
1492
Victor Cherginets
and increase in acidic properties of melt. The increase of KCl molar fraction over 0.6 led to sharp reduction of ZrO2 solubility because the decrease of acid (PO3-) concentration could not be compensated by the depolymerization process [21.3.17]. Deanhardt and Stern47,48 studied solubilities of NiO and Y2O3 in molten NaCl and Na2SO4 at 1100oC. Y2O3 solubility product was (1.4-2.2)×10-36 and (4.5-6.4)×10-31 in the chloride and the sulphate melt respectively. The formation of YO2was observed when oxide ion was in excess. The formation of peroxide ions in the presence of O2 over melt was favored by thermodynamic properties. Watson and Perry49 have studied ZnO solubility in molten KCl and found that it was 2.3×10-8 mole/kg for ZnO and 3.2×10-12 for K2ZnO2. We have studied oxide solubilities in molten CsCl-KCl-NaCl at 600 and 700oC in order to estiFigure 21.3.2. The formal ionic moments (I, rX-1 , Å-1) and polariza- mate oxide solubility changes with temperature.34-36,50 Oxide soltions (α) of halide-ions. ubilities were lower than those predicted by Shreder’s equation but estimations of their thermal changes usually were in a good agreement with those calculated from Shreder’s equation. Oxides solubilities (in molar fractions) in KCl-NaCl and CsCl-KCl-NaCl have been found to be close at the same temperature. Some works are related to oxo-acidity studies in alkaline chloride melts with cations of high acidity: MgO-BaCl2-CaCl2-NaCl,37 CaO-CaCl2-KCl-NaCl,51 CeCl3-KCl-NaCl,52 ZnO-NaCl-ZnCl.2.53,54 Acidic cations have been found to affect acid-base interactions by fixing oxide ions. Oxide solubilities37,51 were higher than those for alkaline chloride mixtures without acidic admixtures. The same conclusions came from studies55 on MgO solubility in CaCl2-CaO mixture, Ks=10-6.2 at 1133 K. The oxide solubility products in Ca-based chlorides are increased by 4 orders of magnitude. Solubilities (as pP) of MgO (7,61±0,06), NiO (7,44±0,09), CoO (6,25±0,12), MnO (5,20±0,17), ZnO (4,97±0,07) and CdO (3,21±0,07) in molten eutectic SrCl2-KCl-NaCl (0.22:0.42:0.36) at 700oC have been determined.56 The precipitation of PbO from solutions containing Pb2+ was not observed although these cations demonstrated appreciable acidic properties (pK=2,58±0,05). The oxide solubility products in molar fractions have been increased by 1.95 orders of the magnitude for all the oxides. A similar study was performed in two Ba-based melts: BaCl2-KCl-NaCl (0.43:0.29:0.28) and BaCl2-KCl (0.26:0.74) at 700oC.57 Solubility products in molar fractions of CdO were 5.43 and 5.62, for ZnO these magnitudes were 6.97 and 7.14 in molten BaCl2-KCl-NaCl and BaCl2-KCl, respectively. The corresponding pP in KCl-NaCl were 2.01 and 1.83. The oxide solubility products in alkaline earth and Li-based chlorides are presented in Table 21.3.2. -
21.3 Oxide solubilities in ionic melts
1493
Table 21.3.2. Solubility products of some oxides in melts based on moderately acidic cations at 700oC (molar fractions, at the confidence level 0.95) Oxide/composition
MgO
KCl-LiCl
24,25
56
57
57
SrCl2-KCl-NaCl
BaCl2-KCl-NaCl
BaCl2-KCl
0.4:0.6
0.22:0.42:0.36
0.43:0.29:0.28
0.26:0.74
8.38±0.05
9.63±0.06
9.76±0.12
7.32±0.17
7.38±0.06
MnO CoO
6.94±0.11
8.37±0.12
8.12±0.12
NiO
7.85±0.2
9.56±0.09
9.26±0.11
ZnO
7.09±0.07
6.97±0.1
7.11±0.1
CdO
5.33±0.07
5.43±0.2
5.62±0.2
21.3.3.4 Other alkaline halides NiO solubility in molten KF-LiF at 550oC determined by potentiometric method is 1.3×10-6.58 This value considerably exceeds values obtained for the above chloride melts. The increase of NiO solubility occurred due to formation of fluoride complexes which are more stable than chloride ones. Demirskaya, Cherginets and Khailova conducted oxide solubility studies in molten CsBr⋅2KBr59-61 and CsI62 at 700oC. Results are presented in Table 21.3.3. These data are useful to evaluate the effect of halide ion of melt on changes of oxide solubilities. There is a significant reduction of oxide solubility with anion exchange Cl-→Br-, pP changes were of order of 1-2 units for all oxides studied, a degree of dissociation decreases simultaneously. Oxide solubilities in molten CsI possess intermediate values (Table 21.3.3.). The changes of solubility in halide melts may be explained by changes of complexation abilities of halide ions. In general, the complexation ability increases together with the formal ionic moment, rX-1- , Å-1, and the polarization, α, Å3, of anions. In the sequence F->Cl->Br->I- the formal ionic moments decrease and polarization increases as in Figure 21.3.2. The superimposition of these factors leads to the extremum in the bonding energy in the halide complexes, therefore, its minimum for chloride or bromide complexes should be expected. The studies33,59-62 show that this leads to minimal solubilities in molten bromides. Table 21.3.3. Solubility products, of some oxides in molten alkaline halides (molar fractions).[After references 33,59-62] Oxide
CsCl-KCl-NaCl (0.45:0.25:0.30)
CsBr-KBr (0.66:0.34)
CsI
MgO
12.68±0.11
11.68±0.04
12.76±0.06
CaO
7.73±0.05
6.86±0.13
9.93±0.06
7.50±0.14
SrO
5.81±0.40
5.54±0.13
6.82±0.12
6.28±0.30
BaO
4.75±0.10
4.50±0.20
5.16±0.30
4.98±0.15
MnO
9.73±0.15
9.49±0.14
10.70±0.04
CoO
10.92±0.06
10.50±0.03
11.33±0.20
NiO
12.60±0.08
11.42±0.06
12.71±0.30
1494
Victor Cherginets
Oxide
CsCl-KCl-NaCl (0.45:0.25:0.30)
CsBr-KBr (0.66:0.34)
CsI
Cu2O
6.95±0.10
6.35±0.05
ZnO
8.80±0.07
8.15±0.03
11.03±0.16
CdO
7.54±0.48
7.09±0.05
9.35±0.16
8.00±0.06
SnO
11.28±0.05
PbO
8.21±0.05
7.04±0.05
8.52±0.07
7.80±0.10
21.3.4 ON THE POSSIBILITY TO PREDICT OXIDE SOLUBILITIES ON THE BASE OF THE EXISTING DATA Studies33,59-62 show that the estimation of oxide solubilities in non-studied halide melts are possible, at least, for 1:1 oxides, i.e., MeO. 21.3.4.1 The estimation of effect of anion It has been found that oxide solubilities (pP) are related to cation radius.33 For Na- and K-based halide melts, a plot for solubility products expressed in molar fractions is described by the following equation: −2 pPN = pPN , 0 + a rMe 2+
[21.3.20]
where: rMe2 +
the cation of oxide radius, nm.
The value of a is ~0.053 and values of pPN,0 are 1.8, 3.2 and 2.6 for chloride, bromide and iodide melts, respectively. The accurate values are presented in Table 21.3.4. Table 21.3.4. Coefficients of plots [21.3.20] for molten alkaline halides at the confidence level 0.95. After references [38,59-62] t, oC
pPN,0
a
KCl-NaCl
700
1.8±0.9
0.053±0.01
CsCl-KCl-NaCl
600
1.7±2.0
0.057±0.01
CsCl-KCl-NaCl
700
1.7±0.3
0.054±0.02
CsBr-KBr
700
3.2±1.8
0.053±0.01
CsI
700
2.65±0.58
0.052±0.01
Melt
21.3.4.2 The estimation of effect of melt acidity Papers25,26,57,58 give us the possibility to estimate the oxide solubilities for melts with cations more acidic than Na+. For acidic melts the following equation applies: pPN = pPN , KCl − NaCl − pI I
[21.3.21]
where pIl is ~3.5 for KCl-NaCl, ~2 for SrCl2-KCl-NaCl, ~4 for CaCl2-KCl-NaCl and ~1,8 for BaCl2-KCl. If the composition of acidic cation is different from that of melts described in the previous section the value pIl in [21.3.21] can be corrected as follows:
21.3 Oxide solubilities in ionic melts
pI l l = pI l + log
NMe n + , l
l
NMe n + , l
1495
[21.3.22]
where: pNMen +, l , NMen +, l l
molar fractions of the most acidic cation in non-studied and the studied melts, respectively.
21.3.4.3 The estimation of effect of temperature The pP value obtained according the above equations can be corrected for temperature if it differs from 700oC. It has been shown that at relatively small temperature changes oxide solubility is close to value predicted by the Shreder’s equation.37 This allows one to estimate the solubility products at different temperatures using the following equation: pPT 2 = pPT * −
5.2Tmp (T * −T2 )
(T * + T2 )
[21.3.23]
where: T* T2 Tmp
temperature at which pPT* is known temperature at which pP should be estimated melting point of the oxide
21.3.5 CONCLUSIONS The existing data allow • to estimate level of contamination of melts by oxide containing impurities • to predict oxide solubility in molten salts having different anion composition and acidic properties It should be emphasized, however, that there are problems which require further work: • quantitative estimation of surface effect on oxide solubility • the removal of oxide ion admixtures by the their conversion to oxide deposit. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
S.E.Lumkis, Izv.AN SSSR, Otd.Tekh.Nauk, N12, 100 (1958). H.Lux, Z. Elektrochem., 45, 303 (1939). N.N.Ovsyannikova and Yu.F.Rybkin, Ukr.Khim.Zhurn., 42, 151 (1976). A.M.Shams El Din and A.A.El Hosary, J.Electroanal.Chem., 16, 551 (1968). A.M.Shams El Din, A.A.El Hosary, H.D. Taki El Din, Electrochim.Acta, 13, 407 (1968). I.Slama, Coll.Czechoslov.Chem.Commun., 28, 985 (1963). I.Slama, Coll.Czechoslov.Chem.Commun., 28, 1069 (1963). D.Naumann and G.Reinhardt, Z.anorg.allg.Chem., 343, 165 (1966). A.V.Volkovich, Rasplavy , N4, 24 (1991). Yu.K.Delimarsky and V.N.Andreeva, Zhurn.Neorg.Khim, 5, 1123 (1960). R.Combes, F.De Andrade, A.De Barros and H.Ferreira, Electrochim. Acta, 25, 371 (1980). M.Frederics and R.B.Temple, Inorg.Chem., 11, 968 (1972). Shen’ Tsin Nan and Y.K.Delimarsky, Ukr.Khim.Zhurn., 27, 454 (1961). G.Delarue, J.Electroanalyt.Chem., 1, 13 (1959). G.Delarue, Bull.Soc.Chim.France., N8, 1654 (1960). A.N.Shvab, A.P.Timchenko and A.V.Gorodysky, Ukr.khim.zhurn., 40, 90 (1974). H.A.Laitinen and B.B.Bhatia, J.Electrochem.Soc., 107, 705 (1960). V.L.Cherginets, Rasplavy , N1, 62 (1991). V.I.Shapoval and O.F.Pertchik, Elektrokhimiya, 10, 1241 (1974). V.I.Shapoval, V.F.Makogon and O.F.Pertchik, Ukr.khim.zhurn., 45, 7 (1979). V.I.Shapoval and V.F.Makogon, Ukr.khim.zhurn., 45, 201 (1979).
1496
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62
Victor Cherginets
J.Picard, F.Seon and B.Tremillon, Electrochim.Acta, 22, 1291 (1978). F.Seon, J.Picard and B.Tremillon, J.Electroanalyt.Chem., 138, 315(1982). V.L.Cherginets and T.P.Rebrova, Zhurn.Fiz.Khim., 73, 687 (1999). V.L.Cherginets and T.P.Rebrova, Electrochim.Acta, 45, 471 (1999). V.I.Shapoval and O.G.Tsiklauri, Conf."Phys. chemistry and electrochemistry of molten salts and solid electrolytes", Sverdlovsk, June 5-7, 1973, Part II, Sverdlovsk,1973,pp.32-33. V.I.Shapoval, O.G.Tsiklauri and N.A.Gasviani, Soobsch.AN Gruz.SSR, 89, 101 (1978). R.Combes, J.Vedel and B.Tremillon, Electrochim.Acta, 20, 191 (1975). Y.K.Delimarsky,V.I.Shaponal and N.N.Ovsyannikova, Ukr.khim.zhurn., 43, 115 (1977). R.Combes, J.Vedel and B.Tremillon, Anal.Lett., 3, 523 (1970). F.De Andrade, R.Combes and B.Tremillon, C.R.Acad.Sci., C280, 945 (1975). R.Combes, B.Tremillon and F.De Andrade, J.Electroanalyt.Chem., 83, 297 (1987). V.L.Cherginets and V.V.Banik, Rasplavy , N1, 66 (1991). T.P.Boyarchuk, E.G.Khailova and V.L.Cherginets, Electrochim.Acta, 38, 1481 (1993). V.L.Cherginets and E.G.Khailova, Zhurn.neorg.khim., 38, 1281 (1993). V.L.Cherginets and E.G.Khailova, Electrochim.Acta, 39, 823 (1994). P.Bocage, D.Ferry and J.Picard, Electrochim.Acta, 36, 155 (1991). V.L.Cherginets and A.B.Blank, Ukr.Khim.Zhurn, 57, 936 (1991). V.E.Komarov and V.E.Krotov, Coll.Inst.Electrochem. of Ural Sci.Centre of Acad.Sci.USSR, N27, 61 (1978). N.M.Barbin, V.N.Nekrasov, L.E.Ivanovsky, P.N.Vinogradov and V.E.Petukhov, Rasplavy, N2, 117 (1990). N.M.Barbin, V.N.Nekrasov, Electrochim.Acta, 44, 4479 (1999) Y.Kaneko and H.Kojima, Denki Kagaku, 42, 304 (1974). V.L.Cherginets, Uspekhi Khimii, 66, 661 (1997). V.L.Cherginets, Electrochim.Acta, 42, 3619 (1997). H.A.Laitinen and W.E.Harris, Chemical analysis, 2nd edition, McGraw-Hill Book Company, London, 1975. F.F.Grigorenko, A.V.Molodkina, V.M.Solomaha and M.S.Slobodyanik, Visnyk Kyivskogo Universitetu.Ser.Khim., N14, 38 (1973). M.L.Deanhardt and K.H.Stern, J.Electrochem.Soc., 128, 2577 (1981). M.L.Deanhardt and K.H.Stern, J.Electrochem.Soc., 129, 2228 (1982). R.F.Watson and G.S.Perry, J.Chem.Soc.Faraday Trans., 87, 2955 (1991). E.G.Khailova,T.A.Lysenko and V.L.Cherginets, Zhurn.Neorg.khim., 38, 175 (1993). G.T.Kosnyrev, V.N.Desyatnik, N.A.Kern and E.N.Nosonova, Rasplavy , N2, 121 (1990). R.Combes, M.N.Levelut and B.Tremillon, Electrochim.Acta, 23, 1291 (1978). Y.Ito, H.Kotohda, J.Uchida and S.Yoshizawa, J.Chem.Soc.Jap. Chem.and Ind.Chem., 6, 1034 (1982). D.Ferry,Y.Castrillejo and G.Picard, Electrochim.Acta, 34, 313 (1989). G.N.Kucera and M.-L.Saboungi, Met.Trans., B7, 213 (1976). V.L.Cherginets and T.P.Rebrova, Zhurn.Fiz.Khim., 74, 244 (2000). V.L.Cherginets and T.P.Rebrova, Electrochem.Commun., 1, 590 (1999). Y.Ito,H.Hayashi,Y.Itoh and S.Yoshizawa, Bull.Chem.Soc.Jpn., 58 3172 (1985). O.V.Demirskaya,E.G.Khailova and V.L.Cherginets, Zhurn.Fiz.khim., 69, 1658 (1995). V.L.Cherginets, E.G.Khailova and O.V.Demirskaya, Electrochim.Acta, 41, 463 (1996). V.L.Cherginets and E.G.Khailova, Ukr.Khim.Zhurn.,62, 90 (1996). V.L.Cherginets, E.G.Khailova and O.V.Demirskaya, Zhurn.Fiz.khim.,1997, 371 (1997).
21.4 Alternative cleaning technologies
1497
21.4 ALTERNATIVE CLEANING TECHNOLOGIES/DRYCLEANING INSTALLATIONS Kaspar D. Hasenclever Kreussler & Co.GmbH, Wiesbaden, Germany
21.4.1 DRYCLEANING WITH LIQUID CARBON DIOXIDE (LCD) 21.4.1.1 Basics The use of compressed CO2 for dissolving oils and fats from different substrates under industrial conditions has been published in German journals since 1982 (Quirin, KW; FSA 84; 460-468). The activity of surfactants under pressure and the formulation of reverse micelles has been published by Johnston, Lemert and McFann in Am. Chem. Soc. Ser. 406 (1989). The structure of reverse micelle and microemulsion phases in near critical and supercritical fluid as determined from dynamic light scattering studies has been published by Johnston and Penninger in: Supercritical Fluid Science and Technology, ACS Symp. Ser. Washington DC, 1989. The use of perfluoropolyether microemulsions in liquid and supercritical CO2 has been published by Chittofrati, Lenti, Sanguinetti, Visca, Gambi, Senatra and Zhou in Progr. Colloid & Polym. Sci., 79, 218-225 (1989). A process for cleaning or washing of clothing in liquid and supercritical CO2 is the issue of the German Patent DP 39 04 514 A1 by Schollmeyer and Knittel of 23.08.1990. CO2 is a slightly toxic, colorless gas with a pungent, acid smell. It will not burn or support combustion. The gas is 1.4 times heavier than air and sublimes at atmospheric pressure at minus 78°C. CO2 is not corrosive to steel, as long as it is free of water. With water it reacts to H2CO3, which can cause rapid corrosion to steel. Chromesteel or aluminum should be used if contact with water is unavoidable. CO2 will react violently with strong bases, ammonia and amines. The critical data of CO2 are: pressure 73.81 bar temperature 31.3°C volume 0,096 l density 0.468 g/ml Above Tcrit, CO2 cannot be liquefied not even under highest pressure. Above critical data substances are in supercritical condition. (Supercritical CO2 = ScCD). At 20°C and 55.4 bar CO2 is liquid (LCD). The physical properties are: density 0.77 g/ml viscosity 0.1 mPas surface tension 5 mN/m solvent power about 20 (K.B.) LCD is a solvent for apolar substances. Its activity can be widened by combination with surfactants. Micro-emulsions in LCD can be created with different surfactants (AOT/ F-Surf.) and water. CO2 is a natural resource, non-flammable, non-smog producing, physically stable. LCD can be stored and transported under pressure without harm. Containers must be treated
1498
Kaspar D. Hasenclever
in accordance to the national regulations (in Germany Druckbehälter Verordnung certification). Tanks and/or containers for LCD have three to four times the weight of their net capacity. 21.4.1.2 State of the art The stability of textiles, textile dyes, buttons, zips, interlinings under influence of high pressure in LCD and ScCD was tested at DTNW, Krefeld, DWI/TH Aachen. Both, LCD and ScCD, if they are pure, do not harm textiles and dyes. After treatment in ScCD buttons and plastic zips are destroyed, when the decompression of CO2 runs fast. LCD and ScCD are able to penetrate into apolar polymers such as polyester and polyamide, and plasticize the material so that dispersed substances can migrate into them. This behavior is used for dyeing processes with dispersion dyestuffs in ScCD, which allows textile dyeing without waste. The same procedure can cause greying in drycleaning, if resolved pigments are dispersed in LCD cleaning fluids. In Journal of Supercritical Fluids, (1990), 3, 51-65 Consani and Smith of Battelle Pacific Northwest Laboratories report the Observation on the Solubility of Surfactants and Related Molecules in Carbon Dioxide at 50°C. Nearly all known surfactants are classified. In US patent 5,467,492 Hughes Aircraft Company claims a dry cleaning in LCD with the distinguishing feature of a non moving basket together with jet agitation to the load by current circulation of the LCD cleaning fluid. In order to reach this aim, the pressure vessel contains a cylindrical perforated basket to take the load. After the pressure vessel is totally filled with LCD, the load is then set into motion and agitated by high velocity fluid jets. A dry cleaning machine presented at Las Vegas in 1997 was equipped with a 135 l pressure cylinder and the capability to clean a load of textiles of about 10 kg. At the same exhibition Miccel, Technologies, Inc.; North Carolina State University presented the MiCARE Garment Cleaning Fluid System, which consists of LCD together with patented surfactants as cleaning fluid and a dry cleaning machine for LCD, equipped with a moving basket, creating agitation in the conventional manner. In Science, Feb. 1996, E. Goldbaum published a report on the development of a team of scientists of the University of Texas, the University of Nottingham and the University of Colorado, using water in CO2 microemulsions with fluorinated surfactants in place of conventional solvents, such as chlorinated hydrocarbons or hydrocarbons. This research was funded by a Department of Energy grant of the US. The Research Institute Hohenstein together with other German Research Institutes and industrial Partners (Kreussler) are working on a basic research project, to define the interactions between LCD, textiles, surfactants, dissolved, emulsified and/or dispersed matter in this system. The Dutch Research Institute TNO, Delft works together with industrial partners on a project to develop the complete background for textile cleaning in LCD under practical conditions. As part of this project Kreussler is responsible for the research and development of detergents and the cleaning process. 21.4.1.3 Process technology Outer garments - apparel - in average of European countries contain about 15 g/kg soiling, when they are brought to dry cleaning. This “soiling” contains: 50% pigments aerosols, carbon black, iron oxides, dust; 30% water soluble perspiration, salt, body excrements;
21.4 Alternative cleaning technologies
1499
10% polymers albumin (blood, milk), starch (food); 10% solvent soluble oils, fat, wax, grease. The main problem in dry cleaning is the removal of pigments, water soluble material and polymers. Apolar solvents as perc, hydrocarbons or LCD are not able to achieve this. Drycleaning detergents (DD) will widen the activity of the drycleaning process in removing pigments and polar substances. In order to gain a maximal benefit from DD, the process technology must meet the special requirements. DD’s offer dispersing and emulsifying activity to solvents, they activate water additions into micro-emulsions in order to achieve removal of polar matter in apolar solvent. To optimize this action, a multi bath process is used, which works under the conditions of extraction from high to low contaminated cleaning fluids. Regeneration of cleaning fluids is realized by filtration and distillation. For filtration a drum-pump-filter-drum circuit is used. For distillation a distilling vessel, condensing equipment and clean solvent tank are necessary. The dimensions must be constructed with regard to the requirements of the quantity of removed “soil” during a full working day, which means with a 10 kg machine 750 g pigments, 600 g “salts” and polymers, 100 g oils and fats, 1000 g DD and about 10 l of water. In order to save costs, the cleaning baths are as short as possible, which means, about 3 l of solvent per kg load, so that per process one full bath can be distilled. In order to separate clean, average and high contaminated cleaning fluid, most drycleaning machines are equipped with 3 tanks, a filter and a distilling vessel with the capacity of the biggest tank. In order to build less expensive machines, sometimes drycleaning machines are offered which are equipped with insufficient size for distilling and less than 3 tanks. In this case the possible cleaning result will not meet the necessary hygienic and esthetic requirements of customers. The same rules apply to LCD as to conventional solvents. A one tank machine and distilling from tank to tank instead from cage to clean tank, combined with an insufficient filter size, cannot meet the minimum requirements, even if the solvent is LCD. 21.4.1.4 Risks The solubility of water in LCD is low (0.1%); the solubility of CO2 in water is high; at 4 bar and 20°C 1 l of water dissolves 4 l of CO2 gas, of which about 0.1 % reacts to H2CO3, a corrosive substance to iron and steel. Up to now the interaction in the system LCD, surfactants, water and textiles together with solved and dispersed contaminants has not yet been studied under practical conditions. The influence of surfactants on the equilibrium of water in the system of high pressure without airspace (free water, water in LCD emulsion, water adsorbed on textiles) is not yet known. In order to achieve the removal of polar substances from textiles, in conventional cleaning systems a desorption of moisture from textiles is necessary. If this method is compatible with LCD is not yet known. The low risk of shrinkage on natural textiles in dry cleaning is due to the fact, that the adsorption of moisture is reduced, after fibers are soaked with apolar solvent beyond condensation conditions of water. In presence of moisture within condensation conditions natural textiles will shrink in dry cleaning more than in aqueous processes. The LCD cleaning process works completely within condensation conditions of water. This means, that LCD cleaning can bear a high risk of shrinkage, in particular, when water additions are used in order to remove apolar soiling.
1500
Kaspar D. Hasenclever
The material and manufacturing costs of LCD machines are more expensive than normal drycleaning machines, and substantially more expensive than wet cleaning machines. The operating costs depend on the expenses of the distribution of LCD, which seems to need a total new network. With costs of about DM 2.50 per kg ($US 1.20/kg), LCD is more expensive than perc and similar to HCS. At the same time the consumption of LCD seems to be four times larger than perc and eight times larger than HCS. Not to mention water in wet cleaning. Present Care Labeling considers perc, HCS and water, no LCD. If any textile damage will occur, it will be the user’s responsibility. The investment climate in drycleaning is very weak, so that there is no tendency to invest into an unknown technology. 21.4.1.5 Competition LCD stands in competition to dry cleaning in TCE and in HCS as well as to wet cleaning. The following comparison is done under the supposition of a capacity of 20 kg/h for investment costs and a quantity of 100 kg for consumption costs (Table 21.4.1). All costs in DEM. The result of this comparison is very clear: most competitive is Wet Cleaning, followed by HCS dry cleaning. In order to avoid garment risks on sensitive textiles in wet cleaning, a combined installation with 70% wet cleaning and 30% HCS cleaning would be the optimum. In order to make LCD competitive, solvent costs must be reduced rapidly and basic research work has to be done in order to increase garment safety, cleaning results and the technical reliability. Table 21.4.1. Cost comparison of textile cleaning methods Costs/Properties
TCE
HCS
120,000.00
80,000.00
40,000.00
160,000.00
Space, m³
2.20
2.50
1.80
5.00
Solvent
5.40
5.80
2.00
30.00
Additives
12.00
8.00
16.00
12.00
Energy
23.80
21.40
11.20
15.00
Waste
12.50
8.00
0.00
4.00
Consumables total
53.70
43.20
29.20
61.00
Machine
Wet Clean
LCD
Reliability
good
sufficient
very good
high risk
Maintenance
good
good
very good
not known
Cleaning results
good
sufficient
very good
not known
low
very low
sufficient
not known
high risk
low risk
no risk
no risk
Garment risks Environment
21.4 Alternative cleaning technologies
1501
21.4.2 WET CLEANING The new technique results from intensive research work of Kreussler. The processing methods and in particular their translation into production-ready machine technology was developed in co-operation of Kreussler with Miele of Gütersloh. What does this new textile cleaning system look like, what prospects does it offer, what are its limitations and how it can be integrated into the practical operations of a Figure 21.4.1. Miele System Kreussler washing matextile cleaning company? chine and dryer. 21.4.2.1 Kreussler textile cleaning system For achieving drycleaning with water it is essential to use a washing machine with at least 80 cm drum diameter, centrifugal acceleration of 0.9 g-force in the wash cycle and 450 g-force during spinning (at least 1000 rpm), appropriate drum and lifter rib design, and accurate regulation of temperature and liquor flow, as well as precise control of mechanical action and dosing. Also necessary is a large-capacity tumble dryer, with drum at least 100 cm in diameter, having precisely controlled centrifugal acceleration; horizontally directed air flow, parallel to the drum axle; and precision electronics to regulate temperature and residual moisture level. These requirements, comprising washer-extractor and tumble dryer together with other important know-how, are provided by the “Miele System Kreussler”. Miele System Kreussler The washer-extractor (Miele WS 5220 TR), with a drum volume of 220 litres, has a loading capacity of 8-10 kg for delicate outerwear, 12-15 kg of ordinary outerwear and 20-22 kg of normal textiles designated washable by the care label. The special outerwear cleaning is carried out using the LANADOL process, for which approx. 140 litres of water are used per load. Cycle time is about 25 minutes. The dryer (Miele 6559 TR) has a drum diameter of 110 cm, with a 550 litre volume. The drying air flows through the load horizontally - parallel to the drum axle - from the rear towards the door, thus achieving optimum evaporation level with a short drying time. The necessary temperature progression and precise residual moisture is ensured with precision electronics. A significant aspect is that not only the LANADOL process but all other wash programs can be carried out in the Miele System Kreussler machines, providing comprehensive textile cleaning facilities. LANADOL processing technique Most textiles offered to drycleaners for processing can be handled with three basic processes: • Extra: Garments without wash symbol • Normal: Garments with 30/40°C wash symbol • Proofing: Poplin and sportswear
1502
Kaspar D. Hasenclever
Figure 21.4.2. (left) Polyester/wool trousers after drying.(right) Dosing method for polyester/wool trousers after drying LANADOL AKTIV and LANADOL APRET.
Figure 21.4.3. Knitwear and silk, unfinished after drying.
Within each of these processing groups, the work should be sorted into light/medium/heavy weight, the best load combination consisting of items with the same or similar drying characteristics. It would be preferable, too, to differentiate roughly between light and dark colors. Heavily soiled or grease stained areas should be pre-treated by simple pre-spotting with LANADOL AVANT, using PRENETT A-B-C for special staining. Where non-color-fastness is suspected, a dye-fastness check (seam test) should be made with a damp white cotton cloth. After the relevant program is selected, the process runs fully automatically. The exact amount required of the special cleaning and fibre protection agent LANADOL AKTIV is added automatically at precisely the right time. A fibre protection agent, providing retexturing and antistatic finish for drying, is also added automatically. Following spin-drying, woollen textiles have a particularly low residual moisture content of about 30%. After selection of the program suitable for the
21.4 Alternative cleaning technologies
1503
Figure 21.4.4 Down-filled anoraks and ski clothing, washed... ... and impregnated.
load make-up, the tumbler will dry to the pre-determined residual moisture level within 15 25 minutes. With garments structured in several layers - such as men's jackets, ladies' suit jackets and coats - further drying at room temperature or in a drying cabinet is beneficial, before finishing on a hot air - steam garment former. All other garments can be finished normally. 21.4.2.2 Possibilities We will not dwell here on the possibilities for processing textiles which are washable according to their care labels. On removal from the dryer, the items are quite smooth, needing little effort to finish. Treatment of washable outerwear with the LANADOL process is so problem-free, so simple and rapid, the work is so clean, bright and fresh, with such a pleasant handle and freedom from static, that one would not wish to handle this classification in any other way. One immediately realizes that conventional drycleaning would be just second best for such textiles. Customers will also come to appreciate this. Even more interesting, however, is the use of the LANADOL process for textiles which are not washable according to their care label. According to current processing knowledge, the progression from “ideal” to “difficult” garments can be listed as follows: Silk, knitwear, cotton, wool, viscose, linen and structured garments (jackets). “Difficult” in this context refers mainly to finishing requirements. It can be quite possible that a man’s jacket, of which the interlining has shrunk, could in addition to treatment on the garment former, need additional hand ironing for up to 10 minutes for perfect results. Usually, however, suit jackets can be processed without problems, although not particularly fast. Without doubt, men’s and ladies’ jackets can be drycleaned faster, simpler and with fewer problems. After the LANADOL process, however, they will be fresher, cleaner and nicer to wear. Other classifications which are not indicated by the care label to be suitable for domestic washing can generally be treated without problems by the LANADOL process, with which processing risks can in most cases be assessed as less than for drycleaning. Of special interest is treatment by the LANADOL process of textiles which are problematic for drycleaning. Down-filled anoraks, raincoats, impregnated goods - and also covers for rheumatism sufferers, bulky textiles and glittery items. The LANADOL process is
1504
Kaspar D. Hasenclever
ideal for these - quick and safe with excellent results. Wet proofing is just in a class of its own. Bulky textiles without sweals, without perc residues, without peculiar odors but clean, fresh, fluffy and soft, and all after a short process demonstrate particularly well LANADOL processing’s superiority for this group of classifications. 21.4.2.3 Limitations Where there is light, there is also shadow. To evaluate the new system, coming to terms with the shadowy side is vital. There are darker aspects to the method in the high demands it places on operators and the limits, often not obvious, to its care possibilities for certain textiles. System limits The compromise necessary to achieve low textile shrinkage, good cleaning results and reasonable processing time implies a process-technology tightrope walk where even small deviations from the norm can lead to a fall. For this, read damaged textiles. It is essential, therefore, to adhere to all the specified parameters. This involves not only the equipment together with the types and amounts of chemicals used, but also application of the correct process for the classification to be treated. At the moment, we will not risk even a transfer of the process to other machine sizes of the same make. Skill limits To sort work into classifications for the appropriate process methods, personnel with knowledge of textiles gained after comprehensive training in textile cleaning are required. Furthermore, experienced finishing personnel are needed, particularly if textiles not washable according to their care labels are to be treated with the LANADOL process, even more so if jackets are to be cleaned. Limits set by textiles The main problem item, the jacket, has already been mentioned several times. Problems can be caused by: • Seam shrinkage • Interlining shrinkage • Lining shrinkage In most cases seam shrinkage can be remedied by expert finishing methods. Remedying shrunk interlining can be simple or very difficult. The amount of finishing work required will vary considerably, according to the adhesive and fabric construction used. Pure viscose lining material can shrink considerably in some cases. If the prescribed residual moisture level was kept to precisely during the drying process, this shrinkage can be remedied. However, if there was over-drying then recovery is frequently impossible. If all three eventualities have occurred on the same jacket, then it is a goner. On the other hand, this really happens very rarely. The limits outlined here accurately reflect experiences to date. With wider use, the catalogue might possibly grow more extensive - but so also would the experience to deal with such problems. 21.4.2.4 Adapting to working practices The LANADOL process is ideal to supplement an existing drycleaning process, so that one can readily envisage the proportion growing a step at a time. With increasing confidence in this processing method, soon not only those textiles which are washable according to their care label will be cleaned in water, but also more general outerwear. It is then only a question of time in a textile care company before most work will be cleaned with water and the
21.4 Alternative cleaning technologies
1505
smaller part, comprising articles requiring more complicated treatment, will be treated with solvent. Not ideal, but nevertheless possible, is a situation where an attractive plant location uses the LANADOL process exclusively, should local conditions prohibit use of solvent equipment or permit it only with great difficulty. In this case, with a well equipped finishing department and qualified operators, a drycleaner’s complete service range is feasible, with possibly just very few exceptions. 21.4.3 FUTURE Dry cleaning with TCR is at the end of its development. HCS dry cleaning at present is in a phase of consolidation. Wet cleaning is in its early beginnings and LCD cleaning isn’t in practice yet. TCE will probably soon be rendered a “knock out property” by its environmental behavior. Also the operating costs will increase rapidly because of more specific regulations with regard to storage and handling as well as to the waste removal of TCE. Today 80% of dry cleaning is done in TCE. Possibly more than 90% of existing TCE machines will be replaced by other systems within the next 10 years. HCS has no probable “knock out property”. Possible new developments may reduce the solvent consumption and increase the cleaning results. Because of low operating costs HCS will be the most important replacement for perc machines in the near future. Wet Cleaning has no “knock out property”. New developments will reduce the garment risks, the finishing work and the operating speed. Minimizing the water consumption is already solved technically though these technologies have not been put to use due to the low cost availability of water resources. New services to customers and the revitalization of professional textile care will move on with Wet Cleaning. Most of the new operations will be installed with wet cleaning equipment only or with a combination of wet- and HCS cleaning. LCD bears the risk of three “knock out properties”: reliability, cleaning result, and garment damage. In order to solve the problems connected with the removal of polar “soiling” and the prevention against greying on synthetic fibers and shrinking of natural fibers, a lot of basic research has to be done. Additionally a lot of developmental work has to be done in order to make LCD economically competitive to existing processes. That means not only a reduction of LCD consumption without increasing the process time and the use of energy, but also build up of a simple distribution of this solvent.
22
Solvent Recycling, Removal, and Degradation 22.1 ABSORPTIVE SOLVENT RECOVERY Klaus-Dirk Henning CarboTech Aktivkohlen GmbH, Essen, Germany
22.1.1 INTRODUCTION A variety of waste gas cleaning processes1-4 are commercially available. They are used to remove organic vapors (solvents) from air emissions from various industries and product lines. The main applications of solvents in various industries are given in Table 22.1.1. Table 22.1.1. Application of organic solvents. [After references 2-4] Application
Solvents
Coating plants: Adhesive tapes, Mag- Naphtha, toluene, alcohols, esters, ketones (cyclohexanone, methyl netic tapes, Adhesive films, Photo- ethyl ketone), dimethylformamide, tetrahydrofuran, chlorinated hydrographic papers, Adhesive labels carbons, xylene, dioxane Chemical, pharmaceutical, food in- Alcohols, aliphatic, aromatic and halogenated hydrocarbons, esters, kedustries tones, ethers, glycol ether, dichloromethane Fibre production Acetate fibers Viscose fibers Polyacrylonitrile
Acetone, ethanol, esters carbon disulfide dimethylformamide
Synthetic leather production
Acetone, alcohols, ethyl ether
Film manufacture
Alcohols, acetone, ether, dichloromethane
Painting shops: Cellophane, Plastic Amyl acetate, ethyl acetate, butyl acetate, dichloromethane, ketones (acfilms, Metal foils, Vehicles, Hard pa- etone, methyl ethyl ketone), butanol, ethanol, tetrahydrofuran, toluene, pers, Cardboard, Pencils benzene, methyl acetate Cosmetics industry
Alcohols, esters
Rotogravure printing shops
Toluene, xylene, heptane, hexane
Textile printing
Hydrocarbons
Drycleaning shops
Hydrocarbons, tetrachloroethane
1508
Klaus-Dirk Henning
Application
Solvents
Sealing board manufacture
Hydrocarbons, toluene, trichloroethane
Powder and explosives production
Ether, alcohol, benzene
Degreasing and scouring applications Metal components Hydrocarbons, trichloroethylene, tetrachloroethylene, dichloromethane Leather Tetrachloroethylene Wool Tetrachloroethane Loading and reloading facilities (tank Gasoline, benzene, toluene, xylene venting gases)
The basic principles of some generally accepted gas cleaning processes for solvent removal are given in Table 22.1.2. For more than 70 years adsorption processes using activated carbon, in addition to absorption and condensation processes, have been used in adsorptive removal and recovery of solvents. The first solvent recovery plant for acetone was commissioned for economic reason in 1917 by Bayer. In the decades that followed solvent recovery plants were built and operated only if the value of the recovered solvents exceeded the operation costs and depreciation of the plant. Today such plants are used for adsorptive purification of exhaust air streams, even if the return is insufficient, to meet environmental and legal requirements. Table 22.1.2. Cleaning processes for the removal and recovering of solvents Process
Basic principle
Remarks
Incineration
The solvents are completely destroyed by: thermal oxidation (>10 g/m3) Only waste heat recovery possible catalytic oxidation (3-10 g/m3)
Condensation
High concentration of solvents (>50 g/m3) - Effective in reducing heavy emissions, are lowered by direct or indirect conden- - good quality of the recovered solvents, sation with refrigerated condensers - difficult to achieve low discharge levels
Absorption
Scrubbers using non-volatile organics as scrubbing medium for the solvent removal. Desorption of the spent scrubbing fluid
Advantageous if: - scrubbing fluid can be reused directly without desorption of the absorbed solvent - scrubbing with the subcooled solvent itself possible
Membrane permeation processes are principle suited for small volumes with high Membrane permeation concentrations e.g. to reduce hydrocarbon3 concentration from several hundred g/m down to 5-10 g/m3
To meet the environmental protection requirements a combination with other cleaning processes (e.g. activated carbon adsorber) is necessary
Biological treatment
Biological decomposition by means of Sensitive against temperature and solvent bacteria and using of bio-scrubber or concentration bio-filter
Adsorption
Adsorption of the solvents by passing the waste air through an adsorber filled with activated carbon. Solvent recovery by steam or hot gas desorption
By steam desorption the resulting desorbate has to be separated in a water and an organic phase using a gravity separator (or a rectification step)
This contribution deals with adsorptive solvent removal2,12,13 and recovery and some other hybrid processes using activated carbon in combination with other purification steps.
22.1 Absorptive solvent recovery
1509
Due to more stringent environmental protection requirements, further increases in the number of adsorption plants for exhaust air purification may be expected. 22.1.2 BASIC PRINCIPLES 22.1.2.1 Fundamentals of adsorption To understand the adsorptive solvent recovery we have to consider some of the fundamentals of adsorption and desorption1-9 (Figure 22.1.1). Adsorption is the term for the enrichment of gaseous or dissolved substances on the boundary surface of a solid (the adsorbent). The surfaces of adsorbents have what we call active sites where the binding forces between the individual atoms of the solid structure are not completely saturated by neighboring atoms. These active sites can bind foreign molecules which, Figure 22.1.1. Adsorption terms (after reference 2). when bound, are referred to as adsorpt. The overall interphase boundary is referred to as adsorbate. The term, desorption, designates the liberation of the adsorbed molecules, which is a reversal of the adsorption process. The desorption process generates desorbate which consists of the desorpt and the desorption medium.2 A distinction is made between two types of adsorption mechanisms: chemisorption and physisorption. One, physisorption, is reversible and involves exclusively physical interaction forces (van der Waals forces) between the component to be adsorbed and the adsorbent. The other, chemisorption, is characterized by greater interaction energies which result in a chemical modification of the component adsorbed along with its reversible or irreversible adsorption.2 Apart from diffusion through the gas-side boundary film, adsorption kinetics are predominantly determined by the diffusion rate through the pore structure. The diffusion coefficient is governed by both the ratio of pore diameter to the radius of the molecule being adsorbed and other characteristics of the component being adsorbed. The rate of adsorption is governed primarily by diffusional resistance. Pollutants must first diffuse from the bulk gas stream across the gas film to the exterior surface of the adsorbent (gas film resistance). Due to the highly porous nature of the adsorbent, its interior contains by far the majority of the free surface area. For this reason, the molecule to be adsorbed must diffuse into the pore (pore diffusion). After their diffusion into the pores, the molecules must then physically adhere to the surface, thus losing the bulk of their kinetic energy. Adsorption is an exothermic process. The adsorption enthalpy, decreases as the load of adsorbed molecules increases. In activated carbon adsorption systems for solvent recovery, the liberated adsorption enthalpy normally amounts to 1.5 times the evaporation enthalpy at the standard working capacities which can result in a 20 K or more temperature increase. In the process, exothermic adsorption mechanisms may coincide with endothermic desorption mechanisms.2
1510
Klaus-Dirk Henning
22.1.2.2 Adsorption capacity The adsorption capacity (adsorptive power, loading) of an adsorbent resulting from the size and the structure of its inner surface area for a defined component is normally represented as a function of the component concentration c in the carrier gas for the equilibrium conditions at constant temperature. This is known as the adsorption isotherm x = f (c )T. There are variety of approaches proceeding from different model assumptions for the quantitative description of adsorption isotherms (Table 22.1.3) which are discussed in detail in the literature.2,9,11 Table 22.1.3. Overview of major isotherm equations (After reference 2) Adsorption isotherm by
Isotherm equation
Notes
Freundlich
x = K × pin
low partial pressures of the component to be adsorbed
Langmuir
x=
xmaxbpi 1 + bpi
homogeneous adsorbent surface, monolayer adsorption
Brunauer, Emmet, Teller (BET)
1 pi C − 1 pi = + V ( ps − pi ) VMBC VMBC ps
Dubinin, Raduskevic
RT p log s log V = log Vs − 2.3 pi ε β o
homogeneous surface, multilayer adsorption, capillary condensation n
potential theory; n = 1,2,3; e.g., n = 2 for the adsorption of organic solvent vapors on micro-porous activated carbon grades
where: K, n - specific constant of the component to be adsorbed; x - adsorbed mass, g/100 g; xmax - saturation value of the isotherm with monomolecular coverage, g/100 g; b - coefficient for component to be adsorbed, Pa; C - constant; T - temperature, K; pi - partial pressure of component to be adsorbed, Pa; ps - saturation vapor pressure of component to be adsorbed, Pa; V - adsorpt volume, ml; VMB - adsorpt volume with monomolecular coverage, ml; Vs - adsorpt volume on saturation, ml; VM - molar volume of component to be adsorbed, l/mol; R - gas constant, J(kg K); εo - characteristic adsorption energy, J/kg; β - affinity coefficient
The isotherm equation developed by Freundlich on the basis of empirical data can often be helpful. According to the Freundlich isotherm, the logarithm of the adsorbent loading increases linearly with the concentration of the component to be adsorbed in the carrier gas. This is also the result of the Langmuir isotherm which assumes that the bonding energy rises exponentially as loading decreases. However, commercial adsorbents do not have a smooth surface but rather are highly porous solids with a very irregular and rugged inner surface. This fact is taken into account by the potential theory which forms the basis of the isotherm equation introduced by Dubinin. At adsorption temperatures below the critical temperature of the component to be adsorbed, the adsorbent pores may fill up with liquid adsorpt. This phenomenon is known as capillary condensation and enhances the adsorption capacity of the adsorbent. Assuming cylindrical pores, capillary condensation can be quantitatively described with the aid of the Kelvin equation, the degree of pore filling being inversely proportional to the pore radius. If several compounds are contained in a gas stream the substances compete for the adsorption area available. In that case compounds with large interacting forces may displace less readily adsorbed compounds from the internal surface of the activated carbon.
22.1 Absorptive solvent recovery
1511
Figure 22.1.2. Idealized breakthrough curve of a fixed bed adsorber.
As previously stated, the theories of adsorption are complex, with many empirically determined constants. For this reason, pilot data should always be obtained on the specific pollutant adsorbent combination prior to full-scale engineering design. 22.1.2.3 Dynamic adsorption in adsorber beds For commercial adsorption processes such as solvent recovery not only the equilibrium load, but also the rate at which it is achieved (adsorption kinetics) is of decisive importance.9-15 The adsorption kinetics is determined by the rates of the following series of individual steps: • transfer of molecules to the external surface of the adsorbent (border layer film diffusion) • diffusion into the particle • adsorption The adsorbent is generally in a fixed adsorber bed and the contaminated air is passed through the adsorber bed (Figure 22.1.2). When the contaminated air first enters through the adsorber bed most of the adsorbate is initially adsorbed at the inlet of the bed and the air passes on with little further adsorption taking place. Later, when the inlet of the absorber becomes saturated, adsorption takes place deeper in the bed. After a short working time the activated carbon bed becomes divided into three zones: • the inlet zone where the equilibrium load corresponds to the inlet concentration, co • the adsorption or mass transfer zone, MTZ, where the adsorbate concentration in the air stream is reduced • the outlet zone where the adsorbent remains incompletely loaded In time the adsorption zone moves through the activated carbon bed.18-21 When the MTZ reaches the adsorber outlet zone, breakthrough occurs. The adsorption process must be stopped if a pre-determined solvent concentration in the exhaust gas is exceeded. At this
1512
Klaus-Dirk Henning
point the activated carbon bed has to be regenerated. If, instead the flow of contaminated air is continued on, the exit concentration continues to rise until it becomes the same as the inlet concentration. It is extremely important that the adsorber bed should be at least as long as the mass transfer zone of the component to be adsorbed. The following are key factors in dynamic adsorption and help determine the length and shape of the MTZ: • the type of adsorbent • the particle size of the adsorbent (may depend on maximum allowable pressure drop) • the depth of the adsorbent bed • the gas velocity • the temperature of the gas stream and the adsorbent • the concentration of the contaminants to be removed • the concentration of the contaminants not to be removed, including moisture • the pressure of the system • the removal efficiency required • possible decomposition or polymerization of contaminants on the adsorbent 22.1.2.4 Regeneration of the loaded adsorbents In the great majority of adsorptive waste gas cleaning processes, the adsorpt is desorbed after the breakthrough loading has been attained and the adsorbent reused for pollutant removal. Several aspects must be considered when establishing the conditions of regeneration for an adsorber system.2-4,6,13-16 Very often, the main factor is an economic one, to establish that an in-place regeneration is or is not preferred to the replacement of the entire adsorbent charge. Aside from this factor, it is important to determine if the recovery of the contaminant is worthwhile, or if only regeneration of the adsorbent is required. The process steps required for this purpose normally dictate the overall concept of the adsorption unit. Regeneration, or desorption, is usually achieved by changing the conditions in the adsorber to bring about a lower equilibrium-loading capacity. This is done by either increasing the temperature or decreasing the partial pressure. For regeneration of spent activated carbon from waste air cleaning processes the following regeneration methods are available (Table 22.1.4). Table 22.1.4. Activated carbon regeneration processes (After references 2,16) Regeneration method
Pressure-swing process
Temperature-swing process: Desorption
Reactivation
Principle Alternating between elevated pressure during adsorption and pressure reduction during desorption
Steam desorption or inert gas desorption at temperatures of < 500°C Partial gasification at 800 to 900°C with steam or other suitable oxidants
Application examples Gas separation, e.g., N2 from air, CH4 from biogas, CH4 from hydrogen
Special features Raw gas compression; lean gas stream in addition to the product gas stream
Solvent recovery, process Reprocessing of desorbate waste gas cleanup All organic compounds Post-combustion, if readsorbed in gas cleaning quired scrubbing of flue applications gas generated
22.1 Absorptive solvent recovery
Regeneration method Extraction with: Organic solvents
Caustic soda solution Supercritical gases
Principle
1513
Application examples
Extraction with carbon Sulphur extraction disulfide or other solvents Sulfosorbon process Percolation with caustic Phenol-loads activated carbon soda e.g. extraction with super- Organic compounds critical CO2
Special features Desorbate treatment by distillation, steam desorption of solvent Phenol separation with subsequent purging Separation of CO2/organic compounds
• Pressure-swing process • Temperature-swing process • Extraction process In solvent recovery plants, temperature-swing processes are most frequently used. The loaded adsorbent is direct heated by steam or hot inert gas, which at the same time serves as a transport medium to discharge the desorbed vapor and reduce the partial pressure of the gas-phase desorpt. As complete desorption of the adsorpt cannot be accomplished in a reasonable time in commercial-scale systems, there is always heel remaining which reduces the adsorbent working capacity. Oxidic adsorbents offer the advantage that their adsorptive power can be restored by controlled oxidation with air or oxygen at high temperatures. 22.1.3 COMMERCIALLY AVAILABLE ADSORBENTS Adsorbers used for air pollution control and solvent recovery predominantly employ activated carbon. Molecular sieve zeolites are also used. Polymeric adsorbents can be used but are seldom seen. 22.1.3.1 Activated carbon Activated carbon1-8 is the trade name for a carbonaceous adsorbent which is defined as follows: Activated carbons5 are non-hazardous, processed, carbonaceous products, having a porous structure and a large internal surface area. These materials can adsorb a wide variety of substances, i.e., they are able to attract molecules to their internal surface, and are therefore called adsorbents. The volume of pores of the activated carbons is generally greater than 0.2 mlg-1. The internal surface area is generally greater than 400 m2g-1. The width of the pores ranges from 0.3 to several thousand nm. All activated carbons are characterized by their ramified pore system within which various mesopores (r = 1-25 nm), micropores (r = 0.4-1.0 nm) and submicropores (r < 0.4 nm) each of which branch off from macropores (r > 25 nm). Figure 22.1.3. shows schematically the pore system important for adsorption and desorption.17 The large specific surfaces are created predominantly by the micropores. Activated carbon is commercially available in shaped (cylindrical pellets), granular, or powdered form. Due to its predominantly hydrophobic surface properties, activated carbon preferentially adsorbs organic substances and other non-polar compounds from gas and liquid phases. Activated alumina, silica gel and molecular sieves will adsorb water preferentially from a gas-phase mixture of water vapor and an organic contaminant. In Europe cylindrically-shaped activated carbon pellets with a diameter of 3 or 4 mm are used for solvent recovery, because they assure a low pressure drop across the adsorber system. Physical and
1514
Klaus-Dirk Henning
Figure 22.1.3. Schematic activated carbon model (After reference 17).
chemical properties of three typical activated carbon types used for solvent recovery are given in Table 22.1.5. Table 22.1.5 Activated carbon types for waste air cleaning (After reference 2) Applications (typiAdsorbent cal examples)
Compacted density, 3 kg/m
Pore volume for pore size d < 20 nm, ml/g
Pore volume for pore size d > 20 nm, ml/g
Specific surface 2 area, m /g
Specific heat capacity, J/kgK
Activated carbon, fine-pore
Intake air and exhaust air cleanup, odor control, Adsorption of low-boiling hydrocarbons
400 - 500
0.5 - 0.7
0.3 - 0.5
1000-1200
850
Activated carbon, mediumpore
Solvent recovery, Adsorption of medium-high boiling hydrocarbons
350 - 450
0.4 - 0.6
0.5 - 0.7
1200-1400
850
Activated carbon, wide-pore
Adsorption and recovery of high-boiling hydrocarbons
300 - 400
0.3 - 0.5
0.5 - 1.1
1000-1500
850
22.1.3.2 Molecular sieve zeolites Molecular sieve zeolites2,10,11 are hydrated, crystalline aluminosilicates which give off their crystal water without changing their crystal structure so that the original water sites are free for the adsorption of other compounds. Activation of zeolites is a dehydration process accomplished by the application of heat in a high vacuum. Some zeolite crystals show behavior opposite to that of activated carbon in that they selectively adsorb water in the presence of nonpolar solvents. Zeolites can be made to have specific pore sizes that impose limits on the size and orientation of molecules that can be adsorbed. Molecules above a specific size cannot enter the pores and therefore cannot be adsorbed (steric separation effect).
22.1 Absorptive solvent recovery
1515
Substitution of the greater part of the Al2O3 by SiO2 yields zeolites suitable for the selective adsorption of organic compounds from high-moisture waste gases. 22.1.3.3 Polymeric adsorbents The spectrum of commercial adsorbents2 for use in air pollution control also includes beaded, hydrophobic adsorber resins consisting of nonpolar, macroporous polymers produced for the specific application by polymerizing styrene in the presence of a crosslinking agent. Crosslinked styrene divinylbenzene resins are available on the market under different trade names. Their structure-inherent fast kinetics offers the advantage of relatively low desorption temperatures. They are insoluble in water, acids, lye and a large number of organic solvents. 22.1.4 ADSORPTIVE SOLVENT RECOVERY SYSTEMS Stricter regulations regarding air pollution control, water pollution control and waste management have forced companies to remove volatile organics from atmospheric emissions and workplace environments. But apart from compliance with these requirements, economic factors are decisive in solvent recovery. Reuse of solvent in production not only reduces operating cost drastically but may even allow profitable operation of a recovery system. 22.1.4.1 Basic arrangement of adsorptive solvent recovery with steam desorption Adsorptive solvent recovery units have at least two, but usually three or four parallel-connected fix-bed adsorbers which pass successively through the four stages of the operation cycle.1-4,18-23 • Adsorption • Desorption • Drying • Cooling Whilst adsorption takes place in one or more of them, desorption, drying and cooling takes place in the others. The most common adsorbent is activated carbon in the shape of 3 or 4 mm pellets or as granular type with a particle size of 2 to 5 mm (4 x 10 mesh). A schematic flow sheet of a two adsorber system for the removal of water-insoluble solvents is shown in Fig 22.1.4. The adsorber feed is pre-treated if necessary to remove Figure 22.1.4. Flow sheet of a solvent recovery system with steam solids (dust), liquids (droplets or desorption.
1516
Klaus-Dirk Henning
Figure 22.1.5. Principle of steam desorption.
aerosols) or high-boiling components as these can hamper performance. Frequently, the exhaust air requires cooling. To prevent an excessive temperature increase across the bed due to the heat of adsorption, inlet solvent concentrations are usually limited to about 50 g/m3. In most systems the solvent-laden air stream is directed upwards through a fixed carbon bed. As soon as the maximum permissible breakthrough concentration is attained in the discharge clean air stream, the loaded adsorber is switched to regeneration. To reverse the adsorption of the solvent, the equilibrium must be reversed by increasing the temperature and decreasing the solvent concentration by purging. For solvent desorption direct steaming of the activated carbon bed is the most widely used regeneration technique because it is cheap and simple. Steam (110-130°C) is very effective in raising the bed temperature quickly and is easily condensed to recover the solvent as a liquid. A certain flow is also required to reduce the partial pressure of the adsorbate and carry the solvent out of the activated carbon bed. A flow diagram for steam desorption for toluene recovery is given in Figure 22.1.5. First, the temperature of the activated carbon is increased to approx. 100°C. This temperature increase reduces the equilibrium load of the activated carbon. Further reduction of the residual load is obtained by the flushing effect of the steam and the declining toluene partial pressure. The load difference between spent and regenerated activated carbon - the “working capacity” - is then available for the next adsorption cycle. The counter-current pattern of adsorption and desorption favors high removal efficiencies. Desorption of the adsorbed solvents starts after the delay required to heat the activated carbon bed. The specific steam consumption increases as the residual load of the activated carbon decreases (Figure 22.1.6). For cost reasons, desorption is not run to completion. The desorption time is optimized to obtain the acceptable residual load with a minimum specific steam consumption. The amount of steam required depends on the interaction forces between the solvent and the activated carbon. The mixture of steam and solvent vapor from the adsorber is condensed in a condenser. If the solvent is immiscible with water the condensate is led to a gravity separator (making use of the density differential) where it is separated into a aqueous and solvent fraction.
22.1 Absorptive solvent recovery
1517
Figure 22.1.6. Desorption efficiency and steam consumption.
Figure 22.1.7. Options for solvent recovery by steam desorption (After references 2, 3,12).
If the solvents are partly or completely miscible with water, additional special processes are necessary for solvent separation and treatment (Figure 22.1.7). For partly miscible solvents the desorpt steam/solvent mixture is separated by enrichment condensation, rectification or by a membrane processes. The separation of completely water-soluble solvents requires a rectification column.2 After steam regeneration, the hot and wet carbon bed will not remove organics from air effectively, because high temperature and humidity do not favor complete adsorption. The activated carbon is therefore dried by a hot air stream. Before starting the next adsorption cycle the activated carbon bed is cooled to ambient temperature with air. Typical operating data for solvent recovery plants and design ranges are given in Table 22.1.6.
1518
Klaus-Dirk Henning
Table 22.1.6. Typical operating data for solvent recovery plants Operating data
Range of design
Air velocity (m/s): Air temperature (°C): Bed height (m): Steam velocity (m/s): Time cycle for each adsorber Adsorption (h): Steaming (h): Drying (hot air, h): Cooling (cold air, h):
0.2-0.4 20-40 0.8-2 0.1-0.2 2-6 0.5-1 0.2-0.5 0.2-0.5
Solvent concentration (g/m3): Solvent adsorbed by activated carbon per cycle (% weight): Steam/solvent ratio: Energy (kWh/t solvent): Cooling water (m3/t solvent): Activated carbon (kg/t solvent):
1-20 10-20 2-5:1 50-600 30-100 0.5-1
22.1.4.2 Designing solvent recovery systems 22.1.4.2.1 Design basis1-3,18-23 Designing an activated carbon system should be based on pilot data for the particular system and information obtained from the activated carbon producer. The basic engineering and the erection of the system should be done by an experienced engineering company. Solvent recovery systems would also necessitate the specification of condenser duties, distillation tower sizes, holding tanks and piping and valves. Designing of a solvent recovery system requires the following information (Table 22.1.7). Table 22.1.7. Design basis for adsorption systems by the example of solvent recovery (After references 1-3)
Site-specific conditions
Emission sources: Production process(es), continuous/shift operation Suction system: Fully enclosed system, negative pressure control Available drawings: Building plan, machine arrangement, available space, neighborhood Operating permit as per BimSchG (German Federal Immission Control Act)
Exhaust air data (max., min., normal)
- Temperature (°C) - Volumetric flow rate (m3/h) - Relative humidity (%) - Pressure (mbar) - Solvent concentration by components (g/m3) - Concentration of other pollutants and condition (fibers, dust)
Characterization of solvents
- Molecular weight, density - Boiling point, boiling range - Critical temperature - Explosion limits - Water solubility - Corrosiveness - Stability (chemical, thermal) - Maximum exposure limits (MEL) - Safety instructions as per German Hazardous Substances Regulations (description of hazards, safety instructions, safety data sheets) - Physiological effects: irritating, allergenic, toxic, cancerogenic
Utilities (type and availability)
- Electric power - Cooling brine
Waste streams (avoidance, recycling, disposal)
- Solvent
- Steam - Cooling water - Compressed air, instrument air - Inert gas
- Steam condensate
- Adsorbent
- Filter dust
22.1 Absorptive solvent recovery
Guaranteed performance data at rated load
1519
- Emission levels: pollutants, sound pressure level (dB(A)) - Utility consumption, materials- Availability
After this information has been obtained, the cycle time of the system must be determined. The overall size of the unit is determined primarily by economic considerations, balancing low operating costs against the capital costs (smaller system → shorter cycle time; bigger adsorber → long cycle time). The limiting factor for the adsorber diameter is normally the maximum allowable flow velocity under continuous operating conditions. The latter is related to the free vessel cross-section and, taking into account fluctuations in the gas throughput, selected such (0.1 to 0.5 m/s) that it is sufficiently below the disengagement point, i.e., the point at which the adsorbent particles in the upper bed area are set into motion. Once the adsorber diameter has been selected, the adsorber height can be determined from the required adsorbent volume plus the support tray and the free flow cross-sections to be provided above and below the adsorbent bed. The required adsorbent volume is dictated by the pollutant concentration to be removed per hour and the achievable adsorbent working capacity. From the view point of adsorption theory, it is important that the bed be deeper than the length of the transfer zone which is unsaturated. Any multiplication of the minimum bed depth gives more than proportional increase in capacity. In general, it is advantageous to size the adsorbent bed to the maximum length allowed by pressure drop considerations. Usually fixed-bed adsorber has a bed thickness of 0.8 - 2 m. 22.1.4.2.2 Adsorber types Adsorbers for solvent recovery1-3 units (Figure 22.1.8) are build in several configurations: • vertical cylindrical fixed-bed adsorber • horizontal cylindrical fixed-bed adsorber • vertical annular bed adsorber • moving-bed adsorber • rotor adsorber (axial or radial) For the cleanup of high volume waste gas streams, multi adsorber systems are preferred. Vertical adsorbers are most common. Horizontal adsorbers, which require less overhead space, are used mainly for large units. The activated carbon is supported on a cast-iron grid which is sometimes covered by a mesh. Beneath the bed, a 100 mm layer of 5 - 50 mm size gravel (or Al2O3 pellets) is placed which acts as a heat accumulator and its store part of the desorption energy input for the subsequent drying step. The depth of the carbon bed, chosen in accordance with the process conditions, is from 0.8 to 2 m. Adsorbers with an annular bed of activated carbon are less common. The carbon in the adsorber is contained between vertical, cylindrical screens. The solvent laden air passes from the outer annulus through the outer screen, the carbon bed and the inner screen, then exhausts to atmosphere through the inner annulus. The steam for regeneration of the carbon flows in the opposite direction to the airstream. Figure 22.1.8 illustrates the operating principle of a counter-current flow vertical moving bed adsorber. The waste gas is routed in counter-current to the moving adsorbent and such adsorption systems have a large number of single moving beds. The discharge system ensures an uniform discharge (mass flow) of the loaded adsorbent. In this moving bed system adsorption and desorption are performed in a separate column or column sections. The process relies on an abrasion resistant adsorbent with good flow proper-
1520
Klaus-Dirk Henning
Figure 22.1.8. Adsorber construction forms.
ties (beaded carbon or adsorber resins). After the adsorption stage of a multi-stage fluidized-bed system, the loaded adsorbent is regenerated by means of steam or hot inert gas in a moving-bed regenerator and pneumatically conveyed back to the adsorption section.2 When using polymeric adsorbents in a comparable system, desorption can be carried out with heated air because the structure-inherent faster kinetics of these adsorbents require desorption temperatures of only 80 to 100°C. Rotor adsorbers allow simultaneous adsorption and desorption in different zones of the same vessel without the need for adsorbent transport and the associated mechanical stresses. Rotor adsorbers can be arranged vertically or horizontally and can be designed for
22.1 Absorptive solvent recovery
1521
axial or radial gas flow, so that they offer great flexibility in the design of the adsorption system. The choice of the rotor design2,27,33 depends on the type of adsorbent employed. Rotor adsorbers accommodating pellets or granular carbon beds come as wheel type baskets divided into segments to separate the adsorption zone from the regeneration zone. Alternatively, they may be equipped with a structured carrier (e.g., ceramics, cellulose) coated with or incorporating different adsorbents such as • activated carbon fibers • beaded activated carbon • activated carbon particles • particles of hydrophobic zeolite The faster changeover from adsorption to desorption compared with fixed-bed adsorbers normally translates into a smaller bed volume and hence, in a reduced pressure loss. Hot air desorption is also the preferred method for rotor adsorbers. Hot air rates are about one tenth of the exhaust air rate. The desorbed solvents can either be condensed by chilling or disposed of by high-temperature or catalytic combustion. In some cases, removal of the concentrated solvents from the desorption air in a conventional adsorption system and subsequent recovery for reuse may be a viable option. The structural design of the adsorber is normally governed by the operating conditions during the desorption cycle when elevated temperatures may cause pressure loads on the materials. Some solvents when adsorbed on activated carbon, or in the presence of steam, oxidize or hydrolyze to a small extent and produce small amounts of corrosive materials. As a general rule, hydrocarbons and alcohols are unaffected and carbon steel can be used. In the case of ketones and esters, stainless steel or other corrosion resistant materials are required for the solvent-wetted parts of the adsorbers, pipework and condensers, etc. For carbon tetrachloride and 1,1,1-trichloroethane and other chlorinated solvents, titanium has a high degree of resistance to the traces of acidity formed during recovery. 22.1.4.2.3 Regeneration The key to the effectiveness of solvent recovery is the residual adsorption capacity (working capacity) after in-place regeneration. The basic principles of regeneration are given in Sections 22.1.2.4 and 22.1.4.1. After the adsorption step has concentrated the solvent, the design of the regeneration system1-3,13-16 depends on the application and the downstream use of the desorbed solvents whether it be collection and reuse or destruction.Table 22.1.8 may give some indications for choosing a suitable regeneration step. Table 22.1.8 Basic principle of regeneration Regeneration step
Off-site reactivation
Basic principle Transportation of the spent carbon to a reactivation plant. Partial gasification at 800-900°C with steam in a reactivation reactor. Reuse of the regenerated activated carbon
Typical application - low inlet concentration - odor control application - solvents with boiling points above 200 °C - polymerization of solvents
1522
Regeneration step
Klaus-Dirk Henning
Basic principle
Typical application
Steam desorption
Increasing the bed temperature by direct Most widely used regeneration technique steaming and desorption of solvent. The for solvent recovery plants. steam is condensed and solvent recovConcentration range from 1-50 g/m3 ered.
Hot air desorption
Rotor adsorber are often desorbed by hot Concentration by adsorption and incinerair ation of the high calorific value desorbate
Recovery of partly or completely water Desorption by means of an hot inert gas soluble solvents. Hot inert gas desorption recirculation. Solvent recovery can be Recovery of reactive solvents (ketones) achieved with refrigerated condenser or solvents which react with hot steam (chlorinated hydrocarbons) Atmospheric pressure adsorption and Pressure swing adsorp- heatless vacuum regeneration. tion/vacuum regenera- The revaporized and desorbed vapor is tion transported from the adsorption bed via a vacuum pump.
Low-boiling and medium-boiling hydrocarbon vapor mixes from highly concentrated tank venting or displacement gases in large tank farms. Recovery of monomers in the polymer industry.
Recovery of reactive solvents of the Reduced pressure and Steam desorption at low temperature and ketone-type. low temperature steam reduced pressure. Minimizing of side-reactions like oxidizdesorption ing, decomposing or polymerizing.
22.1.4.2.4 Safety requirements Adsorption systems including all the auxiliary and ancillary equipment needed for their operation are subject to the applicable safety legislation and the safety requirements of the trade associations and have to be operated in accordance with the respective regulations.2,19 Moreover, the manufacturer’s instructions have to be observed to minimize safety risks. Activated carbon or its impurities catalyze the decomposition of some organics, such as ketones, aldehydes and esters. These exothermic decomposition reactions25,26 can generate localized hot spots and/or bed fires within an adsorber if the heat is allowed to build up. These hazards will crop up if the flow is low and the inlet concentrations are high, or if an adsorber is left dormant without being completely regenerated. To reduce the hazard of bed fires, the following procedures are usually recommended: • Adsorption of readily oxidizing solvents (e.g., cyclohexanone) require increased safety precautions. Instrumentation, including alarms, should be installed to monitor the temperature change across the adsorber bed and the outlet CO/CO2 concentrations. The instrumentation should signal the first signs of decomposition, so that any acceleration leading to a bed fire can be forestalled. Design parameters should be set so as to avoid high inlet concentrations and low flow. A minimum gas velocity of approximately 0.2 m/s should be maintained in the fixed-bed adsorber at all times to ensure proper heat dissipation. • A virgin bed should be steamed before the first adsorption cycle. Residual condensate will remove heat.
22.1 Absorptive solvent recovery
1523
• Loaded activated carbon beds require constant observation because of the risk of hot spot formation. The bed should never be left dormant unless it has been thoroughly regenerated. • After each desorption cycle, the activated carbon should be properly cooled before starting a new adsorption cycle. • Accumulation of carbon fines should be avoided due to the risk of local bed plugging which may lead to heat build up even with weakly exothermic reactions. • CO and temperature monitors with alarm function should be provided at the clean gas outlet. • The design should also minimize the possibility of explosion hazard • Measurement of the internal adsorber pressure during the desorption cycle is useful for monitoring the valve positions. • Measurement of the pressure loss across the adsorber provides information on particle abrasion and blockages in the support tray. • Adsorption systems must be electrically grounded. 22.1.4.3 Special process conditions 22.1.4.3.1 Selection of the adsorbent The adsorption of solvents on activated carbon1-4 is controlled by the properties of both the carbon and the solvent and the contacting conditions. Generally, the following factors, which characterize the solvent-containing waste air stream, are to be considered when selecting the most well suited activated carbon quality for waste air cleaning: for solvent/waste air: • solvent type (aliphatic/aromatic/polar solvents) • concentration, partial pressure • molecular weight • density • boiling point, boiling range • critical temperature • desorbability • explosion limits • thermal and chemical stability • water solubility • adsorption temperature • adsorption pressure • solvent mixture composition • solvent concentration by components • humidity • impurities (e.g., dust) in the gas stream for the activated carbon type: General properties: • apparent density • particle size distribution • hardness • surface area
1524
Klaus-Dirk Henning
• activity for CCl4/benzene • the pore volume distribution curve For the selected solvent recovery task:
• the form of the adsorption isotherm • the working capacity • and the steam consumption The surface area and the pore size distribution are factors of primary importance in the adsorption process. In general, the greater the surface area, the higher the adsorption capacity will be. However, that surface area within the activated carbon must be accessible. At low concentration (small molecules), the surface area in the smallest pores, into which the solvent can enter, is the most efficient surface. With higher concentrations (larger molecules) the larger pores become more efficient. At higher concentrations, capillary condensation will take place within the pores and the total micropore volume will become the limiting factor. These molecules are retained at the surface in the liquid state, because of intermolecular or van der Waals forces. Figure 22.1.9 shows the relationship between maximum effective pore size and concentration for the adsorption of toluene according to the Kelvin theory. It is evident that the most valuable information concerning the adsorption capacity of a given activated carbon is its adsorption isotherm for the solvent being adsorbed and its pore volume distribution curve. Figure 22.1.10 presents idealized toluene adsorption isotherms for three carbon types: • large pores predominant • medium pores predominant • small pores predominant The adsorption lines intersect at different concentrations, depending on the pore size distribution of the carbon. The following applies to all types of activated carbon: As the toluene concentration in the exhaust air increases, the activated carbon load increases.
Figure 22.1.9. Relationship between maximum effective pore diameter and toluene concentration.
22.1 Absorptive solvent recovery
1525
Figure 22.1.10. Idealized toluene adsorption isotherms.
Figure 22.1.11. Adsorption isotherm of five typical solvents (After reference 13).
From the viewpoint of adsorption technology, high toluene concentration in the exhaust air is more economic than a low concentration. But to ensure safe operation, the toluene concentration should not exceed 40% of the lower explosive limit. There is, however, an optimum activated carbon type for each toluene concentration (Figure 21.1.10). The pore structure of the activated carbon must be matched to the solvent and the solvent concentration for each waste air cleaning problem. Figure 22.1.11 shows Freundlich
1526
Klaus-Dirk Henning
adsorption isotherms of five typical solvents as a function of the solvent concentrations in the gas stream. It is clear that different solvents are adsorbed at different rates according to the intensity of the interacting forces between solvent and activated carbon. Physical-chemical properties of three typical activated carbons used in solvent recovery appear in Table 22.1.9. Table 22.1.9. Activated carbon pellets for solvent recovery (After reference 13) AC-Type
C 38/4
Appearance
C 40/4
D 43/4
cylindrically shaped 3
380±20
400±20
430±20
Moisture content, wt%
<5.0
<5.0
<5.0
Ash content, wt%
<6.0
<6.0
<5.0
4
4
4
Surface area (BET), m /g
>1250
>1250
>1100
Carbon tetrachloride activity, wt%
80±3
75±3
67±3
Bulk density (shaken), kg/m
Particle diameter, mm 2
Producers13 of activated carbon are in the best position to provide technical advice on • selecting the right activated carbon type • contributing to the technical and economical success of a solvent recovery plant 22.1.4.3.2 Air velocity and pressure drop In solvent recovery systems, air velocity rates through the bed should be between 0.2 to 0.4 m/s. The length of the MTZ is directly proportional to the air velocity. Lower velocities (<0.2 m/s) would lead to better utilization of the adsorption capacity of the carbon, but there is a danger that the heat of adsorption not be carried away which would cause overheating and possibly ignition of the carbon bed. The power to operate the blowers to move waste air through the system constitutes one of the major operating expenses of the system. Pressure drop (resistance to flow) across the system is a function of • air velocity • bed depth • activated carbon particle size Small particle size activated carbon will produce a high pressure drop through the activated-carbon bed. Figure 22.1.12 compares the pressure drop of cylindrically-shaped activated carbon pellets with activated-carbon granulates. The activated carbon particle diameter must not be excessively large, because the long diffusion distances would delay adsorption and desorption. Commercially, cylindrical pellets with a particle diameter of 3 to 4 mm have been most efficient. 22.1.4.3.3 Effects of solvent-concentration, adsorption temperature and pressure For safety reasons the concentration of combustible solvent vapors should be less than 50 % of the lower explosive limit. The adsorption capacity of adsorbents increases as the concentration of the solvents increases. But the length of the MTZ is proportional to the solvent concentration. Because of the adsorption heat, as the adsorption front moves through the
22.1 Absorptive solvent recovery
1527
Figure 22.1.12. Pressure drop for various activated carbon types.
Figure 22.1.13. Adsorption isotherm of tetrahydrofuran for several temperatures.
bed also a temperature front follows in the same direction. To deal with the adsorption heat the inlet solvent concentration is usually limited to about 50 g/m3. The adsorption capacity of the adsorbent increases with pressure because the partial pressure of the solvent increases. An increase in adsorber temperature causes a reduction in adsorption capacity. Because the equilibrium capacity is lower at higher temperatures, the dynamic capacity (working capacity) of the activated carbon adsorber will also be lower. To enhance adsorption, the inlet temperature of the adsorber should be in the range of 20-40°C. In Figure 22.1.13 the adsorption isotherms of tetrahydrofuran on activated carbon D43/3 for several temperatures are shown.13
1528
Klaus-Dirk Henning
Figure 22.1.14. Water isotherm on activated carbon D43/4.
Figure 22.1.15. Dynamic toluene adsorption from humidified air (after reference 24).
22.1.4.3.4 Influence of humidity Activated carbon is basically hydrophobic; it adsorbs preferably organic solvents. The water adsorption isotherm (Figure 22.1.14) reflects its hydrophobic character. Below a relative gas humidity of about 40% co-adsorption of water can be neglected in most applications. However, higher humidity of the waste gases may affect the adsorption capacity of the activated carbon. Figure 22.1.15 demonstrates, using toluene as an example, how the relative humidity of the exhaust air influences the activated carbon loads. Experimental work24 has shown that
22.1 Absorptive solvent recovery
1529
• relative humidity rates below 30% will neither reduce the adsorption capacity nor the adsorption time,
• relative humidity rates above 70% substantially reduce the adsorption capacity, • the humidity of the gas will affect the adsorption capacity much more with low toluene concentrations than with high concentrations,
• with toluene concentrations between 10 and 20 g/m3 the negative influence of the humidity is small. The water content of the activated carbon after desorption may constitute another problem. The purification efficiency of each activated carbon is the better the less water is present after desorption. Unfortunately, the desorbed activated carbon in the vicinity of the adsorber walls usually contains high proportions of water (approx. 10 to 20%). With such a high water content it is difficult to remove all the water even when drying with hot-air over longer periods. The wet and poorly-regenerated activated carbons in these zones frequently lead to higher solvent concentrations in the purified air, and this is even at the beginning of the adsorption cycle. Some proposals for improvement include: • good insulation of the adsorber walls, • sufficient drying after each regeneration cycle, • cycle the desorption steam and the drying air counter-current to the direction of the adsorptive stream. 22.1.4.3.5 Interactions between solvents and activated carbon The majority of solvents are effectively recovered by adsorption on activated carbon and, when this is the case, the operation of the plant is straightforward. But some solvents may decompose, react or polymerize when in contact with activated carbon during the adsorption step and the subsequent steam desorption.19,25,26 Chlorinated hydrocarbon solvents can undergo hydrolysis to varying degrees on carbon surfaces, resulting in the formation of hydrogen chloride. Each activated carbon particle which is in contact with a metal screen or other constructional component may then act as a potential galvanic cell in the presence of moisture and chloride ions. Carbon disulfide can be catalytically oxidized to sulphur which remains on the internal surface of the activated carbon. Esters such as ethyl acetate are particularly corrosive because their hydrolysis results in the formation of e.g. acetic acid. Aldehydes and phenol or styrene will undergo some degree of polymerization when in contact with hot activated carbon. During the adsorptive removal of ketones such as methyl ethyl ketone or cyclohexanone a reduced adsorption capacity has been measured. Corrosion problems were also apparent and in some cases even spontaneous ignition of the activated carbon occurred. Activated carbon acts, in the presence of oxygen, as a catalyst during adsorption and even more frequently during desorption because of higher temperature.25,26 Figure 22.1.16 shows that the catalytic reactions occurring with cyclohexanone are predominantly of the oxidation type. Adipic acid is first formed from cyclohexanone. Adipic acid has a high boiling point (213°C at a 13 mbar vacuum). Further products identified were: cyclopentanone as a degradation product from adipic acid, phenol, toluene, dibenzofuran, aliphatic hydrocarbons and carbon dioxide. The high-boiling adipic acid cannot be desorbed from the activated carbon in a usual steam desorption. Consequently the life of the activated carbon becomes further reduced after each adsorption/desorption cycle.
1530
Klaus-Dirk Henning
Figure 22.1.16. Surface reaction of cyclohexanone on activated carbon (After reference 25).
Figure 22.1.17. Surface reaction of methyl ethyl ketone on activated carbon (After reference 25).
During adsorption of methyl ethyl ketone in the presence of oxygen, some catalytic surface reactions occur (Figure 22.1.17). The reaction products are acetic acid, di-acetyl, and presumably also di-acetyl peroxide. Di-acetyl has an intensive green coloration and a distinctive odor. Important for the heat balance of the adsorber is
that all of the reactions are strongly exothermic. Unlike the side products of cyclohexanone, acetic acid the reaction product of the MEK (corrosive!) will be removed from the activated carbon during steam desorption. Adsorption performance is thus maintained even after a number of cycles. Operators of ketones recovery plants should adhere to certain rules and also take some precautions25 • Adsorption temperature should not be higher than 30°C (the surface reaction rates increase exponentially with temperature). • The above requirement of low-adsorption temperatures includes: a) adequate cooling after the desorption step; b) a flow velocity of at least 0.2 m/s in order to remove the heat of adsorption. Since ketones will, even in an adsorbed state, oxidize on the activated carbon due to the presence of oxygen, these further precautions should be taken: • keep the adsorption/desorption cycles as short as possible, • desorb the loaded activated carbon immediately at the lowest possible temperatures, • desorb, cool and blanket the equipment with inert gas before an extended shut down.
22.1 Absorptive solvent recovery
1531
22.1.4.3.6 Activated carbon service life Activated carbon in solvent recovery service will have a useful service life of 1 to 10 years, depending on the attrition rate and reduction in adsorption capacity. Attrition rates are usually less than 1-3% per year. The actual rate will depend on carbon hardness. Particle abrasion and the resulting bed compaction leads to an increased pressure loss after several years of service. After 3-5 years of service screening to its original size is necessary. Adsorption capacity can be reduced by traces of certain high-boiling materials (resins, volatile organosilicone compounds) in the waste air which are not removed during the desorption cycle. Some solvents may decompose, react or polymerize when in contact with activated carbon and steam (Section 22.1.4.3.5). This gradual loading will decrease the carbon’s activity. While carbon can remain in service in a reduced capacity state, it represents a non-optimal operation of the system that results in • reduced amount of solvent removed per cycle • increased steam consumption • higher emissions • shorter adsorption cycles Eventually, recovering capacity will diminish; it is more economical to replace carbon than continue its use in a deteriorated state. In replacing spent activated carbon there are two options: • Replacing with virgin carbon and disposal of the spent carbon. • Off-site reactivation of the spent carbon to about 95% of its virgin activity for about one-half of the cost of new carbon Some activated carbon producers offer a complete service including carbon testing, off-site reactivation, transportation, adsorber-filling and carbon make-up.13 22.1.5 EXAMPLES FROM DIFFERENT INDUSTRIES Some examples of solvent recovery systems in different industries show that reliable and trouble-free systems are available. 22.1.5.1 Rotogravure printing shops Process principle Adsorptive solvent recovery with steam desorption and condensation units with gravity separator and a stripper have become a standard practice in modern production plants. The solvents-laden air (toluene, xylene) is collected from emission points, e.g., rotogravure printing machines, drying ducts by means of a blower and passed through the recovery plant. Design example30 In 1995, one of Europe’s most modern printing shops in Dresden was equipped with an adsorptive solvent recovery system (Supersorbon® process (Figure 22.1.18.)). Design data (First stage of completion) Exhaust air flow rate 240,000 m3/h (expandable to 400,000 m3/h) Solvent capacity 2,400 kg/h (toluene) Exhaust air temperature 40°C Toluene concentration in clean air max. 50 mg/m3 (half-hour mean)
1532
Klaus-Dirk Henning
Figure 22.1.18. Supersorbon® process for toluene recovery (After reference 30).
Solvent-laden air is exhausted at the three rotogravure printing presses by several fans operating in parallel and is routed in an upward flow through four adsorbers packed with Supersorbon® activated carbon. The solvent contained in the air is adsorbed on the activated carbon bed. Adsorption continues until breakthrough, when the full retentive capacity of the adsorbent for solvent vapors is used up. The purity of the clean exhaust air complies with the regulation for emission level of less than 50 mg/m3. Regeneration of the adsorbent takes place by desorbing the solvent with a countercurrent flow of steam. The mixture of water and toluene vapors is condensed and the toluene, being almost insoluble in water, is separated by reason of its different density in a gravity separator. The recovered toluene can be re-used in the printing presses without further treatment. Very small amounts of toluene dissolved in the wastewater are removed by stripping with air and returned to the adsorption system inlet together with the stripping air flow. The purified condensate from the steam is employed as make-up water in the cooling towers. Operating experience The adsorption system is distinguished by an extremely economical operation with a good ratio of energy consumption to toluene recovery. Solvent recovery rate is about 99.5 %. 22.1.5.2 Packaging printing industry In the packaging printing industry solvents like ethyl acetate, ethanol, ketones, tetrahydrofuran (THF), hexane and toluene are used in printing. The solvent laden air generally contains between 2 and 15 g/m3 of solvent and, depending on the season, 5-18 g/m3 of water. Adsorptive recovery with steam desorption was widely used for many years but its disadvantage is that the solvents are recovered in various mixtures containing large amounts of water. The recovery of solvents is complicated because most solvents produce azeotropes with water which are not easily separated and, consequently, the waste water fails to meet the more stringent environmental regulations. In the last 10 years the adsorbent regen-
22.1 Absorptive solvent recovery
1533
eration with hot inert gas has became the more favored process in the packaging printing industry. 22.1.5.2.1 Fixed bed adsorption with circulating hot gas desorption31,32 Activated carbon preferentially adsorbs non-polar organic solvents, while adsorbing relatively little water. Thus activated carbon allows the passage of 95-98 % of the water while retaining the solvents from the gas stream. During the adsorption step solvent laden air or gas is delivered to the bed through the appropriate valves. As adsorption starts on a fresh bed, the effluent gas contains traces of solvent. Through time, the solvent level increases and when it reaches a predetermined value, the adsorption is stopped by closing the feed gas valves. The adsorption step usually lasts for 4-16 hours. An example of a solvent recovery plant with 5 activated carbon beds and hot gas desorption is shown in Figure 22.1.19. If necessary, feed air is cooled to 35°C and sent by the blower V-1 to the main heater 6 from which it enters 4 beds simultaneously, while one bed is being regenerated. The cleaned air is collected in header 7 and vented. After the adsorption step, the air is displaced from the bed by an inert gas such as nitrogen. A circulating inert gas at 120-240°C serves for heating and stripping the solvent from the carbon. The solvent is recovered by cooling and chilling the circulating gas. The optimum chilling temperature depends on the boiling point of the solvent. Generally, chilling temperatures are between +10 and -30°C. During the heating step, the temperature of the effluent gas gradually increases until a predetermined value (for example, 150°C) is reached, after which the heater is bypassed and the bed is cooled down by cold gas. Typical solvent recovery plants will have from 2 to 8 beds. All beds go through the same adsorption, inertization, heating and cooling steps but each at different time. Referring to the flow sheet in Figure 22.1.19 the regeneration gas is circulated by blower V-2 to the gas heater, the header 8 to the bed being regenerated. The effluent gas is passed to the header 9, cooler, molecular sieves (or water condenser), to the chiller for sol-
Figure 22.1.19. Fixed bed adsorption with circulating hot gas desorption (After references 31,32).
1534
Klaus-Dirk Henning
vent condensation and back to the blower V-2. At the end of the adsorption step the bed contains 10-30% solvent and 1-2% water. There are various ways to remove the water separately and to recover a solvent containing between 0.1 and 1.5 % water (Table 22.1.10). Table 22.1.10. Process conditions and water content in the solvent (After references 32,33) Water in the raw solvent
Investment cost, %
Heat needed, kWh/kg S
Power, kWh/kg S
Basic process with molsieve beds
0.1
100
2.2
1.0
Chiller instead of molsieve
1.5
91
1.5
1.0
0.4 - 1.0
93
1.1
1.0
0.1
96
1.1
1.0
Chiller and 1 carbon bed Chiller, and 1 carbon bed molsieve
Using molecular sieves Beds of molecular sieves are used to recover a solvent with 0.1%. This process requires around 2-2.5 kWh heat/kg of recovered solvent. Regeneration loop pressure drops are high, because of the additional molsieve beds and valves. Separate condensation of water Initially, when regenerative heating of the bed starts, very little solvent is desorbed, but much of the water (about 87%) is desorbed. The circulating gas is first passed through a cooler and then a separate chiller for the condensation or freezing of the water. Subsequently the gas is passed to the chiller where the water is condensed or frozen. From 1 to 1.8% water is recovered. This process is simpler than the molecular sieves process. Two separate chillers Two separate chillers are used in another adsorption process which offers further substantial improvements. A third activated carbon bed is added so that while two of the beds are being regenerated one of these is being cooled and transferring its heat to the other bed which is being heated. This not only gives excellent heat recovery but also provides a means of reducing the water content of the recovered solvent. The cooled bed absorbs the water vapor from the gas stream coming from chiller. This dry gas stream is transported to the bed which is being heated and the desorbed solvent from the heated bed remains dry. Solvents recovered at this stage contain only 0.4 to 0.5% water. Solvent which is removed from the bed being cooled is readsorbed on the bed being heated. The process, which is patented, brings important benefits of regeneration gas flow rates. Such low flow rates mean that the process requires less heat, less refrigeration and a lower cost regeneration loop. The overall efficiency is high because the low residual solvent loading of the regenerated beds leads to increased solvent recovery. Two separate chillers plus molecular sieve bed A two chiller system with a molecular sieve bed has to be regenerated only about once per week, since most of the water is removed in the chiller and readsorbed on the cooled carbon bed. This process offers a high solvent recovery rate giving a solvent with only 0.1 % water at greatly reduced heat consumption, lower investment, and higher solvent recovery rate.
22.1 Absorptive solvent recovery
1535
Key plants with all the above mentioned processes are offered by special engineering companies.31,32 The investment cost for a typical recovery plant with hot gas regeneration can be estimated by the following equation: COST (in 1000 Euro) = 200 + 0.4 x A0.7 + 15 x S0.7 where: A S
feed air rate in Nm3/h solvent flow rate in kg/h
22.1.5.2.2 Solvent recovery with adsorption wheels Adsorption wheels33 are used for the continuous purification of large volumes of exhaust air containing relatively low solvent concentrations. The adsorption wheel consists of a number of identical chambers arranged axially around a vertical axis. All chambers contain adsorbent. The wheel rotates and each chamber passes in sequence over an exhaust air duct and solvent molecules are adsorbed. As wheel continues to rotate the chamber in which adsorption had occurred now moves into a desorption position in which hot air is passed through the chamber and over the adsorbent. This removes the adsorbed solvent. The hot air flow in the desorption sector is at relatively low flow rate compared to that of contaminated gas stream. The number of chambers in the desorption zone is much greater than the number in the absorption zone so the desorption stream is many times more concentrated in solvent than was the exhaust stream. At this higher concentration, the desorption stream can now be economically treated in one of the ways: • The adsorption stream is purified by means of recuperative or regenerative oxidation. This approach is advantageous in painting applications with solvent mixtures that cannot be reused in production. • For solvent recovery by condensation, the desorption stream is cooled down in a cooling aggregate and the liquid solvent is recovered for reuse. Water contents below 1% can be achieved without further purification. Example: A manufacturer who specializes in flexible packaging, i.e., for confectionary, always adds the same solvent mixture (ethanol, ethyl acetate, ethoxypropanol) to his printing ink. Adsorptive solvent removal of the solvent-mixture by use of an adsorption wheel (Figure 22.1.20) and solvent recovery via condensation proved the technically and cost effective. 10,000 to 65,000 Nm3/h of exhaust air from printing machines and washing plants at a temperature maximum of 45°C and a maximum solvent loading of 4.6 g/m3 is to be cleaned. Depending on the air volume two adsorption wheels with a capacity of 26,000 Nm3/h and 39,000 Nm3/h and separate desorption circuits are used either alternatively or together. The total desorption air of max. 11,000 Nm3/h is being concentrated to max. 27 g/m3, which corresponds to 50% of the lower explosion limit. The condensation unit for the solvent recovery process is gradually adjusted to small, medium or large exhaust air volumes. The recovered solvents are stored and returned to the production process. 22.1.5.3 Viscose industry In plants which produce viscose fibre (stable fibre), viscose filament yarn (rayon) and viscose film (cellophane), large volumes of exhaust air contaminated with carbon disulfide
1536
Klaus-Dirk Henning
Figure 22.1.20. Adsorption wheel for solvent recovery in packaging printing (After reference 33).
(CS2) and hydrogen sulphide (H2S) have to be cleaned. Typical waste gas volumes and concentration of CS2 and H2S are given in Table 22.1.11. Different cleaning/recovery processes are available for the removal of each of the two sulfur containing compounds:2,17,34,35 Hydrogen sulfide • Absorption of H2S in a NaOH-scrubber • Catalytic oxidation of H2S to elemental sulphur on iodine impregnated activated carbon (Sulfosorbon-process) 2 H2S + O2 →
2 S + 2 H2O
The sulphur is adsorbed at the internal surface at the carbon-catalyst. Table 22.1.11. Waste gas in the viscose industry (After reference 35) Product
Rayon Staple fibre Viscose fibre
Spec. waste gas volume, m3/t
Concentration (mg/m3) H2S
CS2
400,000 - 700,000
60 -130
300 - 800
50,000 - 90,000
700 -1800
2300 - 4000
100,000 -150,000
280 - 400
1000 - 3000
22.1 Absorptive solvent recovery
1537
Carbon disulfide Adsorptive removal on activated carbon and recovery by steam desorption. For simultaneous H2S and CS2 removal the Sulfosorbon-process uses adsorbers packed with two different activated carbon types • Iodine impregnated wide-pore activated carbon for H2S oxidation in the bottom part (gas inlet) of the adsorber • Medium-pore activated carbon for the adsorption of CS2 in the upper layer of the fixed bed. As soon as the CS2 concentration in the treated air approaches the emission limit, the exhaust air stream is directed to a regenerated adsorber. After an inert gas purge, the carbon disulfide is desorbed with steam at 110 to 130°C. The resulting CS2/steam mixture is routed through a condenser and cooler, before entering a gravity separator where phase separation occurs. At the usual CS2 and H2S concentrations in viscose production exhaust air, the CS2 adsorption/desorption cycles can be run for a prolonged periods before the first layer loaded with elemental sulphur has to be regenerated. The regeneration process of the sulphur loaded activated carbon involves the following steps: • washing out of sulphuric acid with water • extraction of elemental sulphur with carbon disulfide • desorption of carbon disulfide with steam • air drying and cooling of activated carbon The elemental sulphur present in the carbon disulfide in dissolved form can be separated by distillation and recovered as high-purity sulphur. Design example34 In the production of viscose filament yarn a Supersorbon®-system combined with a NaOH-scrubber system (Figure 22.1.21) has been in use since 1997. A very high standard of safety engineering is implemented in the treatment system owing to the flammability of the CS2 and the toxic nature of the constituents to be removed from the exhaust air. System concept First treatment stage: Absorption of H2S in two NaOH jet scrubbers and one water-operated centrifugal scrubber. Second treatment stage: Fixed-bed adsorption for purification of exhaust air and CS2 recovery. Treated air stream Exhaust air from viscose filament yarn plant. Design data: Exhaust air flow rate 12,000 Nm3/h Solvent capacity 27 kg CS2/h and 7.5 kg H2S/h Exhaust air temperature 15 - 30°C Clean air solvent concentration 100 mg/Nm3 (24-hour mean) CS2: max. 5 mg/Nm3 (24-hour mean) H2S: max.
1538
Klaus-Dirk Henning
Figure 22.1.21. H2S and CS2 removal in the filament yarn production (After reference 34).
H2S-removal The exhaust air, saturated with water vapor, is first treated in an absorption unit for removal of H2S by routing it through two successive jet scrubbers working with dilute caustic soda solution. Downstream of these, a centrifugal scrubber is installed as an entrainment separator. The sulphide-containing solution rejected from the scrubbers is used to precipitate zinc in the waste water treatment system of the production plant. By products adsorbed on the activated carbon, such as sulphuric acid and elemental sulphur, are removed periodically by water and alkaline extraction. CS2-removal and recovery The pre-cleaned airstream containing carbon disulfide vapors passes through two or three parallel adsorbers, in which the CS2 is absorbed on a bed of Supersorbon® activated carbon. As soon as the adsorbent is saturated it is regenerated by desorbing the solvent with a countercurrent flow of steam. The resulting mixture of water and CS2 vapors is condensed and separated in gravity settlers. The recovered CS2 is returned to the viscose production without further treatment. The condensed steam is stripped of residual CS2 in the centrifugal scrubber and used as dilution water for operation of the jet scrubbers. Operating experience The specified purity of the treated exhaust air is reliably achieved with the Supersorbon® process34 with respect to both CS2 and H2S. The CS2 recovery rate is about 95%. The purity of the recovered solvent meets viscose production specifications. A widely varying CS2 concentration in the exhaust air has not adversely affected operation of the adsorption system.
22.1 Absorptive solvent recovery
1539
22.1.5.4 Refrigerator recycling Condensation process Condensation processes are especially suitable for the cleaning of low flow highly concentrated streams of exhaust gas.36 The entire waste gas stream is cooled below the dew point of the vapors contained therein, so that these can condense on the surface of the heat exchanger (partial condensation). Theoretically, the achievable recovery rates depend only on the initial concentration, the purification temperature and the vapor pressure of the condensables at that temperature. In practice however, flow velocities, temperature profiles, the geometry of the equipment, etc. play decisive roles, as effects such as mist formation (aerosols), uneven flow in the condensers and uncontrolled ice formation interfere with the process of condensation and prevent an equilibrium concentration from being reached at the low temperatures. The Rekusolv Process37 uses liquid nitrogen to liquefy or freeze vapors contained in the exhaust gas stream. In order to reduce the residual concentrations in the exhaust to the legally required limits, it is often necessary to resort to temperatures below minus 100°C. The Rekusolv process is quite commonly used in the chemical and pharmaceutical industry and at recycling plants for solvent recovery. Example of solvent recovery in refrigerator recycling36,37 In refrigerator recycling plants R11 or pentane is released from the insulating foam of the refrigerator during shredding. Figure 22.1.22. shows the Rekusolv process as it is used by a refrigerator recycling company. At this plant around 25 refrigerators per hour are recycled thereby generating 8 kg/h of polluting gases. The Rekusolv plant is capable of condensing almost all of this. The unit is designed to operate for 10 to 12 hours before it has to be defrosted. The plant is operated during the day and is automatically defrosted at night. The concentration of pollutants in the exhaust gas is reduced from 20 to 40 g/m3 to 0.1 g/m3 - a recovery rate of more than 99.5%. 22.1.5.5 Petrochemical industry and tank farms Vapor recovery units are installed at petrochemical plants, tank farms and distribution terminals of refineries. Tank venting gases are normally small in volume, discharged intermittently at ambient temperature and pressure and loaded with high concentrations of organic vapors.2 Processes used for the cleanup of such waste air streams with organic vapor up to saturation point are often combined processes:2,38,39 • absorption and pressure-swing adsorption • membrane permeation and pressure-swing adsorption • condensation and adsorption Adsorption on activated carbon and vacuum regeneration Figure 22.1.23 represents the basic principle of the systems which are used to recover vapors displaced from tank farms and loading stations and blend them back into the liquids being loaded. The plant consists of two fixed-bed adsorbers packed with activated carbon which operate alternatively. The vapor pressure gradients required as a driving force for desorption is generated by vacuum pumps. As the raw gas entering the adsorber is saturated with hydrocarbon vapors, the heat of adsorption causes local heating of the well insulated adsorbent bed. The higher temperature of the spent activated carbon supports subsequent desorption as the pressure is being lowered by means of a vacuum pump.
1540
Klaus-Dirk Henning
Figure 22.1.22. Block diagram of the Rekusolv process (After references 36,37).
Figure 22.1.23. Vapor recovery by adsorption and vacuum regeneration (After reference 39).
22.1 Absorptive solvent recovery
1541
The desorbed vapor is transported from the adsorption bed via a vacuum pump. A liquid ring pump using water for the seal fluid is normally applied for safety purposes in operations where air is allowed to enter the adsorbers. The desorption stream is processed through a cooler and absorber. The absorber uses gasoline (or the recovered solvent) as a sorbent and serves as a hydrocarbon reduction stage for the entire system. Loading facilities with the inlet vapor concentrations of over 10% by volume can demonstrate recovery efficiencies of over 99%. 22.1.5.6 Chemical industry In the chemical industry a variety of solvents and solvent mixtures are in use. For many gas cleaning operations special adsorption processes are required. Two-stage adsorption40 In many cases, a two-stage process offers significant advantages. This process makes use of two adsorbers that are operating alternately in series. Four-way reverse valves ensure safe change-over. During operation, the air passages of the valves are always open so that any blocking of the flow is impossible. Solvent laden air first passes into one of the adsorbers. Purified air is released into the atmosphere only during the time when the second adsorber is being steamed. During this phase of the process, the on-line adsorber is only slightly charged with solvent which means that it will be completely adsorbed. In general the time required for steaming an adsorber is at most 50% of the time provided for adsorption. After the adsorber has been steamed, the outlet valve is switched over. The steamed adsorber is then dried with solvent-free air taken from the laden adsorber. A feature of this process is that the steamed adsorber is always dried with solvent-free air. An advantage of this two-stage process is that the laden adsorber can be loaded beyond the breakthrough capacity since the second adsorber is unladen and can therefore accept any excess solvent. The condensation phase and reprocessing of the desorbate are carried out in the same manner as described for the single-stage plant. For example, a two-stage adsorption plant for the recovery of dichloromethane is able to clean a gas stream of 3000 m3/h to a residual solvent content of less than 20 mg/m3. REFERENCES 1 2 3 4 5 6 7 8 9 10 11
J.W. Leatherdale, AirPollution Control by Adsorption, Carbon Adsorption Handbook, Ann Arbor Science Publisher Inc., Michigan, 1978 P.N. edited by Cheremisinoff, F. Ellerbusch. VDI 3674, Waste gas cleaning by adsorption, Process and waste gas cleaning, Beuth Verlag GmbH, 10772 Berlin, 1998. H. Menig, Luftreinhaltung durch Adsorption, Absorption und Oxidation, Deutscher Fachschriften-Verlag, Wiesbaden, 1977. H. von Kienle, E. Bäder, Aktivkohle und ihre industrielle Anwendung, Ferdinand Enke Verlag, Stuttgart, 1980. Test methods for activated carbon. European Council of Chemical Manufacturers` Federations/CEFIC Brussels (1986) 7. A. Yehaskel, Activated Carbon, Manufacture and Regeneration, Noyes DATA Corporation, Park Ridge, New Jersey, USA, 1978. M. Smisek, S. Cerny, Activated Carbon. Elsevier Publishing Company, Amsterdam-London- New York, 1970. R.Ch. Bansal, J.-B. Donnet and F. Stoeckli, Acitvated Carbon, Marcel Dekker Inc., New York, 1988. H. Jüntgen, Grundlagen der Adsorption, Staub-Reinhaltung Luft, 36, (1976) No. 7, pp. 281-324. H. Seewald, Technische verfügbare Adsorbentien. Vortragsveröffentlichung Haus der Technik, H. 404, pp. 24-34, Vulkan-Verlag, Essen, 1977. W. Kast, Adsorption aus der Gasphase. VCH-Verlag, Weinheim, 1988.
1542
12 13 14 15
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Klaus-Dirk Henning
H. Krill, Adsorptive Abgasreinigung - eine Bestandsaufnahme, VDI-Berichte 1034, pp. 339-371, VDI-Verlag, Düsseldorf, 1993. K.-D. Henning, Activated Carbon for Solvent Recovery, Company booklet CarboTech Aktivkohlen GmbH, D-45307 Essen, Germany. A. Mersmann, G.G. Börger, S. Scholl, Abtrennung und Rückgewinnung von gasförmigen Stoffen durch Adsorption, Chem.-Ing.-Techn., 63, No. 9, pp. 892-903. H. Jüntgen, Physikalisch-chemische und verfahrenstechnische Grundlagen von Adsorptionsverfahren, Vortragsveröffentlichung Haus der Technik, H. 404, pp. 5-23, Vulkan-Verlag, Essen, 1977. K.-D. Henning, Science and Technology of Carbon, Eurocarbon, Strasbourg/France, July 5-9, 1998. K.-D. Henning, S. Schäfer, Impregnated activated carbon for environmental protection, Gas Separation & Purification, 1993, Vol. 7, No. 4, p. 235. P.N. Cheremisinoff, Volatile Organic Compounds, Pollution Engineering, March 1985, p.30 C.S. Parmele, W.L. O`Connell, H.S. Basdekis, Vapor-phase cuts, Chemical Engineering, Dec. 31, 1979, p. 59. M.G. Howard, Organic solvents recovered efficiently by dry processes, Polymers Paint Colour Journal, May 18, 1983, p. 320. W.D. Lovett, F.T. Cunniff, Air Pollution Control by Activated Carbon, Chemical Engineering Progress, Vol. 70, No. 5, May 1974, p. 43. H.L. Barnebey, Activated Charcoal in the Petrochemical Industry, Chemical Engineering Progress, Vol. 67, No. 11, Nov. 1971, p. 45. K.-D. Henning, W. Bongartz, J. Degel, Adsorptive Removal and Recovery of organic Solvents, VfT Symposium on Non-Waste Technology, Espo/Finnland, June 20-23, 1988. G.G. Börger, A. Jonas, Bessere Abflugtreinigung in Aktivkohle-Adsorben durch weniger Wasser, Chem.-Ing.-Techn., 58 (1986), p. 610-611, Synopse 15.12. K.-D. Henning, W. Bongartz, J. Degel, Adsorptive Recovery of problematic solvents, Paper presented at Nineteenth Biennial Conference on Carbon, June 25-30, 1989, Pennstate University/Pennsylvaia/USA. A.A. Naujokas, Loss Preventation, 12 (1979) p. 128. G. Konrad, G. Eigenberger, Rotoradsorber zur Abluftreinigung und Lösemittelrückgewinnung. Chem.-Ing.-Techn., 66 (1994) No. 3, pp. 321-331. H.W. Bräuer, Abluftreinigung beim Flugzeugbau, Lösemittelrückgewinnung in der Wirbelschicht, UTA Umwelt Technologie Aktuell (1993),No. 2, pp. 115-120. W.N. Tuttle, Recent developments in vacuum regenerated activated carbon based hydrocarbon vapour recovery systems, Port Technology International, 1, pp. 143-146, 1995. Exhaust air purification and solvent recovery by the Supersorbon(r) Process, Company booklet, Lurgi Aktivkohlen GmbH, D-60388 Frankfurt, Gwinnerstr. 27-33, Germany. G. Caroprese, Company booklet, Dammer s.r.l., I-20017 Passirana die Rhol (Mi), Italy. R. Peinze, Company booklet, Air Engineering, CH-5465 Mellikon, Switzerland. Environmental Technology, Company booklet, Eisenmann, D-71002 Böblingen, Germany. Exhaust air purification and CS2 recovery by the Supersorbon® process with front- and H2S absorption, Company booklet, Lurgi Aktivkohlen GmbH, D-60388 Frankfurt, Germany. U. Stöcker, Abgasreinigung in der Viskose-Herstellung und -Verarbeitung, Chem.-Ing.-Techn., 48, 1976, No. 10. Exhaust gas purification, Company booklet, Messer Griesheim GmbH, D-47805 Krefeld, Germany. F. Herzog, J. Busse, S. Terkatz, Refrigerator recycling - how to deal with CFCs and pentane, Focus on gas 13, Company booklet, Messer Griesheim GmbH, D-47805 Krefeld, Germany. C.J. Cantrell, Vapour recovery for refineries and petrochemical plants, CEP, October 1982, pp. 56-60. Kaldair Vapor Recovery Process, Company booklet, Kaldair Ltd., Langley Berks SL3 6EY, England. Waste Air Purification, Company booklet, Silica Verfahrenstechnik GmbH, D-13509 Berlin, Germany.
22.2 Solvent recovery
1543
22.2 SOLVENT RECOVERY Isao Kimura Kanken Techno Co., Ltd., Osaka, Japan
22.2.1 ACTIVATED CARBON IN FLUIDIZED BED ADSORPTION METHOD This section discusses a solvent recovery process developed by Kureha Engineering Co. Ltd., Japan.1 In this continuous process, spherical particles of activated carbon (AC) circulate in the adsorption and desorption columns by fluidization. In the adsorption column, the particles form fluidized beds on multi-trays to adsorb the solvent in counter-current contact with the feed gas. The cleaned gas is released from the top of the adsorption-column to the atmosphere. The carbon absorber with adsorbed solvent is electrically heated (150-250°C) at the upper part of the desorption column and solvent is desorbed by nitrogen, which is supplied from the bottom of the desorption column. Nitrogen is continuously recycled in the desorption column. The solvent in the carrier gas is fed to a condenser for its recovery. Features of the process: • Atmospheric condensate is the only waste product because nitrogen (or air) is used as the carrier gas. • High flow rate per unit area because the process is conducted by fluidization. • Low thermal energy loss because both the adsorption and desorption are conducted at constant temperature. • Low electric energy consumption because the fluidization needs less blower driving power. Figure 22.2.1 shows the process diagram and Table 22.2.1 shows application data.
Figure 22.2.1. Flow diagram of the process.
1544
Isao Kimura
Table 22.2.1. Application data. Activated carbon fluidized bed adsorption process Solvents
Manufacturing plants Flow rate, m3/min Feed conc., ppm
phenol, naphthalene
semiconductors
50
100*
curry odor
food processing
240
4,000
IPA, phenol, acetone, ethanol
electronics
120
220
terpineol
IC
300
30
dichloromethane
semiconductors
80
500
IPA, xylene
semiconductors
150
110
IPA
semiconductors
80
200
manure odor
fertilizers
600
3,000*
2,000
4
phenol, formaldehyde, ammonia shell mold casting butyl acetate, IPA, etc.
semiconductors
420
50
terpineol
IC
300
30
butyl acetate, etc.
solvents
550
50
*odor concentration
22.2.2 APPLICATION OF MOLECULAR SIEVES Molecular sieves is a term synonymous in this context to aluminosilicate hydrate also called zeolite. The zeolite is applied to deodorization, gas separation and some other processes. Since this material is, unlike activated carbon, non-combustible, it has increasingly been applied to concentrate volatile organic compounds, VOC. The concentration process discussed here applies to the “VOC concentrator” developed by Seibu Giken Co. Ltd. Japan.2 The element referred to here as the VOC concentrator is non-combustible and has a honeycomb structure. The key features are summarized below. Key features of honeycomb rotor concentrator • Hydrophobic zeolite is embedded into a calcinated ceramic honeycomb substrate with an inorganic binder and recalcinated at high temperature to improve the compatibility of zeolite with the ceramic substrate. • Various types of VOCs can be efficiently purified and concentrated. • The most suitable zeolite and composition are selected for application. • Absorption medium is non-combustible • Heat resistance up to 500°C • Desorption temperature: 150 - 220°C Functions of the VOC concentrator (1) Purification - The VOC laden exhaust air stream is purified by zeolite or other adsorbent while passing through the rotor, and the purified air stream is vented into the atmosphere. (2) Concentration - The VOC laden exhaust air stream is adsorbed in the process zone and desorbed in the desorption zone into a heated desorption air stream with much less air volume than that of process air stream. The VOC is concentrated nearly equal to the air volume ratio (5-15) of the process and desorption (QP/QD).
22.2 Solvent recovery
1545
Basic design of VOC concentrator VOC concentrator consists of a VOC rotor, a rotor driving device, a rotor casing with a set of seals, a pair of chambers (front/rear) with a zone partition wall, and a desorption heater as shown in Figure 22.2.2.
Figure 22.2.2. VOC concentrator.
Advantages VOC concentrators, outstanding in their purifying and concentrating efficiency, have been used worldwide as a high-quality and safe product owing to material engineering expertise. By being combined with oxidizers, and other recovery equipment, the initial investment and running cost of the total system can be drastically diminished. Even for such VOCs as cyclohexane/MEK/ammonia/methyl alcohol/ethyl alcohol/styrene/formaldehyde/isophorone/phenol, which have been found unsuitable for the existing technology using activated carbon, due to economic or safety concerns, the VOC concentrator is costeffective and safe. • Non-combustible components - The development of non-flammable rotors has been achieved by engineering the appropriate materials for the honeycomb substrate, the adsorbent, and the binder and by application of special material processing technique. • Treatment of VOCs which have high boiling point - Taking advantages of zeolite rotors, non-combustible component, and high heat resistance, concentrators can use adsorption air at high temperatures. Accordingly, VOCs that could not be treated by carbon material due to its desorption temperature limit have turned out to be easily treated. • Inertness - VOCs that can easily polymerize, such as, styrene, cyclohexanone, etc., can be effectively treated by hydrophobic zeolite. • Cleaning and reactivation - Zeolite rotors can be calcinated under high temperature, due to their all inorganic content (including binder). The rotors can be easily washed. Zeolite rotors can be reactivated by heat treatment. Table 22.2.2 shows the typical applications of the VOC concentrator.
1546
Isao Kimura
Table 22.2.2. Typical applications of the VOC concentrator Industry
Facilities
Treated VOCs
Automobile, Steel structure manPainting booth ufacturing
Toluene, xylene, esters, alcohols
Steel furniture
Painting booth, oven
Toluene, xylene, esters, alcohols
Printing
Dryer
Toluene, xylene, esters, alcohols
Adhesive tape Hook and loop fastener
Coating process Cleaning unit
Ketones (MEK, cyclohexanone, methyl isobutyl ketone, etc.)
Chemicals
Oil refinery, reactors
Aromatic hydrocarbons, organic acids, aldehydes, alcohols
Synthetic resin adhesive
Plastics Plywood manufacturing process
Styrene, aldehydes, esters
Semiconductor
Cleaning unit
Alcohols, ketones, amines
22.2.3 CONTINUOUS PROCESS FOR AIR CLEANING USING MACROPOROUS PARTICLES AS ADSORPTION AGENTS In the photoresist segment of the semiconductor process a very low concentration of VOC is generated. These low concentration VOCs can be removed by a concentration step followed by a catalytic oxidation process. The industrial application of this method is discussed below. Dilute VOC is fed to the rotating zeolite concentrator discussed in Section 22.2.2 (honeycomb rotor) where it is concentrated to 10 times of the original concentration and burned on catalysts at relatively low temperature. This saves fuel cost. The VOC is a mixture of IPA, acetone, MEK, toluene, ethyl acetate, n-hexane, propylene glycol monoethyl ether, propylene glycol monoethyl ether acetate, and tetramethyl ammonium hydroxide. The mix-
Figure 22.2.3. Flow diagram of continuous air cleaning process by zeolite concentrator.
22.2 Solvent recovery
1547
Figure 22.2.4. Continuous air cleaning unit by zeolite concentrator.
ture is exhausted from the photoresist at approximately 100 Nm3/min, 20°C and 60% RH. Figure 22.2.3 shows the process flow diagram. The VOC laden feed gas (exhausted from the photoresist) is pre-treated by filters in two steps, first by a bag filter and then by activated carbon. The former removes dusts and the latter removes any readily polymerizable vapors such as, acrylic acid, organic silicon comFigure 22.2.5. Block diagram. Continuous air cleaning process pounds e.g. hexamethylene disilazane by zeolite concentrator. (HMDS), phosphorous-compounds and halogen-compounds which are all catalyst poisons. The pretreated gas is fed to the zeolite honeycomb rotor for concentration. (Figure 22.2.4) Table 22.2.3 shows measurements at the designated points in the block diagram (see Figure 22.2.5). VOC removal reached 91 % at outlet of the rotor and 98 % at outlet of the catalytic oxidation unit. Table 22.2.3. Measurements according to block diagram (Figure 22.2.5). Continuous air cleaning process by zeolite concentrator No.
Point of detection
Temp., °C
Conc., ppm
Flow rate, Nm3/min
1
Concentrator, inlet
26
430
102
2
Concentrator, outlet
33
40
116
3
Cat. oxidation, inlet
450
3,700
4
Cat. oxidation, outlet
490
83
5
Exhaust
39
1548
Isao Kimura
22.2.4 SOLVENT RECOVERY FROM HAZARDOUS WASTES Table 22.2.5. Cleaning performance of solvents Table 22.2.4. Properties of cycloparaffins as a cleaning solvent 2 Cleaning solvent
Property Appearance
Value transparent
Color (Seibolt)
+30
Specific gravity
0.839
Viscosity, 40°C Cst
1.6
Flash point, °C
76
Aniline point
58
Cycloparaffins content, %
80
Alkylbenzene content, %
10
Aromatics content, %
1
The process was previously used for chlorine compounds represented by perchloroethylene and trichloroethylene and for fluorine compounds used for cleaning of press stamped metal parts. However, these compounds are now prohibited for use and production by environmental regulations. The current replacements include emulsions, organic and inorganic solutions, etc. Cycloparaffins originating from petroleum products were quali-
Oils
Residual oil, mg/cm
cycloparaffins
press oil fan press oil bender oil
0.0017 0.0015 0.0017
trichloroethane
press oil fan press oil bender oil
0.0021 0.0013 0.0013
Figure 22.2.6. Vapor pressure and combustion range of cycloparaffins.
Figure 22.2.7. Metal parts cleaning employing cycloparaffins.
22.2 Solvent recovery
1549
Figure 22.2.8. Flow diagram of solvent recovery process.
fied as a good cleaning solvent (see Table
Table 22.2.6. Brief specification of the 22.2.4). cycloparaffins recovery unit Item
Specification
Solvent
cycloparaffins
Capacity
70 l/min
Dimensions
1,050 x 533 x 1,355 mm
Weight
250 kg (approx.)
Steam pressure
6.0 MPa
Steam consumption 25 kg/h Cooling water
<30°C, >2MPa, >20 l/min
Table 22.2.5 shows the cleaning performance of the solvent. Figure 22.2.6 shows the vapor pressure and combustion range of the solvent. Figure 22.2.7 shows a process diagram of cycloparaffins solvent use in cleaning stamped metal parts. The solvent is recycled after recovery by distillation. Combining the cleaning and the distillation process gives a continuous process. Figure 22.2.8 shows a schematic diagram of distillation process. Table 22.2.6 shows specification of the recovery unit.
22.2.5 HALOGENATED SOLVENT RECOVERY 22.2.5.1 Coating process A process to recover ethylene dichloride, EDC, and methylene dichloride, MDC, using activated carbon fibers, ACF, is discussed in this section. The gas at approximately 50°C, containing EDC and MDC is received from a painting process. It is pretreated by a filter to remove dusts and passed through a heat exchanger to cool it down to approximately 30-40°C. The pre-treated gas is fed to adsorption columns filled with ACF where the gas is cleaned sufficiently for it to be safely released to atmosphere.
1550
Isao Kimura
Figure 22.2.9. Flow diagram of ACF solvent recovery process.
Figure 22.2.10. ACF solvent recovery unit.
Figure 22.2.9 shows the process flow diagram. The process consists of two (or more) adsorption columns to ensure a continuous run of adsorption and desorption cycles. Figure 22.2.10 shows an industrial application of the system.
22.2 Solvent recovery
Table 22.2.7. Performance data for the ACF solvent recovery unit
1551
The desorption is done by steam. The desorbed stream is fed to a series of two Operational & per- condensers in which both the steam and solItem vent are condensed. The condensate is fed formance data to a decanter to separate water and solvent Flow rate 100 m3/min, 20°C by gravity. Water containing approximately Solvent 1-2% solvent is fed to a degassing unit and EDC 46.7 l/h the vaporized solvent is joined to the feed. MDC 23.3 l/h The decanted solvent is refined, stored in a tank, analyzed, and adjusted to its appropriGas concentration Inlet 2,900 ppm ate EDC/MDC ratio. The refined solvent is Outlet 58 ppm recycled for painting. Water from the degassing unit conAdsorption efficiency 98 % tains solvent in a concentration relative to 2 ACF specific surface area 900 - 2,000 m /g the solvent’s solubility. For example, if the Adsorption & desorption cycle 10 min solubility of perchloroethylenene in water is designated as 1, the relative solubility of EDC in water would be 58 and of MDC in water would be 88. In degassing at 60°C and 180 l/min x 30 min the residual EDC content in water is brought below 0.01 ppm (the environmental standard limit is 0.04 ppm) but in the case of MDC it remains a problem to reach to the environmental standard of 0.2 ppm. A treatment by activated carbon may be one of the solutions. Table 22.2.7 shows performance data in the industrial application. The performance of the ACF solvent recovery process can be summarized as follows: Since ACF has lager surface area and smaller bulk density than particulate activated carbon, PAC, ACF’s filling density becomes 1/10 of PAC’s. Advantages: • High recovery rate - The solvent recovery rate of the ACF process is higher than that of the PAC, especially in the case of lower boiling point solvents, such as, MDC, benzene, trichloroethane, etc. • High quality - ACF has low catalytic activity and a short adsorption and desorption cycle time (approx. 10 min). The solvent is less decomposed in the process. In the recovery of chlorine-containing solvents, it produces less acids as decomposition products. Thus it is less corrosive to the materials of construction. Also, the recovered solvent is of better quality. • Broader application - It is applicable to polymerizing monomers and high boiling point compounds. • Light weight, compact and safe - The process unit can be compact because of the short adsorption and desorption cycles and it is safer because there is less heat accumulation. • Energy savings - The process needs less steam consumption in the adsorption and desorption. Disadvantages: • Less flexibility regarding capacity - ACF’s filling volume cannot be readily changed.
1552
Isao Kimura
• Difficult in decreasing outlet temperature - The short cycle in adsorption and desorption makes the cooling process difficult to synchronize.
• Expensive - ACF are more expensive than PAC, thus the unit modification cost is more expensive. 22.2.5.2 Tableting process of pharmaceutical products In a process which produces tablets, a mixture of solvents, containing MDC and methanol and binders is used. Solvents are emitted from the tableting unit. The following process is used to recover the emitted solvent. Feed gas at 50°C from the tableting process is pre-treated by a filter to remove dusts and by a rotary wet scrubber to absorb mainly methanol. During the scrubbing MDC also dissolves in water, although at low concentration. This water is treated in two steps, by a degassing unit at room temperature and by an activated carbon filter so that the concentration of residual solvents is decreased below that of the environmental standard. The gas from the degassing unit is joined to the feed gas at the outlet of the demister of the MDC recovery unit. The gas, the pretreated feed, and gas from the degassing are cooled down to 30 - 40°C before feeding to the MDC recovery unit which consists of at least two columns for adsorption and desorption cycles. After processing the gas is released to the atmosphere. Because little methanol is vaporized at room temperature, the water stream is fed to another degassing unit heated to 65°C to vaporize methanol. For safety purposes, the degassing air flow rate is controlled at 10 m3/min. The methanol laden gas from the heated degassing unit is fed to a catalytic oxidation unit where methanol is decomposed before exhausting the gas stream to the atmosphere. Table 22.2.8 shows the process performance. Table 22.2.8. Performance of the solvent recovery in the tableting process Item
Performance
Flow rate at rotary scrubber
1,000 l/min
Feed concentration
MeOH 10,000 ppm MDC 2,000 ppm
Exhaust concentration
MeOH MDC
Catalytic oxidation flow rate
10,000 l/min
Feed concentration
MeOH 8,500 ppm MDC 1 ppm
Exhaust concentration
MeOH
150 ppm 3 ppm
10 ppm
This is a unique example of a process in which the recovery and the catalytic oxidation steps are combined. In addition to its use in the pharmaceutical industry, the adsorption is a suitable method for the separation of chlorine-containing solvents. Figure 22.2.11 shows the process flow diagram.
22.2 Solvent recovery
1553
Figure 22.2.11. Flow diagram of tablet binder recovery process.
22.2.6 ENERGY RECOVERY FROM WASTE SOLVENT Table 22.2.9. Specifications of the VOC Table 22.2.10. Specifications of the waste oxidation unit heat boiler Type of unit
Direct firing
Type of unit
Horizontal, multi-tube
Total gas flow rate
2,000 kg/h
Use for
Steam as heating medium
Solvent flow rate
160 kg/h
Waste gas flow rate
5,000 Nm3/h
Solvent
Acetone, methanol, IPA
Temp., waste gas
Inlet: 800, outlet: 350°C
Feed temp.
60 - 100°C
Supply water temp.
20°C
Oxidation temp.
800 °C
Steam generated
1,300 kg/h
Performance, THC meter 97 % Supplemental fuel
Heavy oil, max170 l/min
It is not always economically justified to recover solvents if they are at very low concentrations and in mixtures. It may be more economical to burn the solvent blend in a boiler to save fuel. Tables 22.2.9 and 22.2.10 contain the specifications of industrial combustion units.
1554
Isao Kimura
Figure 22.2.12. Flow diagram of energy recovery process.
Savings of energy The net calorific values for acetone, methanol and IPA is 6,962, 5,238 and 7,513 kcal/kg, respectively, and the simple mean value of their mixture is 6,600 kcal/kg. Let us assume that solvents are generated from a process at a maximum rate of 160 kg/h. The total heat generation of the VOC oxidation unit is as follows: 6,600 x 160 = 1,056 x 103 kcal/h Heavy oil = 10,200 kcal/kg; specific gravity of heavy oil = 0.86 The saved energy is as follows: 1,056 x 103 / 10,200 = 103.5 kg/h 103.5 / 0.86 = 120 l/h [1] If no solvent is supplied, the required heavy oil is 170 l/h [2] If solvent is supplied at max. rate, the required heavy oil is only 50 l/h. [1] - [2] = 50 l/h Thus, energy saving performance is 71% ([1]/[2] x 100 = 71 %) Figure 22.2.12 shows the process flow diagram and Figure 22.2.13 shows an industrial application of the energy recovery unit. The VOC laden gas from a painting plant is supplied through a filter and is blown to a heat exchanger and then to succeeding units. The fan drive motors are controlled by inverters to adjust the flow rate. In the heat exchanger the feed is preheated with a waste heat boiler’s off gas, at a temperature of 350°C, to 60 - 100°C before it is fed to a VOC oxidation unit. In the oxidation unit, heavy oil and VOC are mixed and burnt. At approximately 800°C, VOC is burnt out with yield of more than 97 %. The off gas at 800°C is fed to a waste heat boiler where it generates 1,300 kg/h steam at 8 MPa. The heat transfer gas at 350°C is used for the preheating the feed gas.
22.3 Solvent treatment in a paints and coating plant
1555
Figure 22.2.13. Energy recovery unit.
REFERENCES 1 2
T.Takahashi and S.Ito, Kogyo Toso, No. 150, 75-79, 1998. Seibu Giken’s Catalog. VOC Concentrator (1997).
22.3 SOLVENT TREATMENT IN A PAINTS AND COATING PLANT Denis Kargol OFRU Recycling GmbH & Co. KG, Babenhausen, Germany
The Swiss coating manufacturer Karl Bubenhofer AG was planning to acquire various wash plants for cleaning batch, mixing and multi-trip containers efficiently. The contaminated solvents obtained during the wash process were then to be treated automatically by distillation. The distillation plant had to have a suitable output, operate continuously and be simple to operate. Another requirement was the ability to combine this plant with the wash plants in a new building complex. A distillation plant suitable for this purpose was supplied by OFRU Recycling in Babenhausen/Germany. Karl Bubenhofer AG, known by the name, “Kabe-Farben”, is regarded as one of the largest paint and coating producers in Switzerland and has been producing coating materials since 1926. The product range covers architectural paints and plasters, powder coatings and industrial liquid paints. The company´s wide range of customers includes industrial, commercial and contract paint companies as well as resellers at home at abroad. Kabe offers a special service in cooperation with regional painters’ associations. For a small charge customers and painters can offload their waste paints and contaminated solvents to Kabe in special multi-trip containers. About 160 tonnes per year of spent solvents
1556
Denis Kargol
Figure 22.3.1. Soiled multi-trip container.
are obtained by way of this disposal system. Even before the planning phase, these contaminated spent solvents were treated with the company’s own small distillation plant. Until now, manual cleaning of these multi-trip containers and large batching and mixing vessels from production was particularly inconvenient and labor-intensive for Kabe. A wash plant for mobile large containers (250-1800 l volume) and for small drums pails (about 80-100 per day) would obviate the unpleasant manual cleaning task in the future. A combination of container wash following by distillation therefore seemed very practicable. The treated clean solvent would feed the wash plants and ensure thorough cleaning. The company was also interested in further processing a part of the distillate for Kabe products. When the solvent requirement of the container wash plant and the customers’s spent solvent quantities were added together, the total was about 200 tonnes of contaminated solvents per year. Solvents such as toluene, ketones, esters, ethyl acetate, xylene, glycol and petroleum spirits were to be separated almost completely from solids such as alkyd, 2-pack, epoxy and polyester resins. The solvent recovery plant had to meet the following requirements: 1. Combination with wash plants and storage tanks 2. Distillation with a high rate of recovery 3. Continuous operation (24 hour) 4. Distillation output of 200-400 l/h 5. Low maintenance 6. Minimum operating and control times 7. Residue to be contaminated only slightly with residual solvent Kabe-Farben found the solution Rio Beer, a wash plant producer from Switzerland. Rio Beer supplied not only the wash plant design but also integrated OFRU Recycling from Babenhausen in Germany with its solvent recovery plants. OFRU Recycling provided the necessary expertise from the paints and coatings sector. Kabe-Farben visited the OFRU pilot plant in Babenhausen and was able to recover 224 kg of clean solvent within a few hours from 300 kg of contaminated solvent. A quantity of 76 kg of solid paint slurry was then discharged from the distillation unit. An analysis afterwards showed that the residue contained only 5.5% solvent. More than 74% of the solvent could therefore be recovered in the first instance. The solution Kabe decided in favor of a vacuum evaporator with an integrated stirrer. The ASC-200 model together with three solvent tanks stand together in an explosion-proof room. The cus-
22.3 Solvent treatment in a paints and coating plant
1557
Figure 22.3.2. Solvent circuit and treatment.
tomer’s contaminated spent solvents are poured in through a large sieve by means of a drum tilting device. Twin diaphragm pumps then convey the contaminated solvent to the first soiled solvent tank. In view of the large proportion of paint in the solvent (5-20 vol.%), the solvent is kept homogeneously in motion in the tank by means of the stirrer. Another soiled solvent tank is connected up to the wash circuit and supplies the solvent for pre-washing. The ASC-200 model used by Kabe operates entirely under vacuum. A considerable increase in capacity is achieved by means of the vacuum equipment; moreover, distillation is milder and residue is largely solvent free. The contaminated product is introduced automatically by suction into the evaporator by means of a magnetic valve and the vacuum until it reaches the height of a capacitive level controller, which keeps the solvent level constant during continuous operaFigure 22.3.3. Wash plant for mixing vessels. tion. The filling operation is adjusted in such a way that no excessive cooling or interruption in the distillation process takes place. The solvent is usually heated by means of a double jacket with its own thermal oil facility. Kabe, on the other hand, operates a catalytic waste air purification plant. The solvent-air mixture obtained in the plant is converted catalytically to heat and stored as an energy store. A small portion of the heat thereby obtained is absorbed in our case by a thermal oil system and released to the distillation plant. A stirrer is fitted inside the ASC-200 distillation plant. Its scrapers rotate inside the inner wall of the vessel. They keep the heat transfer surface free from encrustation and guarantee uniform heat transfer into the boiling solvent. The stirrer keeps the liquid in motion
1558
Denis Kargol
Figure 22.3.5. Residue transferred to PE pail.
and thereby promotes evaporation of the solvent. Finally, the solvent vapor reaches a water-cooler tubular condenser provided with a cooling water control unit. The condensed, clean distillate is pumped from here to the distillate tank. This tank has a level control and now feeds the clean solFigure 22.3.4. ASC-200 distillation plant in the explo- vent directly into the wash plant for a secsion-proof room. ond wash. The distillation process takes place fully automatically and lasts about 2 days. Distillation of the bottom product takes place independently afterwards. After the residue has been thickened to a predetermined viscosity, the equipment shuts down automatically. The still is drained by means of a discharge valve at the bottom of the cone. The 200-300 kg of compact greyish residues are transferred to 20 completely empty 20 liter plastic pails and then fed to an inexpensive solid waste incineration process. No further cleaning operations are required at the plant so the stirred distillation unit is ready for use again immediately. The plant used at Kabe treats about 200 liters of solvent per hour. At about 200 tonnes per year it is therefore being used to 60-70% of its capacity at present. Since its installation, the distillation plant has been operating for 5 days per week with very little maintenance. A few mechanics maintain the plant twice a year without external support.
22.4 Application of solar photocatalytic oxidation
1559
22.4 APPLICATION OF SOLAR PHOTOCATALYTIC OXIDATION TO VOC-CONTAINING AIRSTREAMS K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, and G. C. Glatzmaier National Renewable Energy Laboratory, Golden, CO, USA
22.4.1 SOLVENT DEGRADATION BY PHOTOCATALYTIC OXIDATION Photocatalytic solvent oxidation had been demonstrated at the pilot scale in two recent field tests located at McClellan Air Force Base (AFB) in Sacramento, California and at the Fort Carson U. S. Army Installation in Colorado Springs, Colorado (Watt et al., 1999; Magrini et al., 1998). The objective of the tests was to determine the effectiveness of solar-powered photocatalytic oxidation (PCO) treatment units for destroying emissions of chlorinated organic compounds (trichloroethylene and dichloroethylenes) from an air stripper at ambient temperature and destroying paint solvent emissions (toluene and MEK) from a painting facility at higher temperatures. Goals for field testing these solar-driven systems were to gather real-world treatability data and establish that the systems maintained performance during the duration of the testing. The photocatalytic oxidation process can effectively destroy hazardous organic pollutants in air and water streams. Although treatment systems will vary depending on the type of stream being treated, the basic process remains the same. The key ingredient is the photoactive catalyst titanium dioxide (TiO2), which is an inexpensive, non-toxic material commonly used as a paint pigment. When TiO2 is illuminated with lamps or natural sunlight, powerful oxidizing species called hydroxyl radicals form. These radicals then react with the organic pollutant to tear it apart and ultimately form carbon dioxide (CO2) and water (Phillips and Raupp, 1992). When halogenated organics are treated, dilute mineral acids like HCl form. The process works at both ambient and mildly elevated temperatures (>200oC) (Fu et. al., 1995; Falconer and Magrini, 1998). Researchers throughout the world have been investigating PCO as an advanced oxidation technology for treating air and water streams contaminated with a variety of organic and inorganic compounds (Blake, 1996; Cummings et al., 1996). The susceptibility of an organic species to complete oxidation is typically reported in terms of photoefficiency, defined as the number of molecules of contaminant oxidized to carbon dioxide, water, and simple mineral acids divided by the number of photons incident on the catalyst. These values vary widely, depending on the reactor design, catalyst geometry and the compound of interest. Much of the work on photocatalytic oxidation focuses on treating the halogenated organics trichloroethylene (TCE) and perchloroethylene, contaminants commonly found in ground water sources. These compounds and other chlorinated ethylenes typically react rapidly with TiO2 and photons at efficiencies greater than 100%. These rates are likely due to chain reactions propagated by chlorine radicals (Luo and Ollis, 1996; Nimlos et. al. 1993; Yamazaki-Nishida, 1996). Paint solvent emissions generally consist of toluene, xylenes, ketones and acetate vapors. Measured photoefficiencies for benzene and other aromatic compounds like toluene are typically less than 5% (Gratson et. al., 1995; d’Hennezel
1560
K. A. Magrini, et al.
and Ollis, 1997). Besides exhibiting low photoefficiencies, aromatic species tend to form less reactive, nonvolatile intermediates during gas phase PCO. These intermediates build up on the catalyst surface and block or inhibit the active catalytic sites for further reaction (Larson and Falconer, 1997). The addition of heat and small amounts of platinum to the TiO2 catalyst overcome these problems (Falconer and Magrini, 1998; Fu et. al., 1996). Oxygenated organics such as ethanol and acetone have photoefficiencies typically around 1%-10% (Peral and Ollis, 1992). Several field demonstrations of PCO using sunlight to treat groundwater contaminated with TCE have been reported (Mehos and Turchi, 1993; Goswami et. al., 1993). These field tests found that nontoxic constituents in the water can non-productively react with or “scavenge” the photogenerated hydroxyl radicals and reduce the rate of the desired reaction. Common scavengers such as humic substances and bicarbonate ions increase treatment costs for the technology (Bekbolet and Balcioglu, 1996). Turchi and co-workers, (1994) found that by air stripping the volatile contaminants from the water stream, the regulated compounds at many contaminated sites could be transferred to the air, leaving the radical scavengers behind. The water can then be safely discharged and the air effectively treated with PCO. The improved photoefficiency reduces treatment costs and more than offsets the added cost of air stripping these contaminants from water. Read et. al., (1996) successfully field tested a lamp-driven, PCO system on chloroethylene vapors from a soil vapor extraction well located at DOE’s Savannah River Site. Magrini et al., 1996, and Kittrell et al., 1996, both used modified TiO2 catalysts and lamp-driven reactors to treat VOCs representative of semiconductor manufacturing and contact lens production, respectively. 1,2-dichloroethane, stripped from contaminated groundwater, was successfully treated in a pilot scale PCO demonstration at Dover AFB (Rosansky et al., 1998). The second field test assessed PCO to treat paint solvent vapors. Painting operations for military and civilian vehicles are conducted in ventilated enclosures called paint booths. Filters in the exhaust ducts trap paint droplets from the paint overspray while the VOC-laden air is typically exhausted through roof vents. The vent emissions can contain several hundred parts per million (ppm) of the paint solvents, which continue to evaporate from the vehicle after painting is complete. Most types of paint generally contain significant amounts of VOCs such as toluene, a suspected carcinogen, as well as other hazardous solvents such as methyl ethyl ketone, methyl isobutyl ketone, hexanes, xylenes, n-butyl acetate, and other components in lesser amounts. Current technologies for treating these emissions include catalytic combustion of the vapors over supported Cu and Cr-oxides at temperatures of 350oC (Estropov et. al, 1989); air-flow reduction and recirculation strategies (Ayer and Darvin, 1995); and regenerative thermal oxidation at near incineration temperatures (Mueller, 1988). The use of platinized TiO2 and temperatures of 180-200oC provide significant energy savings in treating these emissions. Our goals for testing the paint booth emissions was to gather real-world treatability data and establish that the system maintained performance during the duration of the testing. 22.4.2 PCO PILOT SCALE SYSTEMS 22.4.2.1 Air stripper application Figure 1 shows a schematic of the pilot-scale system used at McClellan AFB. This system, scaled from an optimized laboratory reactor, was fabricated by Industrial Solar Technology
22.4 Application of solar photocatalytic oxidation
1561
Figure 22.4.1. Schematic of the 10 SCFM solar photoreactor system used at McLellan AFB.
Corporation (Golden, CO). The reactor, 1.22 m wide by 2.44 m, consisted of a fiberglass-reinforced plastic I-beam frame. A transparent fluoropolymer film, treated to accept adhesives, formed the front and back windows of the reactor. The film windows were attached to the reactor frame with foam tape coated on both sides with an acrylic adhesive. The catalyst, titanium dioxide (Degussa P25), was coated onto a structured, perforated polypropylene tubular packing commonly used in oil-water separators. The TiO2 was suspended in water as a slurry and sprayed onto the tubular supports with a new paint sprayer until the supports were opaque. Fluid modeling of airflow through the reactor showed that a 5.1 cm PVC manifold, located at the inlet and outlet of the reactor, would provide an even flow distribution. Small (0.6 cm) holes, drilled into the manifold at one inch intervals, provided the even flow distribution required for efficient contact of the contaminated air with the catalyst. A two-inch diameter pipe from the outlet of the air stripper supplied the contaminated air stream to the reactor. The air stream was first passed through to a tube-in-shell heat exchanger for partial dehumidification. A chiller circulated 2oC ethylene glycol through the tube side of the heat exchanger. Chilling, which reduced the relative humidity of the air stream from near saturation to less than 20% at 20oC, was required because high humidity reduces the TCE destruction rate, likely because of competitive adsorption between moisture and TCE molecules at the catalyst surface (Fan and Yates, 1996). By lowering the humidity to less than 20% (@ 20°C), the reaction rate was sufficient to ensure complete TCE destruction. This effect was observed by Hung and Marinas (1997) in a lamp-illuminated annular reactor in which TCE conversion was not affected by relative humidity up to 20%; conversion deteriorated as the humidity level reached saturation. Laboratory tests of the system demonstrated that at TCE concentrations of near 100 ppmv, less than 15% of the TCE in the airstream transferred to the condensate formed in the heat exchanger. The condensate (approx. 1-2 liters per day) could be fed back into the air stripper for treatment. On the reactor, GC sample ports and temperature and pressure sensors provided monitoring of reactor inlet and outlet VOC concentrations and temperature and pressure drop throughout the system. A portable gas chromatograph provided on-line VOC organic anal-
K. A. Magrini, et al.
1562
ysis of the inlet and outlet air streams. A portable light intensity monitor was used to measure the global horizontal light intensity at the surface of the photoreactor. The calibrated radiometer was mounted on the reactor framework which faced south at an angle of 40o to the horizon. Two blowers in series were used to draw contaminated air into the system. The blowers were located at the end of the air handling system so that the entire system was maintained under slight negative pressure. Because TCE oxidation can produce intermediates such as phosgene, (Nimlos et. al, 1993; Read et. al., 1996; Fan and Yates, 1996), keeping the system under vacuum prevented release of any toxic vapors. The exhaust gases from the blower were returned for final treatment by a caustic contacting tower. Any hydrochloric acid and phosgene formed from TCE oxidation were neutralized and hydrolyzed respectively by the caustic scrubber. The contaminated airstream provided by the air stripper unit contained volatile organic compounds as analyzed from Summa canister tests by Air Toxics, Ltd. using EPA method TO-14. These compounds, listed in Table 22.4.1, consisted primarily of trichloroethylene and trace quantities of dichloromethane, 1,1-dichloroethene, cis-1,2dichloroethane, chloroform, carbon tetrachloride, and benzene. The other compounds present in the outlet presumably formed during treatment. Table 22.4.1. Results from the Summa canister tests of inlet and outlet reactor air streams during the McClellan tests. The two inlet and two outlet samples were taken sequentially and represent approximately replicate samples Compound
Outlet 1, ppbv Outlet 2, ppbv
Detection limit, ppbv
Inlet 1, ppbv
Inlet 2, ppbv
Chloromethane
ND
ND
4.8
4.1
1.5-30
Dichloromethane
36
36
30
32
1.5-30
1,1-Dichloroethane
ND
ND
17
17
1.5-30
1,1-Dichloroethene
200
200
ND
ND
1.5-30
cis-1,2-Dichloroethane
220
240
6
2.2
1.5-30
Chloroform
250
260
280
280
1.5-30
Carbon tetrachloride
700
730
790
790
1.5-30
Benzene
170
190
120
110
1.5-30
1,2-Dichloroethane
ND
ND
12
ND
1.5-30
Trichloroethylene
13000
13000
660
280
1.5-30
Acetone
ND
ND
84
69
9.8-120
1,4-Dioxane
ND
ND
10
ND
9.8-120
22.4.2.2 Paint booth application For the paint solvent application, a slipstream of a paint booth vent located at Fort Carson was analyzed to determine the components present with rapid scan gas chromatograph. On any operational day, several paint types are sprayed. The solvent emissions content thus varies during painting operations. In general, the amounts and types of solvent emissions from the Fort Carson paint booth are listed in Table 22.4.2.
22.4 Application of solar photocatalytic oxidation
1563
Figure 22.4.2. Schematic of the 10 SCFM parabolic trough, solar photoreactor system used at Fort Carson U. S. Army Installation.
Table 22.4.2. Breakdown of solvent emissions from painting operations at Fort Carson in 1994. VOC refers to volatile organic compounds, HAP to hazardous air pollutants, MEK to methyl ethyl ketone, and MIBK to methyl isobutyl ketone Paint Type
Total usage, l/yr
VOC, lbs/yr
HAP, lbs/yr
MEK, lbs/yr
Toluene, lbs/yr
MIBK, lbs/yr
Xylene, lbs/yr
Solvent base
19,656
110,074
5405
3931
491
0
983
983
7235
722
552
0
174
0
Thinner
A small field test system, consisting of a photoreactor packed with Degussa TiO2-coated glass beads, was used to assess treatability of the slipstream and provide design data for the pilot scale system. The small reactor operated with a flow rate of 20 liters/minute provided by a pump connected to the paint booth exhaust vent line and variable reactor temperature (ambient to 200oC). Inlet and outlet VOC concentrations were simultaneously measured by two gas chromatographs directly connected to both lines. This arrangement provided real time monitoring of the inlet and outlet VOC concentrations during painting operations. The resultant treatability data was used to size a 5-SCFM pilot-scale system (Figure 22.4.2) also built by Industrial Solar Technologies. This system consists of a parabolic trough reflector to focus incident sunlight onto a receiver tube containing the catalyst. The trough has a 91 cm aperture width and is 243 cm long. The reflective surface is covered with SA-85 (3M Company), a polymer film that reflects greater than 90% of the incident radiation onto the catalyst tube, providing both heat and light for reaction. The catalyst-containing receiver tube is a 3.81 cm I.D. borosilicate tube packed with 1.9 cm ceramic berl saddles coated with 2.0 wt% platinum on TiO2. A higher weight loading of platinum was used in the field test because of the variable VOC composition of the paint booth stream. Operating temperatures of 200oC were obtained, depending on cloud cover, during most of the day. Note that a thermally heated, lamp-illuminated system will provide similar performance as the solar-based system described here. The reflective trough rotates around the fixed receiver tubes by means of a gear motor. A control and photodiode arrangement activate the gear motor, permitting the trough to track the sun throughout the day. Paint booth emissions were sampled directly from the exhaust stream via 10.1 cm aluminum ducting. On-line analysis of the inlet and outlet concentrations was provided by sample lines connected to the MTI gas samplers. Reactor temperature and inlet flows were measured continuously.
1564
K. A. Magrini, et al.
Figure 22.4.3. Solar insolation (left) and fractional TCE conversion (right) versus illumination time. Flow rate and relative humidity of the incoming air stream was 10 SCFM at 20% RH. Data taken on April 8-9, 1997.
22.4.3 FIELD TEST RESULTS 22.4.3.1 Air stripper application The field test at McClellan AFB took place during a 3-week period in April, 1997. System operating conditions included two flow rates (10 and 20 SCFM) and variable light intensity. The photoreactor system, designed to perform optimally at 10 SCFM with natural sunlight, was also operated at 20 SCFM to challenge the system. Because of varying solar illumination levels during the day, the system can operate for longer periods at lower flow rates. Residence times in the reactor at 10 and 20 SCFM were 0.50 sec and 0.25 sec, respectively. Figure 22.4.3 shows TCE conversion and UV intensity versus illumination time at 10 SCFM for two days of operation. UV intensities were measured with a UV radiometer (UHP, San Gabriel, CA, Model UHX). TCE inlet concentrations varied from 10-15 ppmv. Detection limits for the gas chromatograph were measured at 0.5 ppmv. Figure 22.4.3 indicates that conversions greater than 95% at 10 SCFM are achieved with UV intensities at or greater than 1.5 mW/cm2. At 10 SCFM, continuous destruction greater than 95% was achieved for at least 6 hours per day on two clear days in April at this latitude (37°N). The length of operation for a solar-driven system will vary with time of year, latitude, and altitude due to the effect each parameter has on available solar-UV intensity.
Figure 22.4.4. Solar insolation (left) and fractional TCE conversion (right) versus illumination time during periods of partial cloud cover. Flow rate and relative humidity of the incoming stream was 10 SCFM and 20% RH. Data taken on April 7 and April 16, 1997.
22.4 Application of solar photocatalytic oxidation
1565
Figure 22.4.5. Solar insolation (left) and fractional TCE conversion (right) versus illumination time. Flow rate and relative humidity of the incoming air stream was 20 SCFM at 20% RH. Data taken on April 10, 14, and 15, 1997.
Figure 22.4.4 shows the effect of clouds on solar insolation (left) and TCE conversion (right) as a function of illumination time at 10 SCFM for two separate days. Again TCE conversions greater than 95% were obtained when the insolation levels were greater than 1.5 mW/cm2. In general, clear skies had UV intensities of 2 mW/cm2 or better. Haze and high clouds reduced the intensity to 1.5 mW/cm2, and thicker clouds reduced available intensity to about 1 mW/cm2. Conversions greater than 90% were still achievable during periods of significant cloud cover due to the photoreactor’s ability to use global horizontal (diffuse) UV illumination. TCE conversion versus solar insolation for three days of operation at 20 SCFM is shown in Figure 22.4.5. Doubling the flow rate halves the residence time of the gas in the photoreactor. Conversions greater than 95% are achieved with UV intensities greater than 2.0 mW/cm2. The reproducibility of TCE conversion indicates that the little or no catalyst deactivation occurred during the three weeks of testing. Pooling all of the data taken at the two flow rates provides a relation between TCE conversion and UV intensity (Figure 22.4.6). Significant scatter, which exists for both flow rates, is likely due to the rapidly varying UV intensity levels during cloud events. The variable illumination makes it difficult to establish a precise correlation between an isolated UV measurement and the UV exposure received by the gas as it flows through and reacts with the photocatalyst bed. The 10 SCFM data suggests a square root relationship between UV intensity and conversion. The square root dependence was observed in a continuous flow reactor when TCE concentrations were less than 60 ppmv (Nimlos et. al., 1993). Higher concentrations Figure 22.4.6. Pooled UV intensity versus TCE conversion data for yielded an approximately linear dependence. all 10 SCFM and 20 SCFM runs.
1566
K. A. Magrini, et al.
On the last day of testing, grab samples were collected in Summa canisters and delivered to a local analytical laboratory (Air Toxics, Ltd., Folsom, CA) for TO-14 analysis to confirm the analyses provided by the MTI gas chromatographs and to provide outlet TCE concentrations that were below our GCs detection limits. Replicate samples were reported as 13000 ppbv for TCE in inlet air, and 280 and 660 ppbv TCE in the outlet (Table 22.4.1). This corresponds to 97.8% and 94.9% removal of TCE. The reactor outlet airstream contained most of the dichloromethane and carbon tetrachloride present in the inlet stream. This result was expected, as single carbon haloorganics do not easily photo-oxidize (Jacoby et. al, 1994). Nearly all of the 1,1-dichloroethene and cis-1,2-dichloroethane were destroyed (destruction efficiencies of 97.3% and 99.1% respectively). This too was expected, as dichlorinated ethanes and ethenes photooxidize almost as rapidly as TCE. Small amounts (<20 ppbv) of chloromethane, 1,1-dichloroethane, 1,2-dichloroethane and acetone, and 1,4-dioxane appeared in the reactor outlet and are likely reaction intermediates from TCE oxidation. About 40% of the benzene in the inlet stream did not come out of the reactor. This result is in line with recent results in which benzene was oxidized with a sol-gel TiO2 catalyst (Fu, X. et. al., 1995). Though not all of the benzene reacted, that which did formed only CO2 and H2O. Adding platinum to the TiO2 sol significantly improved benzene oxidation. Transient reaction studies of benzene and Degussa TiO2 thin films indicate that benzene oxidizes rapidly at 300K to form strongly bound surface intermediates that oxidize very slowly (Larson and Falconer, 1997). Other work with thin films of TiO2 and benzene shows that benzene does not form intermediates when water vapor is present. In the absence of water, acetone forms and the catalyst becomes coated with a durable brown material (Sitkiewitz, S. and Heller, A., 1996). Because these tests were run at an average of 20% RH and the catalyst exhibited stable activity, it is likely that benzene was oxidized. Thus, since the catalyst activity did not deteriorate over the course of testing, it may be possible that low concentrations of aromatic species can be tolerated and treated in a solar photocatalytic reactor. It is also possible that the benzene was absorbed by the catalyst support material. 22.4.3.2 Paint booth application Figure 22.4.7 shows the real time test results using the lamp-driven, small field scale reactor for treating VOC emissions from painting operations at Fort Carson. The inlet and outlet VOC concentrations are given in ppm. Test data begins at 8:10 AM on June 25, 1996 and ended at 11:55 for this run. The painting operations are conducted in a batch mode. Vehicles are placed in the paint booth, painted, then removed from the booth to the parking area, where they dry. The mixture of VOCs emitted from the aliphatic polyurethane green military paint being sprayed conFigure 22.4.7. Inlet and outlet VOC concentrations from the Fort Carson paint booth using the small, lamp-driven, sisted of methyl ethyl ketone, methyl field test reactor.
22.4 Application of solar photocatalytic oxidation
1567
isobutyl ketone, toluene, n-butyl acetate, xylenes, and ethyl benzene. When the reactor reached an operating temperature of 170oC, complete VOC destruction was achieved for the entire run except for a period of peak VOC emissions beginning at 9:30 and ending at 9:45. Thus, during this four hour period, the PCO process destroyed virtually all of the VOCs in the slip stream from the paint booth exFigure 22.4.8. The effect of sunlight fluctuations on the haust. The total time averaged photothermal destruction of methyl ethyl ketone in the parabolic conversion is estimated to be trough, photothermal reactor. MEK conversion, insolation level, and greater than 95%. These data were reactor temperature are plotted as a function of continuous outdoor used to design the solar-driven, illumination time. photothermal reactor system. Based on the lamp tests, the 5-SCFM thermal photoreactor system was tested with methyl ethyl ketone (MEK), a paint solvent surrogate. Figure 22.4.8 shows the results of on-sun testing with this system. The Y-axis plots three factors: temperature in oC x100 MEK, conversion as a fraction with 1 equaling 100% conversion, and solar insolation as mW/cm2. The X-axis plots illumination time in minutes. This test was performed on a hazy day with numerous clouds to evaluate the system under challenging conditions. Though conversion and reactor temperature parallel the insolation, they do so in a less than 1:1 ratio. This is likely due to the square root dependence of reaction rate on light intensity and to the heat transfer (insulative) characteristics of the ceramic support. 100% conversion was achieved when insolation levels were above 3.5 mW/cm2 and reactor temperatures above 175oC. The photothermal reactor was tested in July, 1997 for three weeks on paint booth emissions using natural sunlight. A representative data set for July 22, 1997 is shown in Figure 22.4.9 in which the total VOC concentration for the inlet and the outlet of the parabolic trough, photothermal reactor is plotted against illumination time. The solvent stream, which varied with time, generally contained about 140 ppmv of toluene, methyl ethyl ketone, methyl Figure 22.4.9. VOC concentration from paint solvents as a function isobutyl ketone, xylenes, butyl acof illumination time during photocatalytic treatment with a concenetate and ethyl benzene. The opertrating, parabolic trough, solar photoreactor.
1568
K. A. Magrini, et al.
ating temperature was from 190 to 208oC, dependent on available illumination. On-sun refers to turning the concentrating receiver into the sun, which initiates heating and photoreaction. Off-sun refers to turning the receiver away from the sun (photoreaction and heating stops). Inlet and outlet VOC concentrations, which matched closely when the reactor was off-sun, were taken to ensure that thermal reaction as the reactor cooled and adsorption of contaminants onto the catalyst surface were not significant. No attempt was made to decouple photoreaction from thermal reaction because of rapid temperature changes when the trough was rotated off-sun and the difficulty of thermally heating the catalyst. The results of 3 weeks of testing demonstrated that the photothermal treatment system has destruction and removal efficiencies (DREs) in excess of 95% under conditions where incident UV irradiation provided operating temperatures from 150 to 200oC. No loss of performance of the catalyst occurred. Significant results from these pilot-scale field tests: • A non-tracking, ambient-temperature, solar-powered photocatalytic reactor consistently achieved greater than 95% destruction of 10-15 ppmv TCE for 4 to 6 hours of operation per day during April at a latitude of 37°N. Catalyst performance remained constant during three weeks of testing. • 95% destruction was achieved at 10 SCFM with UV intensities greater than 1.5 mW/cm2 and at 20 SCFM at UV intensities greater than 2.0 mW/cm2. • Greater than 95% destruction of a variable paint solvent stream (140 ppm total VOC) was achieved with a concentrating solar photoreactor operated with ambient sunlight at temperatures of 180-200oC. These operating temperatures were achievable for 4-8 hours of operation per day in July at a latitude of 39.5oN. Catalyst performance remained constant during three weeks of testing. 22.4.4 COMPARISON WITH OTHER TREATMENT SYSTEMS The design and fabrication of the photocatalytic reactor system used in the McLellan field test was provided by NREL engineers and the Industrial Solar Technology Corporation (IST), Golden, CO. IST estimates that the cost for a 200-SCFM system is about $14,000.00, depending on the site requirements for hours of operation per day and the site solar potential. Costs to construct the reactors could be significantly reduced through economies of scale and large scale production likely could reduce reactor costs by half.21 Other purchased equipment costs for the photoreactor include a condenser/chiller and an instrumented blower/scrubber skid system for air handling. The level of instrumentation on the skid directly affects cost. The system described here was instrumented more than would be required for a commercial system. Equipment and installation costs for comparably sized carbon adsorption and thermal oxidation systems were calculated using EPAs HAPPRO v.2.1 software. Thus, total direct costs for the three systems are very close and range from an estimate of $131K for thermal oxidation, $119K for PCO, to $97K for carbon adsorption. Indirect costs were estimated using a factor of 0.28 x total purchased equipment cost. This value includes engineering, construction fee, and start-up costs. (Vatavuk, 1994) Total capital required for all three systems is approximately equal and ranges from $121K to $164K. Operating costs are also similar and range from $23 to $28K and total annual costs range from $37K for thermal oxidation to nearly equivalent values for carbon adsorption and PCO at $32K and $33K, respectively. No fuel is required for PCO and carbon adsorption. Details of the calculations are provided in the footnotes following Table 22.4.3.
22.4 Application of solar photocatalytic oxidation
1569
Table 22.4.3. Comparison of estimated costs for similarly sized PCO, carbon adsorption, and thermal oxidation VOC treatment systems Direct Costs
Solar PCO non-concentrating
Carbon Adsorption
Thermal Oxidation
Purchased Equipment (PE) Photoreactor1 Condenser/Chiller2 Instrumented Blower/Scrubber Skid3
14,000 10,625 60,000
Total PE (TPE)4 Installation
91,395 27,419
74,5005 22,350
100,5005 30,150
118,814
96,850
130,650
29,923
24,391
32,904
148,736
121,241
163,554
Operating Costs Electricity Fuel10 Maintenance, Labor and Material
2,880 0 20,376
1,440 0 19,308
1,440 5,000 20,952
Direct Annual Costs Indirect Annual Costs
23,256 8,664
25,748 7,063
27,392 9,527
Total Annual Costs
31,920
32,811
36,919
56,135
25,549
63,456
281
263
318
Total Direct Costs Indirect Costs Indirect Installation Costs7 Total Capital Required8 9
12
Levelized Cost
Levelized Cost/CFM 1
IST Corp., Golden, CO. Non-concentrating reactor cost is current and estimated price reductions anticipated from economies of scale at higher production levels. 2 Peters and Timmerhaus, 1991. 3 Zentox Corp., Ocala, FL. Concentrating vs. non-concentrating is primarily a function of instrumentation level. 4 Includes 0.08 x PE for taxes and freight (Vatavuk, 1994). 5 Calculated using EPAs HAPPRO v.2.1 software. 6 0.38 x PE (Vatavuk, 1994). 7 0.28 x TPE, including engineering costs, construction fee, and start-up costs (Vatavuk, 1994) 8 Includes 6% contingency. 9 Based on $0.08/kWh, 10 kW blower and chiller costs. 10 Based on $0.33/ccf natural gas. 11 0.06 x Total Capital Required for taxes, insurance, and administration. 12 Based on 10% interest rate, 10 year depreciation.
REFERENCES Ayer, J. and Darvin, C. H., 1995, “Cost Effective VOC Emission Control Strategies for Military, Aerospace, and Industrial Paint Spray Booth Operations: Combining Improved Ventilation Systems with Innovative, Low-Cost Emission Control Technologies”, Proceedings of the 88th Annual Meeting of the Air and Waste Management Association, Vol. 4B, 95-WA77A.02. Blake, D.M., 1996, Bibliography of Work on the Heterogenous Photocatalytic Removal of Hazardous Compounds from Water and Air, Update Number 2 to October 1996. NREL/TP-430-22197. Golden, CO: National Renewable Energy Laboratory. Bekbolet, M. and Balcioglu, I., 1996, Water Science and Technology, Vol. 34, No. 9, pp. 73-80.
1570
K. A. Magrini, et al.
Cummings, M. and Booth, S. R., Proc. Int. Conf. Incineration Therm.Treat. Technol. (1996), pp. 701-713. d’Hennezel, O. and Ollis D. F., 1997, Journal of Catalysis, Vol. 167, No. 1, pp. 118-126. Evstropov, A. A., Belyankin, I. A., Krainov, N. V., Mikheeva, T. Ya., Kvasov, A. A., 1989, “Catalytic Combustion of Gas emissions from A Spray-Painting Chamber”, Lakokras. Mater. Ikh. Primen, Issue 1, pp. 115-118. Falconer, J. L. F. and Magrini-Bair, K. A., 1998, “Photocatalytic and Thermal Catalytic Oxidation of Acetaldehyde on Pt/TiO2", accepted for publication in the Journal of Catalysis. Fan, J. and Yates, J. T., Journal of the American Chemical Society, Vol. 118, pp. 4686-4692. Fu, X., Clark, L. A., Zeltner, W. A., Anderson, M. A., 1996, “Effects of Reaction Temperature and Water Vapor Content on the Heterogeneous Photocatalytic Oxidation of Ethylene”, Journal of Photochemistry and Photobiology A, Vol. 97, No. 3, pp. 181-186. Fu, X. , Zeltner, W. A., and Anderson, M. A., 1995, Applied Catalysis B: Environmental, Vol. 6, pp. 209-224. Gratson, D.A.; Nimlos, M.R.; Wolfrum, E.J., 1995, Proceedings of the Air and Waste Management Association. Hung, C.-H. and Marinas, B. J. , 1997, Environmental Science and Technology, Vol. 31, pp. 1440-1445. Jacoby, W.A.; Nimlos, M.R.; Blake, D.M.; Noble, R.D.; Koval, C.A., 1994, Environmental Science and Technology, Vol. 28, No. (9), pp. 1661-1668. Kittrell, J. R., Quinlan, C. W., Shepanzyk, J. W., “Photocatalytic Destruction of Hexane Eliminates Emissions in Contact Lens Manufacture”, Emerging Solutions in VOC Air Toxics Control, Proc. Spec. Conf. (1996), pp. 389-398. Larson, S. A. and Falconer J. L., 1997, Catalysis Letters, Vol. 44, No. 9, pp. 57-65. Luo, Y. and Ollis, D. F., 1996, Journal of Catalysis, Vol. 163, No. 1, pp. 1-11. Magrini, K. A., Watt, A. S., Boyd, L. C., Wolfrum, E. J., Larson, S. A., Roth, C., and Glatzmaier, G. C., “Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams”, ASME Proceedings of the Renewable and Advanced Energy Systems for the 21st Century, pp. 81-90, Lahaina, Maui, April 1999. Magrini, K. A., Rabago, R., Larson, S. A., and Turchi, C. S., “Control of Gaseous Solvent Emissions using Photocatalytic Oxidation”, Proc. Annu. Meet. Air Waste Manage. Assoc., 89th, (1996) pp. 1-8. Mueller, J. H., (1988), “Spray Paint Booth VOC Control with a Regenerative Thermal Oxidizer”, Proceedings-APCA Annual Meeting 81st(5), Paper 88/84.3, 16 pp. Nimlos, M. R., Jacoby, W. A., Blake, D. M. and Milne,T. A., 1993, Environmental Science and Technology, vol. 27, p. 732. Peral, J. and Ollis, D.F., 1992, Journal of Catalysis, Vol. 136, pp. 554-565. Phillips, L.A.; Raupp, G.B., 1992, Journal of Molecular Catalysis, Vol. 7, pp. 297-311. Read, H. W., Fu, X., Clark, L. A., Anderson, M. A., Jarosch, T., 1996, “Field Trials of a TiO2 Pellet-Based Photocatalytic Reactor for Off-Gas Treatment at a Soil Vapor Extraction Well”, Journal of Soil Contamination, Vol. 5, No. 2, pp. 187-202. Rosansky, S. H., Gavaskar, A. R., Kim, B. C., Drescher, E., Cummings, C. A., and Ong, S. K., “Innovative Air Stripping and Catalytic Oxidation Techniques for Treatment of Chlorinated Solvent Plumes”, Therm. Technol., Int. Conf., Rem. Chlorinated Recalcitrant Compounds., 1st (1998) pp. 415-424. Sitkiewitz, S. and Heller, A., 1996, New Journal of Chemistry, Vol. 20, No. 2, pp. 233-241. Turchi, C.S., Wolfrum, E., Miller, R.A., 1994, Gas-Phase Photocatalytic Oxidation: Cost Comparison with Other Air Pollution Control Technologies, NREL/TP-471-7014. Golden, CO: National Renewable Energy Laboratory. Watt, A. S., Magrini, K. A., Carlson-Boyd, L. C., Wolfrum, E. J., Larson, S. A., Roth, C., and Glatzmaier, G. C., “A Pilot-Scale Demonstration of an Innovative Treatment for Vapor Emissions”, J. of Air and Waste Management, November, 1999. Yamazaki-Nishida, S., Fu, X., Anderson, M. A. and Hori, K., 1996, Journal of Photochemistry and Photobiology. A: Chemistry, Vol. 97, pp. 175-179.
23
Contamination Cleanup: Natural Attenuation and Advanced Remediation Technologies 23.1 NATURAL ATTENUATION OF CHLORINATED SOLVENTS IN GROUND WATER Hanadi S. Rifai Civil and Environmental Engineering, University of Houston, Houston, Texas, USA
Charles J. Newell Groundwater Services, Inc., Houston, Texas, USA
Todd H. Wiedemeier Parson Engineering Science, Denver, CO, USA
23.1.1 INTRODUCTION Chlorinated solvents were first produced some 100 years ago and came into common usage in the 1940’s. Chlorinated solvents are excellent degreasing agents and they are nearly nonflammable and non-corrosive. These properties have resulted in their widespread use in many industrial processes such as cleaning and degreasing rockets, electronics and clothing (used as dry-cleaning agents). Chlorinated solvent compounds and their natural degradation or progeny products have become some of the most prevalent organic contaminants found in the shallow groundwater of the United States. The most commonly used chlorinated solvents are perchloroethene (PCE), trichloroethene (TCE), 1,1,1-trichloroethane (TCA), and carbon tetrachloride (CT).1 Chlorinated solvents (CS) undergo the same natural attenuation processes as many other ground water contaminants such as advection, dispersion, sorption, volatilization and biodegradation. In addition, CS are subject to abiotic reactions such as hydrolysis and dehydrohalogenation and abiotic reduction reactions. While many of the physical and chemical reactions affecting chlorinated solvents have been extensively studied, their biodegradation is not as well understood as perhaps it is for petroleum hydrocarbons. Researchers are just beginning to understand the microbial degradation of chlorinated solvents with many degradation pathways remaining to be discovered. Unlike petroleum hydrocarbons, which can be oxidized by microorganisms under either aerobic or anaerobic condi-
1572
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
tions, most chlorinated solvents are degraded only under specific ranges of oxidation-reduction potential. For example, it is currently believed that PCE is biologically degraded through use as a primary growth substrate only under strongly reducing anaerobic conditions. This chapter is focused on the natural attenuation behavior of CS at the field scale. The first part of the chapter reviews many of the physical, chemical and abiotic natural attenuation processes that attenuate CS concentrations in ground water. Some of these processes have been described in more detail in previous chapters in the handbook and are therefore only reviewed in brief. In the second part of this chapter, we will review the biological processes that bring about the degradation of the most common chlorinated solvents, present conceptual models of chlorinated solvent plumes, and summarize data from field studies with chlorinated solvent contamination. 23.1.2 NATURAL ATTENUATION PROCESSES AFFECTING CHLORINATED SOLVENT PLUMES Many abiotic mechanisms affect the fate and transport of organic compounds dissolved in ground water. Physical processes include advection and dispersion while chemical processes include sorption, volatilization and hydrolysis. Advection transports chemicals along ground water flow paths and in general does not cause a reduction in contaminant mass or concentration. Dispersion or mixing effects, on the other hand, will reduce contaminant concentrations but will not cause a reduction in the total mass of chemicals in the aquifer. Sorption or partitioning between the aquifer matrix and the ground water, much like dispersion, will not cause a reduction in contaminant mass. Volatilization and hydrolysis both will result in lower concentrations of the contaminant in ground water. The majority of these processes, with the exception of hydrolysis and dehydrohalogenation chemical reactions, do not break down or destroy the contaminants in the subsurface. Chlorinated solvents are advected, dispersed, and sorbed in ground water systems. They also volatilize although their different components have varying degrees of volatility. Chlorinated solvents additionally hydrolyze and undergo other chemical reactions such as dehydrohalogenation or elimination and oxidation and reduction. These abiotic reactions, as will be seen later in the chapter, are typically not complete and often result in the formation of an intermediate that may be at least as toxic as the original contaminant. 23.1.2.1 Advection Advective transport is the transport of solutes by the bulk movement of ground water. Advection is the most important process driving dissolved contaminant migration in the subsurface and is given by: vx = −
K dH n e dL
where: Vx K ne dH/dL
seepage velocity [L/T] hydraulic conductivity [L/T] effective porosity [L3/L3] hydraulic gradient [L/L]
[23.1.1 ]
23.1 Natural attenuation of chlorinated solvents
1573
Figure 23.1.1. Breakthrough curve in one dimension showing plug flow with continuous source resulting from advection only and from the combined processes of advection and hydrodynamic dispersion. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons. Reprinted by permission of John Wiley & Sons.]
Typical velocities range between 10-7 and 103 ft/day2 with a median national average of 0.24 ft/day. The seepage velocity is a key parameter in natural attenuation studies since it can be used to estimate the time of travel of the contaminant front: t=
x vx
[23.1.2]
where: x t
travel distance (ft or m) time
Solute transport by advection alone yields a sharp solute concentration front as shown in Figure 23.1.1. In reality, the advancing front spreads out due to the processes of dispersion and diffusion as shown in Figure 23.1.1, and is retarded by sorption (Figure 23.1.2) and biodegradation.
Figure 23.1.2. Breakthrough curve in one dimension showing plug flow with continuous source resulting from advection only; from the combined processes of advection and hydrodynamic dispersion; and from the combined processes of advection, hydrodynamic dispersion, and sorption. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
23.1.2.2 Dispersion Hydrodynamic dispersion causes a contaminant plume to spread out from the main direction of ground water flow. Dispersion dilutes the concentrations of the contaminant, and introduces the contaminant into relatively pristine portions of the aquifer where it mixes with more electron acceptors crossgradient to the direction of ground-water flow. As a result of
1574
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
dispersion, the solute front travels at a rate that is faster than would be predicted based solely on the average linear velocity of the ground water. Figure 23.1.1 illustrates the effects of hydrodynamic dispersion on an advancing solute front. Mechanical dispersion is commonly represented by the relationship: Mechanical Dispersion = αxvx
[23.1.3]
where: vx αx
average linear ground-water velocity [L/T] dispersivity [L]
Dispersivity represents the spreading of a contaminant over a given length of flow and is characteristic of the porous medium through which the contaminant migrates. It is commonly accepted that dispersivity is scale-dependent, and that at a given scale, dispersivity may vary over three orders of magnitude3,4 as shown in Figure 23.1.3. Several approaches can be used to estimate longitudinal dispersivity, α x , at the field scale. One technique involves conducting a tracer test but this method is time consuming and costly. Another method commonly used in solute transport modeling is to start with a longitudinal Figure 23.1.3. Relationship between dispersivity and dispersivity of 0.1 times the plume scale. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell lengths.5-7 This assumes that dispersivity and J.T. Wilson, Natural Attenuation of Fuels and varies linearly with scale. Xu and Eckstein8 Chlorinated Solvents in the Subsurface. Copyright © proposed an alternative approach. They 1999 John Wiley & Sons, Inc. Reprinted by permission of evaluated the same data presented by John Wiley & Sons, Inc.] Gelhar et al.4 and, by using a weighted least-squares method, developed the following relationship for estimating dispersivity: α x = 0.83(log10 L p )
2. 414
[23.1.4]
where: αx Lp
longitudinal dispersivity [L] plume length [L]
Both relationships are shown on Figure 23.1.3. In addition to estimating longitudinal dispersivity, it may be necessary to estimate the transverse and vertical dispersivities (αT and αZ, respectively) for a given site. Commonly, αT is estimated as 0.1αx (based on data4), or as 0.33αx.9,10 Vertical dispersivity (αZ) may be estimated as 0.05αx,9 or as 0.025 to 0.1αx.10 23.1.2.3 Sorption Many organic contaminants, including chlorinated solvents, are removed from solution by sorption onto the aquifer matrix. Sorption of dissolved contamination onto the aquifer ma-
23.1 Natural attenuation of chlorinated solvents
1575
trix results in slowing (retardation) of the contaminant relative to the average advective ground-water flow velocity and a reduction in dissolved organic concentrations in ground water. Sorption can also influence the relative importance of volatilization and biodegradation.11 Figure 23.1.2 illustrates the effects of sorption on an advancing solute front. Sorption is a reversible reaction; at given solute concentrations, some portion of the solute is partitioning to the aquifer matrix and some portion is also desorbing, and reentering solution. Sorption can be determined from bench-scale experiments. These are typically performed by mixing water-contaminant solutions of various concentrations with aquifer materials containing various amounts of organic carbon and clay minerals. The solutions are then sealed with no headspace and left until equilibrium between the various phases is reached. The amount of contaminant left in solution is then measured. The results are commonly expressed in the form of a sorption isotherm or a plot of the concentration of chemical sorbed (µg/g) versus the concentration remaining in solution (µg/L). Sorption isotherms generally exhibit one of three characteristic shapes depending on the sorption mechanism. These isotherms are referred to as the Langmuir isotherm, the Freundlich isotherm, and the linear isotherm (a special case of the Freundlich isotherm). The reader is referred to ref. 1 for more details on sorption isotherms. Since sorption tends to slow the transport velocity of contaminants dissolved in ground water, the contaminant is said to be “retarded.” The coefficient of retardation, R, is defined as: R=
vx vc
[23.1.5]
where: R vx vc
coefficient of retardation average linear ground-water velocity parallel to ground-water flow average velocity of contaminant parallel to ground-water flow
The ratio vx/vc describes the relative velocity between the ground water and the dissolved contaminant. The coefficient of retardation for a dissolved contaminant (for saturated flow) assuming linear sorption is determined from the distribution coefficient using the relationship: R = 1+
ρ bK d N
[23.1.6]
where: R ρb Kd N
coefficient of retardation [dimensionless] bulk density of aquifer [M/L3] distribution coefficient [L3/M] porosity [L3/L3]
The bulk density, ρb, of a soil is the ratio of the soil mass to its field volume. In sandy soils, ρb can be as high as 1.81 g/cm3, whereas, in aggregated loams and clayey soils, ρb can be as low as 1.1 g/cm3. The distribution coefficient is a measure of the sorption/desorption potential and characterizes the tendency of an organic compound to be sorbed to the aquifer matrix. The
1576
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
higher the distribution coefficient, the greater the potential for sorption to the aquifer matrix. The distribution coefficient, Kd, is given by: Kd =
Ca Cl
[23.1.7]
where: Kd Ca Cl
distribution coefficient (slope of the sorption isotherm, mL/g). a sorbed concentration (mass contaminant/mass soil or µg/g) dissolved concentration (mass contaminant/volume solution or µg/mL)
Several researchers have found that if the distribution coefficient is normalized relative to the aquifer matrix total organic carbon content, much of the variation in observed Kd values between different soils is eliminated.12-19 Distribution coefficients normalized to total organic carbon content are expressed as Koc. The following equation gives the expression relating Kd to Koc: K oc =
Kd f oc
[23.1.8]
where: Koc Kd foc
soil sorption coefficient normalized for total organic carbon content distribution coefficient fraction total organic carbon (mg organic carbon/mg soil)
Table 23.1.1 presents calculated retardation factors for several LNAPL and DNAPL related chemicals as a function of the fraction of organic carbon content of the soil. It can be seen from Table 23.1.1 that R can vary over two orders of magnitude at a site depending on the chemical in question and the estimated value of porosity and soil bulk density. Table 23.1.1. Calculated retardation factors for several chlorinated solvent-related chemicals Compound
log(Koc)
Fraction of organic compound in aquifer (foc) 0.0001
0.001
0.01
0.1
Carbon tetrachloride
2.67
1.2
3.3
24.0
231.2
1,1,1-TCA
2.45
1.1
2.4
14.9
139.7
PCE
2.42
1.1
2.3
13.9
130.4
1,1- or 1,2-DCA
1.76
1.0
1.3
3.8
29.3
trans 1,2-DCE
1.42
1.0
1.1
2.3
13.9
cis 1,2-DCE
1.38
1.0
1.1
2.2
12.8
TCE
1.26
1.0
1.1
1.9
10.0
Chloroethane
1.25
1.0
1.1
1.9
9.8
Dichloromethane
1.23
1.0
1.1
1.8
9.4
Vinyl chloride
0.06
1.0
1.0
1.1
1.6
Notes: Units of foc: g naturally-occurring organic carbon per g dry soil. Assumed porosity and bulk density: 0.35 and 1.72, respectively. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
23.1 Natural attenuation of chlorinated solvents
1577
23.1.2.4 One-dimensional advection-dispersion equation with retardation In one dimension, the advection-dispersion equation is given by: R
∂C ∂ 2C ∂C = Dx −vx ∂t ∂x ∂x 2
[23.1.9]
where: vx R C Dx T x
average linear ground-water velocity [L/T] coefficient of retardation [dimensionless] contaminant concentration [M/L3] hydrodynamic dispersion [L2/T] time [T] distance along flow path [L]
23.1.2.5 Dilution (recharge) Ground water recharge can be defined as the entry into the saturated zone of water made available at the water-table surface.20 Recharge may therefore include precipitation that infiltrates through the vadose zone and water entering the ground-water system due to discharge from surface water bodies (i.e., streams and lakes). Recharge of a water table aquifer has two effects on the natural attenuation of a dissolved contaminant plume. Additional water entering the system due to infiltration of precipitation or from surface water will contribute to dilution of the plume, and the influx of relatively fresh, electron acceptor-charged water will alter geochemical processes and in some cases facilitate additional biodegradation. Wiedemeier et al.1 present the following relationship for estimating the amount of dilution caused by recharge: L RW VD CL = Co exp− WThV D
[23.1.10]
eliminating the width and rearranging, gives: RL CL = Co exp − Th V 2 ( D)
[23.1.11]
where: CL Co R W L Th VD
concentration at distance L from origin assuming complete mixing of recharge with groundwater (mg/L) concentration at origin or at distance L = 0 (mg/L) recharge mixing with groundwater (ft/yr) width of area where recharge is mixing with groundwater (ft) length of area where recharge is mixing with groundwater (ft) thickness of aquifer where groundwater flow is assumed to completely mix with recharge (ft) Darcy velocity of groundwater (ft/yr)
1578
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
23.1.2.6 Volatilization Volatilization causes contaminants to transfer from the dissolved phase to the gaseous phase. In general, factors affecting the volatilization of contaminants from ground water into soil gas include the contaminant concentration, the change in contaminant concentration with depth, the Henry’s Law constant and diffusion coefficient of the compound, mass transport coefficients for the contaminant in both water and soil gas, sorption, and the temperature of the water.21 The Henry’s Law constant of a chemical determines the tendency of a contaminant to volatilize from ground water into the soil gas. Henry’s Law states that the concentration of a contaminant in the gaseous phase is directly proportional to the compound’s concentration in the liquid phase and is a constant characteristic of the compound:11 Ca = HCl
[23.1.12]
where: H Ca Cl
Henry’s Law constant (atm m3/mol) concentration in air (atm) concentration in water (mol/m3)
Values of Henry’s Law constants for selected chlorinated solvents are given in Table 23.1.2. As indicated in the table, values of H for chlorinated compounds also vary over several orders of magnitude. Chlorinated solvents have low Henry’s Law constants, with the exception of vinyl chloride. Volatilization of chlorinated solvents compounds from ground water is a relatively slow process that generally can be neglected when modeling biodegradation. Table 23.1.2. Physical properties for chlorinated compounds
Constituent
CAS #
Molecular weight
log Koc (@20-25°C)
Diffusion coefficients in air
in water
MW, g/mol
Dair, cm2/s
Ref
Dwat, cm2/s
Ref
Partition log(L/kg)
Ref
Ref
Bromodichloromethane
75-27-4
163.8
22
2.98E-02
22
1.06E-05
22
1.85
22
Carbon tetrachloride
56-23-5
153.8
22
7.80E-02
22
8.80E-06
22
2.67
22
108-90-7
112.6
22
7.30E-02
22
8.70E-06
22
2.46
22
Chloroethane
75-00-3
64.52
22
1.50E-01
22
1.18E-05
22
1.25
22
Chloroform
67-66-3
119.4
22
1.04E-01
22
1.00E-05
22
1.93
22
Chloromethane
74-87-3
51
23
1.28E-01
22
1.68E-04
b
1.40
28
Chlorophenol, 2-
95-57-8
128.6
22
5.01E-02
22
9.46E-06
22
2.11
22
124-48-1 208.29
22
1.99E-02
22
1.03E-05
22
2.05
22
Chlorobenzene
Dibromochloromethane
Dichlorobenzene, (1,2)(-o) 95-50-1
147
22
6.90E-02
22
7.90E-06
22
3.32
22
Dichlorobenzene, (1,4)(-p) 106-46-7
147
22
6.90E-02
22
7.90E-06
22
3.33
22
120.92
22
5.20E-02
22
1.05E-05
22
2.12
22
Dichlorodifluoromethane
75-71-8
23.1 Natural attenuation of chlorinated solvents
Constituent
CAS #
Molecular weight
1579
in air MW, g/mol
log Koc (@20-25°C)
Diffusion coefficients
Ref
Dair, cm2/s
in water Ref
Dwat, cm2/s
Ref
Ref
Partition log(L/kg)
Dichloroethane, 1,1-
75-34-3
98.96
22
7.42E-02
22
1.05E-05
22
1.76
22
Dichloroethane, 1,2-
107-06-2
98.96
22
1.04E-01
22
9.90E-06
22
1.76
22
Dichloroethene, cis-1,2-
156-59-2
96.94
22
7.36E-02
22
1.13E-05
22
1.38
c
Dichloroethene, 1,2-trans
156-60-5
96.94
22
7.07E-02
22
1.19E-05
22
1.46
22
75-09-2
85
22
1.01E-01
22
1.17E-05
22
1.23
22
Tetrachloroethane, 1,1,2,2- 79-34-5
168
22
7.10E-02
22
7.90E-06
22
0.00
22
Dichloromethane
Tetrachloroethene
127-18-4 165.83
22
7.20E-02
22
8.20E-06
22
2.43
28
Trichlorobenzene, 1,2,4-
120-28-1
181.5
22
3.00E-02
22
8.23E-06
22
3.91
22
Trichoroethane, 1,1,1-
71-55-6
133.4
22
7.80E-02
22
8.80E-06
22
2.45
22
Trichloroethane, 1,1,2-
79-00-5
133.4
22
7.80E-02
22
8.80E-06
22
1.75
28
Trichloroethene
79-01-6
131.4
24
8.18E-02
a
1.05E-04
b
1.26
d
Trichlorofluoromethane
75-69-4
137.4
22
8.70E-02
22
9.70E-06
22
2.49
22
Vinyl chloride
75-01-4
62.5
22
1.06E-01
22
1.23E-05
22
0.39
26
a
Calculated diffusivity using the method of Fuller, Schettler, and Giddings [from Reference 25] Calculated diffusivity using the method of Hayduk and Laudie and the reference 25 c Calculated using Kenaga and Goring Kow/solubility regression equation from reference 25 and Kow data from reference 26, log (S, mg/L) = 0.922 log(Kow) + 4.184 d d Back calculated from solubility [see note c, based on Kow from reference 26 and method from reference 27, log(Koc) = 0.00028 + 0.938 log (Kow)] [From RBCA Chemical Database. Copyright © 1995-1997 Groundwater Services, Inc. (GSI). Reprinted with permission.] b
23.1.2.7 Hydrolysis and dehydrohalogenation Hydrolysis and dehydrohalogenation reactions are the most thoroughly studied abiotic attenuation mechanisms. In general, the rates of these reactions are often quite slow within the range of normal ground-water temperatures, with half-lives of days to centuries.29,30 Hydrolysis is a substitution reaction in which a compound reacts with water, and a halogen substituent is replaced with a hydroxyl (OH-) group resulting in the formation of alcohols and alkenes after:31,32 RX + HOH → ROH + HX
[23.1.13]
H3C-CH2X → H2C=CH2 + HX
[23.1.14]
The likelihood that a halogenated solvent will undergo hydrolysis depends in part on the number of halogen substituents. More halogen substituents on a compound will decrease the chance for hydrolysis reactions to occur,29 and will therefore decrease the rate of the reaction. In addition, bromine substituents are more susceptible to hydrolysis than chlo-
1580
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
rine substituents;29 for example, 1,2-dibromoethane is subject to significant hydrolysis reactions under natural conditions. McCarty33 lists TCA (1,1,1-trichloroethane) as the only major chlorinated solvent that can be transformed chemically through hydrolysis (as well as elimination) leading to the formation of 1,1-DCE (l,1-dichlorethene) and acetic acid. Locations of the halogen substituent on the carbon chain may also have some effect on the rate of reaction. The rate also may increase with increasing pH; however, a rate dependence upon pH is typically not observed below a pH of 11.34,35 Rates of hydrolysis may also be increased by the presence of clays, which can act as catalysts.29 Other factors that impact the level of hydrolysis include dissolved organic matter, and dissolved metal ions. Hydrolysis rates can generally be described using first-order kinetics, particularly in solutions in which water is the dominant nucleophile.29 A listing of half-lives for abiotic hydrolysis and dehydrohalogenation of some chlorinated solvents is presented in Table 23.1.3. Note that no distinctions are made in the table as to which mechanism is operating; this is consistent with the references from which the table has been derived.29,36 Table 23.1.3. Approximate half-lives of abiotic hydrolysis and dehydrohalogenation reactions involving chlorinated solvents Compound
Half-Life (years)
Chloromethane
no data
Dichloromethane
70434
Chloroform
3500,34 180037
Carbon tetrachloride
4137
Chloroethane
0.1229
ethanol
37
1,1-Dichloroethane
61
1,2-Dichloroethane
7237
1,1,1-Trichloroethane
1.7,34 1.1,37 2.538
1,1,2-Trichloroethane
Products
37
34
140, 170 37
1,1-DCE
47, 380
1,1,2,2-Tetrachloroethane
29
34
1,1,1,2-Tetrachloroethane
37
0.3, 0.4, 0.8 40
acetic acid, 1,1-DCE
TCE 34
1,1,2-TCA, TCE
37
Tetrachloroethene
0.7, * 1.3E+06
Trichloroethene
0.7,40* 1.3E+0637
1,1-Dichloroethene
1.2E+1037
1,2-Dichloroethene
2.1E+1037
*Butler and Barker36 indicate that these values may reflect experimental difficulties and that the longer half-life [as calculated by Jeffers et al.37] should be used. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
One common chlorinated solvent for which abiotic transformations have been well studied is 1,1,1-TCA. 1,1,1-TCA may be abiotically transformed to acetic acid through a series of substitution reactions, including hydrolysis. In addition, 1,1,1-TCA may be reductively dehalogenated to form 1,1-DCA and then chloroethane (CA), which is then hydrolyzed to ethanol38 or dehydrohalogenated to vinyl chloride.37
23.1 Natural attenuation of chlorinated solvents
1581
Dehydrohalogenation is an elimination reaction involving halogenated alkanes in which a halogen is removed from one carbon atom, followed by the subsequent removal of a hydrogen atom from an adjacent carbon atom. In this two-step reaction, an alkene is produced. Contrary to the patterns observed for hydrolysis, the likelihood that dehydrohalogenation will occur increases with the number of halogen substituents. It has been suggested that under normal environmental conditions, monohalogenated aliphatics apparently do not undergo dehydrohalogenation, and these reactions are apparently not likely to occur.29,41 However, Jeffers et al.37 report on the dehydrohalogenation of CA to VC. Polychlorinated alkanes have been observed to undergo dehydrohalogenation under normal conditions and extremely basic conditions.29 As with hydrolysis, bromine substituents are more reactive with respect to dehydrohalogenation. Dehydrohalogenation rates may also be approximated using pseudo-first-order kinetics. The rates will not only depend upon the number and types of halogen substituent, but also on the hydroxide ion concentration. Under normal pH conditions (i.e., near a pH of 7), interaction with water (acting as a weak base) may become more important.29 Transformation rates for dehydrohalogenation reactions are presented in Table 23.1.3. The organic compound 1,1,1-TCA is also known to undergo dehydrohalogenation.38 In this case, TCA is transformed to 1,1-DCE, which is then reductively dehalogenated to VC. The VC is then either reductively dehalogenated to ethene or consumed as a substrate in an aerobic reaction and converted to CO2. In a laboratory study, Vogel and McCarty38 reported that the abiotic conversion of 1,1,1-TCA to 1,1-DCE has a rate constant of about 0.04 year-1. Jeffers et a1.37 reported that the tetrachloroethanes and pentachloroethanes degrade to TCE and PCE via dehydrohalogenation, respectively. Jeffers et al.37 also report that CA may degrade to VC. 23.1.2.8 Reduction reactions Two abiotic reductive dechlorination reactions that may operate in the subsurface are hydrogenolysis and dihaloelimination. Hydrogenolysis is the simple replacement of a chlorine (or another halogen) by a hydrogen, while dihaloelimination is the removal of two chlorines (or other halogens) accompanied by the formation of a double carbon-carbon bond. While these reactions are thermodynamically possible under reducing conditions, they often do not take place in the absence of biological activity.36,42-45 In general, microbes may produce reductants that facilitate such reactions in conjunction with minerals in the aquifer matrix. Moreover, the reducing conditions necessary to produce such reactions are most often created as a result of microbial activity. It is therefore not clear if some of these reactions are truly abiotic, or if because of their reliance on microbial activity to produce reducing conditions or reactants, they should be considered to be a form of cometabolism. In some cases, truly abiotic reductive dechlorination has been observed;46,47 however, the conditions that favor such reactions may not occur naturally. 23.1.3 BIODEGRADATION OF CHLORINATED SOLVENTS The biodegradation of organic chemicals can be grouped into two broad categories: 1) use of the organic compound as a primary growth substrate, and 2) cometabolism. The use of chlorinated solvents as a primary growth substrate is probably the most important biological mechanism affecting them in the subsurface. Some chlorinated solvents are used as electron donors and some are used as electron acceptors when serving as pri-
1582
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
mary growth substrates (meaning the mediating organism obtains energy for growth). When used as an electron donor, the chlorinated solvent is oxidized. Oxidation reactions can be aerobic or anaerobic. Conversely, when used as an electron acceptor, the chlorinated solvent is reduced via a reductive dechlorination process called halorespiration. It is important to note that not all chlorinated solvents can be degraded via all of these reactions. In fact, vinyl chloride is the only chlorinated solvent known to degrade via all of these pathways. Chlorinated solvents can also be degraded via cometabolic pathways. During cometabolism, microorganisms gain carbon and energy for growth from metabolism of a primary substrate, and chlorinated solvents are degraded fortuitously by enzymes present in the metabolic pathway. Cometabolism reactions can be either oxidation or reduction reactions (under aerobic or anaerobic conditions), however based on data from numerous field sites, it does not appear that cometabolic oxidation will be a significant process in plumes of chlorinated solvents. Anaerobic reductive dechlorination can also occur via cometabolism. The process of cometabolic reductive dechlorination, however, is “sufficiently slow and incomplete that a successful natural attenuation strategy typically cannot completely rely upon it”.48 The types of biodegradation reactions that have been observed for different chlorinated solvents are presented in Table 23.1.4. The remainder of this section will focus on describing the various mechanisms shown in Table 23.1.4.
PCE TCE DCE Vinyl chloride 1,1,1-TCA 1,2-DCA Chloroethane Carbon tetrachloride Chloroform Dichloromethane
X X X X X X
X X
X X X X X X
X
X X
X X
X X
X X X
Anaerobic cometabolism
Aerobic cometabolism
Direct anaerobic oxidation
Halorespiration
Compound
Direct aerobic oxidation
Table 23.1.4. Biological degradation processes for selected chlorinated solvents
X X X X X X X X
[From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
23.1.3.1 Halorespiration or reductive dechlorination using hydrogen Reductive dechlorination is a reaction in which a chlorinated solvent acts as an electron acceptor and a chlorine atom on the molecule is replaced with a hydrogen atom. This results in the reduction of the chlorinated solvent. When this reaction is biological, and the organism is utilizing the substrate for energy and growth, the reaction is termed halorespiration. Only
23.1 Natural attenuation of chlorinated solvents
1583
Figure 23.1.4. Abiotic and biological transformation pathways for selected chlorinated solvents. (From reference 1 after references 29, 53). [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
recently have researchers demonstrated the existence of halorespiration.49 Prior to this research, reductive dechlorination was thought to be strictly a cometabolic process. During halorespiration, hydrogen is used directly as an electron donor. The generalized reaction is given by: H2 + C-Cl → C-H + H+ + Cl-
[23.1.15]
where C-Cl represents a carbon-chloride bond in a chlorinated solvent. In this reaction, H2 is the electron donor which is oxidized and the chlorinated solvent is the electron acceptor which is reduced. Although a few other electron donors (also fermentation products) besides hydrogen have been identified, hydrogen appears to be the most important electron donor for halorespiration. Only in the last four years have researchers begun to fully recognize the role of hydrogen as the electron donor in the reductive dechlorination of PCE and TCE.48,50-52 The hydrogen is produced in the subsurface by the fermentation of a wide variety of organic compounds including petroleum hydrocarbons and natural organic carbon. Because of its importance in the microbial metabolism of the halorespirators, the relative supply of hydrogen precursors compared to the amount of chlorinated solvent that must be degraded is an important consideration when evaluating natural attenuation. In general, reductive dechlorination of the ethenes occurs by sequential dechlorination from PCE to TCE to DCE to VC and finally to ethene. A summary of key biotic and abiotic reactions for the chlorinated ethenes, ethanes, and methanes first developed by Voge129 is shown in Figure 23.1.4. For halorespiration to occur, Wiedemeier et al.1 conclude that the following conditions must exist: “1) the subsurface environmental must be anaerobic and have a low oxidation-reduction potential (based on thermodynamic considerations, reductive dechlorination reactions will occur only after both oxygen and nitrate have been depleted from the aquifer);
1584
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
2) chlorinated solvents amenable to halorespiration must be present; and 3) there must be an adequate supply of fermentation substrates for production of dissolved hydrogen.” Fermentation, the process that generates hydrogen, is a balanced oxidation-reduction reaction, in which different portions of a single substrate are oxidized and reduced, yielding energy. Fermentation yields substantially less energy per unit of substrate compared to oxidation reactions which utilize an external electron acceptor; thus, fermentation generally occurs when these external electron acceptors are not available. Bacterial fermentation which can be important in anaerobic aquifers includes primary and secondary fermentation. Primary fermentation refers to the fermentation of primary substrates such as sugars, amino acids, and lipids yields acetate, formate, CO2 and H2, but also yields ethanol, lactate, succinate, propionate, and butyrate. Secondary fermentation, on the other hand, refers to the fermentation of primary fermentation products such as ethanol, lactate, succinate, propionate, and butyrate yielding acetate, formate, H2, and CO2. Bacteria which carry out these reactions are called obligate proton reducers because the reactions must produce hydrogen in order to balance the oxidation of the carbon substrates. These secondary fermentation reactions are energetically favorable only if hydrogen concentrations are very low (10-2 to 10-4 atm or 8,000 nM to 80 nM dissolved hydrogen, depending on the fermentation substrate). Thus, these fermentation reactions occur only when the produced hydrogen is utilized by other bacteria, such as methanogens which convert H2 and CO2 into CH4 and H2O. In the absence of external electron acceptors, the hydrogen produced by fermentation will be utilized by methanogens (methane producing bacteria). In this case, the ultimate end products of anaerobic metabolism of carbon substrates will be CH4 (the most reduced form of carbon) and CO2 (the most oxidized form of carbon). Methanogens will carry out the last step in this metabolism, the conversion of H2 and CO2 into CH4. However, in the presence of external electron acceptors (halogenated organics, nitrate, sulfate, etc.), other products will be formed.1 There are a number of compounds besides the ones listed above that can be fermented to produce hydrogen. Sewell and Gibson54 noted that petroleum hydrocarbons support reductive dechlorination. In this case, the reductive dechlorination is driven by the fermentation of biodegradable compounds such as the BTEX compounds in fuels. Metabolism of BTEX compounds to produce hydrogen likely requires the involvement of several strains of bacteria. Although the BTEX compounds are common fermentation substrates at chlorinated solvent sites, there are many other hydrocarbon substrates which are naturally fermented at sites and result in the generation of hydrogen such as acetone, sugars, and fatty acids from landfill leachate. As hydrogen is produced by fermentative organisms, it is rapidly consumed by other bacteria. The utilization of H2 by non-fermentors is known as interspecies hydrogen transfer and is required for fermentation reactions to proceed. Although H2 is a waste product of fermentation, it is a highly reduced molecule which makes it an excellent, high-energy electron donor. Both organisms involved in interspecies hydrogen transfer benefit from the process. The hydrogen-utilizing bacteria gain a high energy electron donor, and, for the fermentors, the removal of hydrogen allows additional fermentation to remain energetically favorable. A wide variety of bacteria can utilize hydrogen as an electron donor: denitrifiers, iron reducers, sulfate reducers, methanogens, and halorespirators. Thus, the production of hydrogen through fermentation does not, by itself, guarantee that hydrogen will be available for halorespiration. For dechlorination to occur, halorespirators must successfully compete
23.1 Natural attenuation of chlorinated solvents
1585
against the other hydrogen utilizers for the available hydrogen. Smatlak et al.51 suggest that the competition for hydrogen is controlled primarily by the Monod half-saturation constant Ks(H2), the concentration at which a specific strain of bacteria can utilize hydrogen at half the maximum utilization rate. Ballapragada et al.,52 however, provide a more detailed discussion of halorespiration kinetics and point out that competition for hydrogen also depends on additional factors including the bacterial growth rate and maximum hydrogen utilization rate. Smatlak et al.51 have suggested that the steady-state concentration of hydrogen will be controlled by the rate of hydrogen production from fermentation. Both laboratory results and field observations have suggested, however, that the steady-state concentration of hydrogen is controlled by the type of bacteria utilizing the hydrogen and is almost completely independent of the rate of hydrogen production.52,55-57 Under nitrate reducing conditions, steady-state H2 concentrations were <0.05 nM; under iron reducing conditions, they were 0.2 to 0.8 nM; under sulfate reducing conditions, they were 1-4 nM, and under methanogenic conditions, they were 5-14 nM.56,58,59 Finally, Carr and Hughes57 show that dechlorination in a laboratory column is not impacted by competition for electron donor at high hydrogen concentrations. Thus, it is clear that an increased rate of hydrogen production will result in increased halorespiration without affecting the competition between bacteria for the available hydrogen. 23.1.3.1.1 Stoichiometry of reductive dechlorination Under anaerobic conditions, Gossett and Zinder48 showed that the reductive dehalogenation of the chlorinated ethenes occurs as a series of consecutive irreversible reactions mediated by the addition of 1 mole of hydrogen gas for every mole of chloride removed. Thus, the theoretical minimum hydrogen requirement for dechlorination can be calculated on a mass basis as shown below:1 1 mg H2 will dechlorinate 21 mg of PCE to ethene 1 mg H2 will dechlorinate 22 mg of TCE to ethene 1 mg H2 will dechlorinate 24 mg of DCE to ethene 1 mg H2 will dechlorinate 31 mg of VC to ethene Complete fermentation of BTEX compounds is expected to yield 0.25 to 0.4 mg H2 per mg BTEX. Therefore, for each mg of BTEX consumed, 4.5 to 7 mg of chloride could be released during reductive dechlorination. However, the utilization of hydrogen for dechlorination will never be completely efficient because of the competition for hydrogen in the subsurface discussed previously. One rule of thumb that has been proposed is the following: for reductive dechlorination to completely degrade a plume of dissolved chlorinated solvents, organic substrate concentrations greater than 25 to 100+ times that of the chlorinated solvent are required.60 23.1.3.1.2 Chlorinated solvents that are amenable to halorespiration As shown in Table 23.1.4, all of the chlorinated ethenes (PCE, TCE, DCE, VC) and some of the chlorinated ethanes (TCA, 1,2-DCA) can be degraded via halorespiration; however, dichloromethane has not yet been shown to be degraded by this process. The oxidation state of a chlorinated solvent affects both the energy released by halorespiration and the rate at which the reaction occurs. In general, the more oxidized a compound is (more chlorine atoms on the organic molecule) the more amenable it is for reduction by halorespiration.
1586
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
As with the ethenes, chlorinated ethanes will also undergo halorespiration. Dechlorination of 1,1,1-TCA has been described by Vogel and McCarty38 and Cox et al.,61 but understanding this pathway is complicated by the rapid hydrolysis reactions (e.g., half-life is 0.5-2.5 yrs) that can affect TCA.30 Finally, halorespiration has been observed with highly chlorinated benzenes such as hexachlorobenzene, pentachlorobenzene, tetrachlorobenzene, and trichlorobenzene.62-64 As discussed by Suflita and Townsend,64 halorespiration of aromatic compounds has been observed in a variety of anaerobic habitats, including aquifer materials, marine and freshwater sediments, sewage sludges, and soil samples. However, isolation of specific microbes capable of these reactions has been difficult. 23.1.3.2 Oxidation of chlorinated solvents In contrast to halorespiration, direct oxidation of some chlorinated solvents can occur biologically in groundwater systems. In this case, the chlorinated compound serves as the electron donor, and oxygen, sulfate, ferric iron or other compounds serve as the electron acceptor. 23.1.3.2.1 Direct aerobic oxidation of chlorinated compounds Under direct aerobic oxidation conditions, the facilitating microorganism uses oxygen as an electron acceptor and obtains energy and organic carbon from the degradation of the chlorinated solvent. In general, the more-chlorinated aliphatic chlorinated solvents (e.g., PCE, TCE, and TCA) have not been shown to be susceptible to aerobic oxidation, while many of the progeny products (e.g., vinyl chloride, 1,2-DCA, and perhaps the isomers of DCE) are degraded via direct aerobic oxidation. Hartmans et al.65 and Hartmans and de Bont66 show that vinyl chloride can be used as a primary substrate under aerobic conditions, with vinyl chloride being directly mineralized to carbon dioxide and water. Direct vinyl chloride oxidation has also been reported by Davis and Carpenter,67 McCarty and Semprini,53 and Bradley and Chapelle.68 Aerobic oxidation is rapid relative to reductive dechlorination of dichloroethene and vinyl chloride. Although direct DCE oxidation has not been verified, a recent study has suggested that DCE isomers may be used as primary substrates.68 Of the chlorinated ethanes, only 1,2-dichloroethane has been shown to be aerobically oxidized. Stucki et al.69 and Janssen et al.70 show that 1,2-DCA can be used as a primary substrate under aerobic conditions. In this case, the bacteria transform 1,2-DCA to chloroethanol, which is then mineralized to carbon dioxide. McCarty and Semprini53 describe investigations in which 1,2-dichloroethane (DCA) was shown to serve as primary substrates under aerobic conditions. Chlorobenzene and polychlorinated benzenes (up to and including tetrachlorobenzene) have been shown to biodegrade under aerobic conditions. Several studies have shown that bacteria are able to utilize chlorobenzene,71 1,4-DCB,71-73 1,3-DCB,74 1,2-DCB,75 1,2,4-TCB,76,77 and 1,2,4,5-TeCB,77 as primary growth substrates in aerobic systems. Nishino et al.78 note that aerobic bacteria able to grow on chlorobenzene have been detected at a variety of chlorobenzene-contaminated sites but not at uncontaminated sites. Spain79 suggests that this provides strong evidence that the bacteria are selected for their ability to derive carbon and energy from chlorobenzene degradation in situ. The pathways for all of these reactions are similar, bearing resemblance to benzene degradation pathways.79,80
23.1 Natural attenuation of chlorinated solvents
1587
23.1.3.2.2 Aerobic cometabolism of chlorinated compounds It has been reported that under aerobic conditions chlorinated ethenes, with the exception of PCE, are susceptible to cometabolic oxidation.30,53,81,82 Vogel30 further elaborates that the oxidation rate increases as the degree of chlorination decreases. Aerobic cometabolism of ethenes may be characterized by a loss of contaminant mass, the presence of intermediate degradation products (e.g., chlorinated oxides, aldehydes, ethanols, and epoxides), and the presence of other products such as chloride, carbon dioxide, carbon monoxide, and a variety of organic acids.53,83 Cometabolism requires the presence of a suitable primary substrate such as toluene, phenol, or methane. For cometabolism to be effective, the primary substrate must be present at higher concentrations than the chlorinated compound, and the system must be aerobic. Because the introduction of high concentrations of oxidizable organic matter into an aquifer quickly drives the groundwater anaerobic, aerobic cometabolism typically must be engineered. 23.1.3.2.3 Anaerobic oxidation of chlorinated compounds Anaerobic oxidation occurs when anaerobic bacteria use the chlorinated solvent as an electron donor by utilizing an available electron acceptor such as ferric iron (Fe(III)). Bradley and Chapelle84 show that vinyl chloride can be oxidized to carbon dioxide and water via Fe(III) reduction. In microcosms amended with Fe(III)-EDTA, reduction of vinyl chloride concentrations closely matched the production of carbon dioxide. Slight mineralization was also noted in unamended microcosms. The rate of this reaction apparently depends on the bioavailability of Fe(III). In a subsequent paper, Bradley and Chapelle85 reported “significant” anaerobic mineralization of both DCE and VC in microcosms containing creek bed sediments. The sediments were taken from a stream where groundwater containing chlorinated ethenes conFigure 23.1.5. Reaction sequence and relative rates for tinually discharges. Anaerobic halorespiration of chlorinated ethenes, with other reactions shown. mineralization was observed in [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the both methanogenic and Fe(III) reSubsurface, reaction rates description from reference 88. Copyright ducing conditions. © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
1588
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
23.1.4 BIODEGRADATION RATES FOR CHLORINATED SOLVENTS Overall, dechlorination is more rapid for highly chlorinated compounds than for compounds that are less chlorinated.38,86,87 Figure 23.1.5 qualitatively shows the reaction rate and required conditions for halorespiration of PCE to ethene. PCE (four chlorines) degrades the fastest under all anaerobic environments, while VC (a single chlorine) will degrade only under sulfate-reducing and methanogenic conditions, with a relatively slow reaction rate. At many chlorinated ethene sites, concentrations of cis-1,2-DCE are often higher than any of the parent chlorinated ethene compounds. The reason for the accumulation of 1,2-DCE may be due to either slower rates of DCE halorespiration, or the prevalence of organisms that reduce PCE as far as cis-1,2-DCE over ones that can reduce PCE all the way to ethene.48 Although many researchers have commented that reductive dechlorination will result in the accumulation of VC (e.g., see 84, 89), at many field sites VC accumulation is much lower than cis-1,2-DCE. This may occur because the vinyl chloride in many chlorinated solvent plumes can migrate to zones that can support direct oxidation of VC oxidation, either aerobically and/or anaerobically. Suarez and Rifai90 analyzed data from 138 studies (field and laboratory) to estimate biodegradation coefficients for chlorinated compounds. Suarez and Rifai90 found a total of thirteen studies that reported Michaelis-Menten kinetics, 28 studies that reported zero-order rates, and 97 studies that reported first-order constants. 23.1.4.1 Michaelis-Menten rates The data in Table 23.1.5 present the Michaelis Menten kinetic data from Suarez and Rifai.90 Half-saturation constants varied from 0.6 mg/L to 29.5 mg/L for TCE and from 0.17 mg/L to 28 mg/L for DCE. Maximum specific degradation rates were within the ranges 0.038-478.59 mgcompound/mgprotein-day for TCE, and 0-11,115 mgcompound/mgprotein-day for DCE.
1,1,1-TCA
Continuous reactor
Aerobic
Methylosinus trichosporium OB3b
Growth reactor
AerobicCometabolism (methane)
Mixed methanotrophic culture
Continuous reactor
Aerobic
Continuous reactor
Aerobic
Ref
Initial concentration, So, mg/L
Max. spec. degradation rate, µ max/Y, mg/mg-day
Yield, Y, mg/mg
Half-saturation, Ks, mg/L
µ max, day-1
Culture
Redox environment
Type of study
Compound
Table 23.1.5. Michaelis-Menten parameters for chlorinated solvents
28.46
4.60
>93.10
92
0.43
0-11115
0.01
93
Methylosinus trichosporium OB3b
0.48
0.84
1.94-2.91
92
Methylosinus trichosporium OB3b
7.62
9.26
4.95-6.93
92
1.37
1,1-DCE
1,2-DCA
cis1,2-DCE
Growth reactor
AerobicCometabolism (phenol)
Filamentous phenoloxidizers
Continuous reactor
Aerobic
Methylosinus trichosporium OB3b
2.91
TCE
Continuous reactor
Aerobic
Methylosinus trichosporium OB3b
Growth reactor
AerobicCometabolism (methane)
Mixed methanotrophic culture
0.68
12.6025.20
92
52
14.34
46.19
8.72-14.74
92
0.17
0.00-0.44
4.70
93
12.00
Biofilm reactor
52
0.00 Methanogenic consortium
Ref
Initial concentration, So, mg/L
Max. spec. degradation rate, µ max/Y, mg/mg-day 25.40
94
28.00
Methanogenic fluidized bed Anaerobic reactor PCE
Yield, Y, mg/mg
0.27-1.50
Methanogenic fluidized bed Anaerobic reactor
trans1,2-DCE
Half-saturation, Ks, mg/L
1589
µ max, day-1
Culture
Redox environment
Type of study
Compound
23.1 Natural attenuation of chlorinated solvents
0.99
95
Fed-batch reactor
Anaerobic
Growth reactor
Aerobic
0.37
0.53
14.70
97
Growth reactor
AerobicCometabolism (formate)
8.20
7.60
10.10
97
Growth reactor
AerobicCometabolism (methane)
0.13
0.00-1.13
1.00
93
Mixed methanotrophic culture
Methanogenic fluidized bed Anaerobic reactor Growth reactor
AerobicCometabolism (phenol)
0.47
1.07
96
19.00
Filamentous phenoloxidizers
52
0.10-0.25
94
TCE
Ref
Initial concentration, So, mg/L
Max. spec. degradation rate, µ max/Y, mg/mg-day
Yield, Y, mg/mg
Half-saturation, Ks, mg/L
µ max, day-1
Microcosm
AerobicCometabolism (toluene)
0.771.65
0.52
1.50
0.66
98
Microcosm
AerobicCometabolism (phenol)
0.881.43
0.40
3.00
0.66
98
0.04
3.00
99
AerobicGrowth reacCometabolism tor (propane)
Propaneoxidizing culture
0.60
Batch
Methylomonas methanica 68-1
29.48
0.10
438.59
100
Methylosinus trichosporium OB3b
16.51
0.08
187.70
100
19.00
Aerobic
Batch
Vinyl chloride
Culture
Redox environment
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
Type of study
Compound
1590
Continuous reactor
Aerobic
Methylosinus trichosporium OB3b
Continuous reactor
AerobicCometabolism (methane)
Mixed methanotrophic culture
Methanogenic fluidized bed Anaerobic reactor SmallColumn Microcosm
AerobicCometabolism (methane)
54.71
9.17-13.10
0.01
92
101
23.00
52
1.003.50
1.00-17.00
102
[From M.P. Suarez and H.S. Rifai, Bioremediation Journal, 3, 337-362. Copyright © 1999 Battelle Memorial Institute. Reprinted with permission.]
23.1.4.2 Zero-order rates A summary of more than 40 studies reporting zero-order rates is included in Table 23.1.6. The reported zero-order rates ranged from 0 to 19.8 mg/L/day with mean values for anaerobic rates of 0.04, 2.14, 1.80, 1.74, and 0.11 mg/L/day for carbon tetrachloride, DCE, PCE, TCE, and vinyl chloride, respectively. TCE appeared to be reductively dechlorinated at the fastest rate coefficient, with a median equal to 0.76 mg/L/day. In contrast, vinyl chloride exhibited the slowest rate coefficient of reductive dechlorination with a median value of 0.01 mg/L/day.
23.1 Natural attenuation of chlorinated solvents
1591
Table 23.1.6 Summary of zero-order decay rates for reductive dechlorination obtained from laboratory studies (mg/L-day). CCl4
cis -12-DCE
DCE (all other isomers)
PCE
TCE
Vinyl chloride
8
18
8
29
7
9
0.022
0.013
0.009
0.013
0.314
0.002
25 percentile
0.024
0.183
0.023
0.288
0.511
0.006
median
Number of ratesa minimum th
0.029
0.511
0.250
0.577
0.760
0.011
th
0.042
1.318
1.385
1.040
1.297
0.075
th
90 percentile
0.049
3.348
2.021
2.801
3.798
0.379
maximum
0.054
16.958
3.470
19.800
7.490
0.495
mean
0.034
1.854
0.850
1.863
1.740
0.107
0.012
3.939
1.213
4.162
2.567
0.184
0.000
15.513
1.471
17.323
6.590
0.034
75 percentile
standard deviation geometric mean
b
a All the zero-order rates provided were calculated by the authors of the respective studies. bTo calculate the geometric mean, values equal to zero were included as 10-10. [From M.P. Suarez and H.S. Rifai, Bioremediation Journal, 3, 337-362. Copyright © 1999 Battelle Memorial Institute. Reprinted with permission]
23.1.4.3 First-order rate constants Table 23.1.7 summarizes the first order decay coefficients from both field and laboratory studies. As can be seen in Table 23.1.7, first-order rate constants for chlorinated solvents varied from 0 to 1.03 day-1 in the 90% of the cases, with mean values equal to 0.11, 0.02, 0.14, 0.05, 0.26, 0.17, and 0.23 day-1 for carbon tetrachloride, DCA, DCE, PCE, TCA, TCE, and vinyl chloride, respectively. The range minimum-90th percentile for aerobic rates was 0-1.35 day-1 while for anaerobic rates it was 0-1.11 day-1. Field rates from aerobic/anaerobic studies ranged from 0 to 1.96 day-1. The compound that showed the highest mean value under aerobic conditions was vinyl chloride (1.73 day-1), while TCA exhibited the highest mean anaerobic rate coefficient (0.35 day-1). range reported rates
geometric meanc
90th percentile
standard deviation
mean
Number of rates Number of reported rates Number of calculated ratesb
Table 23.1.7. Summary of first-order decay rates for chlorinated solvents (day-1).
Carbon tetrachloride All studies
13
0
13
Aerobic oxidation In situ & laboratory In situ studiesa Laboratory
1 1
0 0
1 1
0.108
0.134 0.216 0.054
0.0037-0.49
Aerobic cometabolism Field & laboratory In situ studiesa Laboratory
1 1
0 0
1 1
Overall aerobic
2
0
2
0.019
11 7 4
0 0 0
11 7 4
0.124 0.141 0.093
range reported rates
geometric meanc
90th percentile
standard deviation
mean
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier Number of rates Number of reported rates Number of calculated ratesb
1592
0.016-0.022
Aerobic/anaerobic (field studies) Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory
0.140 0.230 0.065 0.174 0.334 0.060 0.075
0.004-0.490 0.004-0.490 0.023-0.160
0.036 0.046 0.001
0-0.131
Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory DCA (all isomers) All studies Aerobic oxidation In situ & laboratory In situ studiesa Laboratory Aerobic cometabolism Field & laboratory In situ studiesa Laboratory Overall aerobic
25
16
9
0.017
2
2
0
0.000
2
2
0
0.000
5
0
5
0.067
0.056 0.128 0.046
0.014-0.131
5
0
5
0.067
0.056 0.128 0.046
0.014-0.131
7
2
5
0.048
0.056 0.126 0.000
0-0.131
18 16 2
14 14 0
4 2 2
0.005 0.002 0.036
0.012 0.016 0.001 0.003 0.004 0.001 0.035
0-0.044 0-0.011 0.028-0.044
0.395 0.257 0.004
0-1.960
Aerobic/anaerobic (field studies) Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory cis-1,2-DCE All studies
34
24
10
0.004
geometric meanc
5 3 2
2 2 0
3 1 2
0.476 0.885 0.187
0.787 1.680 0.476 0.843 1.820 0.885 0.250 0.399 0.187
0.081-1.96 0.281-1.96 0.081-0.434
4
4
0
0.000
0.003 0.006 0.000
0-0.008
25 17 8
18 13 5
7 4 3
0.004 0.002 0.014
0.048 0.069 0.004 0.031 0.013 0.002 0.069 0.117 0.014
0-0.200 0-0.130 0.001-0.200
range reported rates
standard deviation
90th percentile
1593
mean
Number of rates Number of reported rates Number of calculated ratesb
23.1 Natural attenuation of chlorinated solvents
Aerobic oxidation In situ & laboratory In situ studiesa Laboratory Aerobic cometabolism Field & laboratory In situ studiesa Laboratory Overall aerobic Aerobic/anaerobic (field studies) Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory DCE (all other isomers) All studies
27
14
13
0.149
0.302 0.666 0.003
0-1.150
8 4 4
2 0 2
6 4 2
0.458 0.720 0.196
0.416 0.845 0.002 0.316 1.012 0.670 0.347 0.521 0.000
0-1.150 0.390-1.150 0-0.714
19 16 3
12 12 0
7 4 3
0.019 0.003 0.101
0.061 0.012 0.004 0.001 0.005 0.003 0.147 0.220 0.039
0.001-0.270 0.001-0.006 0.010-0.270
Aerobic oxidation In situ & laboratory In situ studiesa Laboratory Aerobic cometabolism Field & laboratory In situ studiesa Laboratory Overall aerobic Aerobic/anaerobic (field studies) Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory
range reported rates
geometric meanc
90th percentile
standard deviation
mean
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier Number of rates Number of reported rates Number of calculated ratesb
1594
Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory PCE All studies
50
31
19
0.051
0.084 0.153 0.000
0-0.410
Aerobic oxidation In situ & laboratory In situ studiesa Laboratory
10 3 7
5 3 2
5 0 5
0.001 0.000 0.001
0.001 0.003 0.000 0.002 0.003 0.000
0-0.004 0.000 0-0.004
3
1
2
0.025
0.000
0-0.054
3
1
2
0.025
0.000
0-0.054
Overall aerobic
13
6
7
0.006
0.015 0.017 0.000
0-0.054
Aerobic/anaerobic (field studies)
1
1
0
36 13 23
23 9 14
13 4 9
0.068 0.010 0.101
0.093 0.185 0.002 0.022 0.022 0.000 0.101 0.212 0.024
0-0.410 0-0.080 0-0.410
Aerobic cometabolism Field & laboratory In situ studiesa Laboratory
Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory
TCA All studies
47
27
20
0.261
0.502 1.026 0.000
0-2.330
Aerobic oxidation In situ & laboratory In situ studiesa Laboratory
11 2 9
7 2 5
4 0 4
0.002
0.007 0.009 0.000
0-0.022
0.003
0.007 0.005 0.000
0-0.022
5
1
4
0.247
0.522 0.723 0.001
0-1.180
5
1
4
0.247
0.522 0.723 0.001
0.-1.180
16
8
8
0.079
0.294 0.030 0.000
0-1.180
Aerobic cometabolism Field & laboratory In situ studiesa Laboratory Overall aerobic Aerobic/anaerobic (field studies)
0.562 1.110 0.003 0.039 0.058 0.000 0.629 1.280 0.007
12 7 5
range reported rates
0.355 0.029 0.511
19 3 16
geometric meanc
standard deviation
31 10 21
90th percentile
mean
Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory
1595
Number of rates Number of reported rates Number of calculated ratesb
23.1 Natural attenuation of chlorinated solvents
0-2.330 0-0.125 0-2.330
Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory TCE All studies
86
52
34
0.173
0.475 0.636 0.001
0-3.130
Aerobic oxidation In situ & laboratory In situ studiesa Laboratory
12 2 10
6 2 0
6 0 10
0.005
0.010 0.025 0.000
0-0.028
0.006
0.011 0.026 0.000
0-0.028
Aerobic cometabolism Field & laboratory In situ studiesa Laboratory
17 3 14
7 2 5
10 1 9
0.586 0.948 0.509
0.566 1.418 0.309 0.582 0.524 1.265 0.269
0.024-1.650 0.105-1.410 0.024-1.650
Overall aerobic
29
13
16
0.346
0.517 1.354 0.001
0-1.650
Aerobic/anaerobic (field studies)
1
1
0
56 32 24
38 26 12
18 6 12
0.086 0.003 0.196
0.434 0.022 0.001 0.005 0.006 0.000 0.654 0.337 0.012
0-3.130 0-0.023 0-3.130
0-1.960
Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory
Vinyl chloride All studies Aerobic oxidation In situ & laboratory In situ studiesa Laboratory Aerobic cometabolism Field & laboratory In situ studiesa Laboratory
26
8
18
0.229
0.476 0.946 0.023
4
0
4
0.087
0.080
0.043-0.125
4
0
4
0.087
0.080
0.043-0.125
4 2 2
0 0 0
4 2 2
1.023 1.730 0.316
0.552 1.715 0.178
0.055-1.960 1.500-1.960 0.055-0.576
Overall aerobic
8
0
8
0.555
0.756 0.107 0.211
0.043-0.120
Aerobic/anaerobic (field studies)
3
2
1
0.004
0.002
0.001-0.009
Reductive dechlorination Field & laboratory Field/in situ studiesa Laboratory
8 4 4
5 4 1
3 0 3
0.153 0.003 0.303
0.228 0.499 0.007 0.001 0.036
0-0.520 0-0.007 0-0.520
Anaerobic oxidation Field & laboratory Field/in situ studiesa Laboratory
7 1 6
1 1 0
6 0 6
0.042
0.048 0.104 0.018
0.001-0.120
0.049
0.048 0.107 0.028
0.008-0.120
range reported rates
standard deviation
geometric meanc
mean
90th percentile
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier Number of rates Number of reported rates Number of calculated ratesb
1596
a
In situ studies include in situ microcosms and in situ columns When enough information was provided by the authors of a study, the authors of this paper calculated the rate coefficient assuming first-order kinetics c To calculate the geometric mean, values equal to zero were included as 10-10 [From M.P. Suarez and H.S. Rifai, Bioremediation Journal, 3, 337-362. Copyright © 1999 Battelle Memorial Institute. Reprinted with permission.] b
The biodegradability under different electron acceptors for each one of the chlorinated solvents was also analyzed by Suarez and Rifai.90 As summarized in Table 23.1.8, DCA presented very high potential for biodegradation via aerobic cometabolism and reductive dechlorination with none of the studies reporting recalcitrance. Median half-lives for this compound were 1,260 days and 15 days for reductive dechlorination and cometabolism, respectively. DCE exhibited high potential for aerobic cometabolism with 11% of the studies showing recalcitrance and a very short median half-life (1 day). None of the 44 studies on reductive dechlorination of DCE reported recalcitrance, which leads to the conclusion that DCE may undergo this process though with a relatively slow rate (median half-life equal to 234 days). Table 23.1.8. Biodegradability of chlorinated solvents Process All Studies
Carbon tetrachloride 13 # rates 0 # rates-recalcitrant 14 half-life (days)a % rates recalcitrant 0% potential for biodegradationb almost always
Aerobic oxidation
1 0 NC 0% NA
Cometabolism
1 0 NC 0% NA
Reductive dechlorination
11 0 9 0% almost always
Anaerobic oxidation
23.1 Natural attenuation of chlorinated solvents
1597
Process All Studies
DCA (all isomers) 25 # rates 2 # rates-recalcitrant 990 half-life (days)a % rates recalcitrant 8% potential for biodegradationb almost always
Aerobic oxidation
Cometabolism
Reductive dechlorination
2 2 NC 100% NA
5 0 15 0% almost always
18 0 1260 0% almost always
13 2 2 15% frequently
44 0 234 0% almost always
3 1 35 35% NA
36 5 32 14% frequently
DCE (all isomers) 61 # rates 3 # rates-recalcitrant 173 half-life (days)a % rates recalcitrant 5% potential for biodegradationb almost always PCE # rates # rates-recalcitrant half-life (days)a % rates recalcitrant potential for biodegradationb
50 14 80 28% sometimes
NC 60% barely
TCA # rates # rates-recalcitrant half-life (days)a % rates recalcitrant potential for biodegradationb
47 14 68 30% sometimes
11 8 NC 73% barely
5 1 53 20% frequently
31 5 24 16% frequently
TCE # rates # rates-recalcitrant half-life (days)a % rates recalcitrant potential for biodegradationb
85 12 151 14% frequently
11 6 NC 55% barely
17 0 3 0% almost always
56 5 201 9% almost always
10 6
Vinyl chloride 27 4 5 # rates 0 0 0 # rates-recalcitrant 14 8 0.462 half-life (days)a % rates recalcitrant 0 0% 0% potential for biodegradationb almost always almost always almost always
Anaerobic oxidation
15 7 0 0 80 58 0% 0% almost always almost always
a Median value from the reported studies; bQuantitative estimation based on % occurrence of recalcitrance; NA Insufficient information; NC Not calculable (λ=0); Scale % recalcitrance - biodegradability: < 10% - Almost always, 10%-25% - Frequently, 25%-50% - Sometimes, 50%-75% - Barely, >75% - Almost never. [From M.P. Suarez and H.S. Rifai, Bioremediation Journal, 3, 337-362. Copyright © 1999 Battelle Memorial Institute. Reprinted with permission.]
The process that exhibited the highest potential for biodegradation of PCE and TCA was reductive dechlorination with 86% and 84% of the analyzed studies showing
1598
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
biotransformation. Median half-lives for reductive dechlorination of PCE and TCA were 34 days and 24 days, respectively. With respect of TCE, none of the 17 studies reporting aerobic cometabolism (most of them laboratory studies) showed recalcitrance and the median half-life was very short (3 days). Reductive dechlorination also appeared to be a very good alternative for biotransformation of TCE with only 9% of 56 studies reporting recalcitrance and median half-life equal to 201 days. Finally, vinyl chloride exhibited very high potential for biodegradation under aerobic conditions with no studies showing recalcitrance and median half-lives of 8 days and 0.462 days for oxidation and cometabolism, respectively. In addition to the data reported above by Suarez and Rifai,90 a groundwater anaerobic biodegradation literature review was performed by Aronson and Howard.102 Based on their review, Aronson and Howard102 developed a range of “recommended values” for the anaerobic biodegradation first order decay rate coefficients. For many of the chlorinated solvents, the authors defined the low-end rate coefficient based on the lowest measured field value, and defined the high-end value as the mean rate coefficient for all the field/in-situ microcosm studies. Table 23.1.9 shows the resulting recommended ranges for first order anaerobic biodegradation rate coefficients for several chlorinated solvents along with the mean value of the field/in-situ microcosm studies (note that some minor discrepancies exist between the reported high-end rates and the mean value for the field/in-situ microcosm studies). Table 23.1.9. Mean and recommended first-order rate coefficients for selected chlorinated solvents presented by Aronson and Howard102 Mean of field/in-situ studies
st
Recommended 1 order rate coefficients
st
number
low-end 1 order rate
coefficients, -1 day
halflives, day
studies used for mean
coefficients, -1 day
PCE
0.0029
239
16
0.00019
3,647
0.0033
210
Lower limit was reported for a field study under nitrate-reducing conditions
TCE
0.0025
277
47
0.00014
4,950
0.0025
277
Lower limit was reported for a field study under unknown redox conditions
Vinyl chloride
0.0079
88
19
0.00033
2,100
0.0072
96
Lower limit was reported for a field study under methanogenic/sulfate-reducing conditions
0.016
43
15
0.0013
533
0.01
69
Range not appropriate for nitrate-reducing conditions. Expect lower limit to be much less
0.0076
91
2
0.0042
165
0.011
63
Range reported from a single field study under methanogenic conditions
Compound
1,1,1-TCA
1,2-DCA
1 order rate
st
high-end st
1 order rate
Comments
coeffihalf-lives half-lives cients, day day -1 day
23.1 Natural attenuation of chlorinated solvents
Mean of field/in-situ studies
1599 st
Recommended 1 order rate coefficients
st
number
low-end 1 order rate
coefficients, -1 day
halflives, day
studies used for mean
coefficients, -1 day
Carbon tetrachloride
0.37
1.9
9
0.0037
187
0.13
5
Range not appropriate for nitrate-reducing conditions. Expect lower limits to be much less
Chloroform
0.030
23
1
0.0004
1,733
0.03
23
Only one field study available. Biodegradation under nitrate-reducing conditions expected
Dichloromethane
0.0064
108
1
0.0064
108
-
-
Rate constant reported from a single field study under methanogenic conditions
Trichlorofluoromethane
-
-
-
0.00016
4,331
0.0016
433
All studies with very low concentrations of this compound
50
2
0.00055
1,260
0.027
26
Range may not be appropriate for nitrate reducing conditions
Compound
1 order rate
2, 4-Dichlorophenol 0.014
st
high-end st
1 order rate
Comments
coeffihalf-lives half-lives cients, day day -1 day
[From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
23.1.5 GEOCHEMICAL EVIDENCE OF NATURAL BIOREMEDIATION AT CHLORINATED SOLVENT SITES 23.1.5.1 Assessing reductive dechlorination at field sites Assessing biological activity at a field site based on monitoring data can be difficult. However, there are a number of monitoring parameters that can be indicative of halorespiration. First, the presence of methane in the groundwater indicates that fermentation is occurring and that the potential for halorespiration exists. Second, the transformation of PCE and TCE has been studied intensely and many researchers report that of the three possible DCE isomers, 1,1-DCE is the least significant intermediate and that cis-1,2-DCE predominates over trans-1,2-DCE.103-105 Third, because chlorinated ethenes are 55 to 85% chlorine by mass, the degradation of these compounds releases a large mass of chloride. Therefore, elevated chloride concentrations also indicate reductive dechlorination. 23.1.5.2 Plume classification schemes Wiedemeier et a1.106 proposed a classification system for chlorinated solvent plumes based on the amount and origin of fermentation substrates that produce the hydrogen that drives halorespiration. Three types of groundwater environments and associated plume behavior, Type 1, Type 2, and Type 3, are described below. While the classification system can be used to represent entire plumes, it can also be used to define different zones within a chlorinated solvent plume. 23.1.5.2.1 Type 1 For highly chlorinated solvents to biodegrade, anaerobic conditions must prevail within the contaminant plume. Anaerobic conditions are typical at sites contaminated with fuel hydro-
1600
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
carbons, landfill leachate, or other anthropogenic carbon because these organics exert a tremendous electron-acceptor demand on the system. This condition is referred to as a Type 1 environment. In a Type 1 environment, anthropogenic carbon is fermented to produce hydrogen which drives halorespiration. The geochemistry of groundwater in a Type 1 environment is typified by strongly reducing conditions. This environment is characterized by very low concentrations of dissolved oxygen, nitrate, and sulfate and elevated concentrations of Fe(II) and methane in the source zone (Figure 23.1.6). The presence of methane is almost always observed and confirms that fermentation has been occurring at the site, generating hydrogen. If measured, hydrogen concentrations are typically greater than 1 nanomolar. Importantly, a Type 1 environment results in the rapid and extensive degradation of the more highlychlorinated solvents such as PCE, TCE, and DCE: PCE → TCE → DCE → VC → Ethene → Ethane In this type of plume, cis-1,2-DCE and VC degrade more slowly than TCE; thus, they tend to accumulate and form longer Figure 23.1.6. Conceptual model of Type 1 environment for chlorinated solvent plumes. [From T.H. plumes (Figure 23.1.6a). In Figure 23.1.6b, Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, the PCE declines to zero and is replaced, in Natural Attenuation of Fuels and Chlorinated Sol- sequence, by a peak in TCE concentrations, vents in the Subsurface, after reference 88. Copyright followed by a peak in cis-1,2-DCE, VC, and © 1999 John Wiley & Sons, Inc. Reprinted by permisethene. Fermentation constituents (BOD and sion of John Wiley & Sons, Inc.] acetate) and inorganics are shown in Figure 23.1.6c and 23.1.6d. Figure 23.1.6d illustrates how the fermentation substrate (represented by BOD) extends beyond the source before being consumed. Both panels show long chloride and methane plumes extending far downgradient from the plume area, because chloride is conservative and methane cannot be biodegraded in an anaerobic environment. The acetate curve indicates where active primary fermentation is occurring; declining acetate concentrations are due to consumption by methanogens in the plume area. 23.1.5.2.2 Type 2 The classification system of Wiedemeier et a1.107 recognized that anaerobic conditions may also result from the fermentation of naturally-occurring organic material in the groundwater that flows through chlorinated solvent source zones. This Type 2 environment occurs in hydrogeologic settings that have inherently high organic carbon concentrations, such as coastal or stream/river deposits with high concentrations of organics, shallow aquifers with
23.1 Natural attenuation of chlorinated solvents
1601
recharge zones in organic-rich environments (such as swamps), or zones impacted by natural oil seeps. A Type 2 environment generally results in slower biodegradation of the highly chlorinated solvents compared to a Type 1 environment. However, given sufficient organic loading, this environment can also result in rapid degradation of these compounds. A Type 2 environment typically will not occur in crystalline igneous and metamorphic rock (see discussion of likely hydrogeologic settings for Type 3 environments). 23.1.5.2.3 Type 3 A Type 3 environment is characterized by a well-oxygenated groundwater system with little or no organic matter. Concentrations of dissolved oxygen typically are greater than 1.0 mg/L. In such an environment, halorespiration will not occur and chlorinated solvents such as PCE, TCE, TCA, and CT will not biodegrade. In this environment, very long dissolved-phase plumes are likely to form. The most significant natural attenuation mechanisms for PCE and TCE will be advection, dispersion, and sorption. However, VC (and possibly DCE) can be rapidly oxidized under these conditions. A Type 3 environment is often found in crystalline igneous and metamorphic rock (fractured or unfractured) such as basalt, granite, schist, phyllite, glacial outwash deposits, eolian deposits, thick deposits of well-sorted, clean, beach sand with no associated peat or other organic carbon deposits, or any other type of deposit with inherently low organic carbon content if no anthropogenic carbon has been released. Two conceptual models are provided for environments in which Type 3 behavior occurs. For sources with PCE and TCE, the major natural attenuation processes are dilution and dispersion alone (no biodegradation). As shown in 23.1.7, the PCE and TCE plumes extend from the source zone and concentrations are slowly reduced by abiotic processes. Chloride concentrations and oxidation-reduction potential will not change as groundwater passes through the source zone and forms the chlorinated ethene plume. If TCA is the solvent of interest, significant abiotic hydrolysis may occur, resulting in a more rapid decrease in TCA concentrations and an increase in chloride concentrations. In Figure 23.1.7, a source releases VC and 1,2-DCA into the groundwater at a Type 3 site (an unlikely occurrence as more highly chlorinated solvents are typically released at sites). Because the VC and 1,2-DCA can be degraded aerobically, these constituents decline in concentration at a significant rate. Chloride is produced, and a depression in dissolved oxygen concentration similar to that occurring at fuel sites, is observed. 23.1.5.2.4 Mixed environments As mentioned above, a single chlorinated solvent plume can exhibit different types of behavior in different portions of the plume. This can be beneficial for natural biodegradation of chlorinated solvent plumes. For natural attenuation, this may be the best scenario. PCE, TCE, and DCE are reductively dechlorinated with accumulation of VC near the source area (Type 1); then, VC is oxidized (Type 3) to carbon dioxide, either aerobically or via Fe(III) reduction further downgradient and does not accumulate. Vinyl chloride is removed from the system much faster under these conditions than under reducing conditions. A less ideal variation of the mixed Type 1 and Type 3 environments is shown in the conceptual model in Figure 23.1.8. An extended TCE and 1,2-DCE plume results because insufficient fermentable carbon results in an anaerobic zone which is too short for complete biodegradation. Therefore, TCE extends well into the aerobic zone where no biodegradation occurs. A long DCE plume also extends into the aerobic zone, indicating in-
1602
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
Figure 23.1.7. Conceptual model of Type 3 environment for chlorinated solvent plumes. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
Figure 23.1.8. Conceptual model of mixed environments with Type 1 environment in source zone and Type 3 environment downgradient of source. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
significant direct aerobic biodegradation was assumed. While a long chloride plume will be observed, the short anaerobic zone means much less methane is produced, allowing dilution/dispersion to limit the extent of the methane plume. 23.1.6 CHLORINATED SOLVENT PLUMES - CASE STUDIES OF NATURAL ATTENUATION 23.1.6.1 Plume databases Two different databases provided chlorinated solvent site data. The first database, the Hydrogeologic Database (HGDB),2 provided information on plume length, plume width, plume thickness, and highest solvent concentration for 109 chlorinated solvent sites. The second database1 condensed extensive site characterization data from 17 Air Force chlorinated solvent sites, with information on parent compounds vs. progeny products concentrations, competing electron acceptors, hydrogen, and metabolic by-products.
23.1 Natural attenuation of chlorinated solvents
1603
The data in the HGDB were broken into two groups: the chlorinated ethenes, where one or more of the chlorinated ethenes (PCE, TCE, DCE, or VC) was reported to be the major contaminant, and other chlorinated solvent sites, where all other chlorinated solvents besides the ethenes (e.g., TCA, DCA, chlorobenzene) were lumped together. As shown in Table 23.1.10, the median length of the 75 chlorinated ethene plumes was 1000 ft, with one site reporting a plume length of 13,200 ft. These median lengths are longer than those reported for fuel hydrocarbon plumes and this may be attributed to the competition for hydrogen during halorespiration. Table 23.1.10. Characteristics of chlorinated solvent plumes from HGDB database Plume length, ft
Plume width, ft
Vertical penetration, ft
Highest concentration, mg/L
Chlorinated ethenes (e.g., PCE, TCE, etc.) Maximum 75th percentile Median 25th percentile Minimum n
13,200 2,500 1,000 600 50 75
4,950 1,000 500 200 15 75
500 100 40 25 5 78
28,000 72 8.467 0.897 0.001 81
Other chlorinated solvents (e.g., TCA, DCA) Maximum 75th percentile Median 25th percentile Minimum n
18,000 2,725 575 290 100 24
7,500 1,000 350 188 100 24
150 51 35 24 8 24
2,500 13.250 3.100 0.449 0.016 28
Note: Highest concentration for chlorinated ethenes (28,000 mg/L) was for TCE, which is above the solubility limit. The highest concentration for “other chlorinated solvents” (2500 mg/L) was for chloromethane and toluene. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
The other category, “other chlorinated solvent sites,” had shorter plumes, with a median plume length of 575 ft compared to 1000 ft for the chlorinated ethene sites. Twelve of the 24 plumes were comprised of TCA, which is degraded biologically via halorespiration and other mechanisms and abiotically by hydrolysis (half life of 0.5 to 2.5 years). Despite the degradability of TCA, the TCA plumes had a median length of 925 ft. The shorter plumes in this database of 24 sites were reported to be comprised of either a general indicator, such as Total Organic Halogens, or individual compounds such as 1,1-dichloroethane, dichloromethane, or chlorobenzene. The median highest concentration at these “other chlorinated solvent sites” was 3.1 mg/L (see Table 23.1.10). Data compiled from 17 Air Force sites using the AFCEE (Air Force Center for Environmental Excellence) natural attenuation protocol107 showed a median plume length of 750 ft (based on 14 plumes). There were significant differences in plume length for different plume classes, with Type 1 plumes (sites with available man-made fermentation substrates such as BTEX) being shorter than Type 3 plumes (sites without available fermentation substrates). Twelve of the sites exhibiting Type 1 plumes had a median plume length of 625 ft,
1604
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
while the two sites with Type 3 plumes had lengths of 1100 and 5000 ft. Four mixed plumes (Type 1 in source zone, Type 3 in the downgradient part of the plume) had a median length of 2538 ft. Site-specific biodegradation rate information was developed using several methods, including one developed by Buscheck and Alcantar,108 one based on the use of conservative tracers (the trimethylbenzenes), and other methods such as model calibration. Rates varied significantly, with half-lives ranging from over 300 years for a Type 3 site located in Utah to 0.2 years for a Type 1 site (see Table 23.1.11). The median first order half-life for the 14 chlorinated plumes was 2.1 years.
1400
40
4.953
60
0.000006
316.4
Other
9
NY
Mixed
4200*
2050
60
774.721
139
0.001*
1.9*
Other
10
NE
Mixed
3500
1400
50
164.010
152
0.000001
1899
Other
8
MA Type 1
1800
1200
50
4.340
106
0.0005
3.8
Other
11
FL
Mixed
1575*
400
15
1258.842
113
0.0009*
2.1*
Other
14
AK
Type 3
1100
250
25
4.899
260
0.0065
0.3
Other
1
SC
Type 1
750
550
5
328.208
1600
-
-
-
7
MA Type 1
750
250
50
50.566
20.8
0.0095
0.2
Conserv. tracer
12
MS
Mixed
750
550
5
0.472
1500
0.01
0.2
Buscheck
4
NE
Type 1
650
450
30
47.909
6.7
0.0006
3.2
Buscheck
3
FL
Type 1
600
350
20
0.429
36
0.0007
2.7
Buscheck
5
WA Type 1
550
300
10
3.006
32.9
0.001
1.9
Buscheck
2
MI
Type 1
375
100
10
0.397
292
-
-
-
6
OH
Type 1
100
60
10
15.736
25
-
-
-
5,000
2,050
60
1,259
1,600
0.0095
316.4
75 percentile
1,744
1,038
48
136
233
0.00375
3.5
Median
750.0
425.0
22.5
10.3
109.5
0.0009
2.1
25 percentile
613
263
10
3.3
34
0.00055
1.1
Minimum
100
60
5
0.397
7
0.000006
0.2
n
14
14
14
14
14
11
11
Maximum th
th
Half-life, years
Method for calculating rate coefficient
Plume thickness, ft
5000
First-order biodegradation rate coeff. for solvents, day -1
Plume width, ft
Type 3
Seepage velocity, ft/yr
Plume length, ft
UT
Total chlor. solvents, mg/L
State
13
Type
No.
Table 23.1.11. Chlorinated solvent plume characteristics
*Plume discharges into stream; may not represent maximum potential plume length. Mixed refers to Type 1 conditions in source zone, Type 3 conditions in downgradient part of plume. Median length Type 1 sites: 625 ft, mixed sites 2538 ft, Type 3: 3050 ft (two sites). [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
23.1 Natural attenuation of chlorinated solvents
1605
23.1.6.2 Modeling chlorinated solvent plumes Very few models exist (analytical or numerical) which are specifically designed for simulating the natural attenuation of chlorinated solvents in ground water. Ideally, a model for simulating natural attenuation of chlorinated solvents would be able to track the degradation of a parent compound through its daughter products and allow the user to specify differing decay rates for each step of the process. This may be referred to as a reactive transport model, in which transport of a solute may be tracked while it reacts, its properties change due to those reactions, and the rates of the reactions change as the solute properties change. Moreover, the model would also be able to track the reaction of those other compounds that react with or are consumed by the processes affecting the solute of interest (e.g., electron donors and acceptors). Two models: BIOCHLOR and RT3D for the natural attenuation of chlorinated solvents have been presented recently in the general literature and will be briefly discussed in this section. 23.1.6.2.1 BIOCHLOR natural attenuation model The BIOCHLOR Natural Attenuation Model109 simulates chlorinated solvent natural attenuation using an Excel based interface. BIOCHLOR simulates the following reductive dechlorination process: k1 k2 k3 k4 →TCE → DCE →VC → ETH PCE
The equations describing the sequential first order biodegradation reaction rates are shown below for each of the components: rPCE = −k1CPCE
[23.1.16]
rTCE = k1CPCE − k 2 CTCE
[23.1.17]
rDCE = k 2 CTCE − k 3 CDCE
[23.1.18]
rVC = k 3 CDCE − k 4 CVC
[23.1.19]
rETH = k 4 CETH
[23.1.20]
where: k1, k2, k3, k4 CPCE, CTCE, CDCE, CVC and CETH
the first order rate constants the aqueous concentration of PCE, TCE, DCE, vinyl chloride, and ethene, respectively.
These equations assume no degradation of ethene. To describe the transport and reaction of these compounds in the subsurface, one-dimensional advection, three-dimensional dispersion, linear adsorption, and sequential first order biodegradation are assumed as shown in the equations below. All equations, but the first, are coupled to another equation through the reaction term. RPCE
d 2CPCE d 2CPCE d 2CPCE dCPCE dCPCE = −v + Dx + Dy + Dz − kC 1 PCE 2 2 dy dx dx dt dz 2
[23.1.21]
1606
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
RTCE
d 2CTCE d 2CTCE d 2CTCE dC dCTCE = −v TCE + Dx + Dy + Dz + kC 1 PCE − k 2CTCE 2 2 dy dx dx dt dz 2
RDCE
d 2CDCE d 2CDCE d 2CDCE dCDCE dCDCE = −v + Dx + Dy + Dz + k 2CTCE − k 3CDCE [23.1.23] 2 2 dy dx dx dt dz 2
RVC
[23.1.22]
d 2CVC d 2CVC d 2CVC dC dCVC = −v VC + Dx + Dy + Dz + k 3CDCE − k 4CVC dz 2 dy 2 dx 2 dx dt
[23.1.24]
d 2CETH d 2CETH d 2CETH dCETH dCETH = −v + Dx + Dy + Dz + k 4CETH 2 2 dy dx dx dt dz 2
[23.1.25]
RETH where:
RPCE, RTCE, RDCE, RVC, RETH V Dx, Dy, Dz
retardation factors seepage velocity dispersivities in the x, y, and z directions.
BIOCHLOR uses a novel analytical solution to solve these coupled transport and reaction equations in an Excel spreadsheet. To uncouple these equations, BIOCHLOR employs transformation equations developed by Sun and Clement.110 The uncoupled equations were solved using the Domenico model, and inverse transformations were used to generate concentration profiles. Details of the transformation are presented elsewhere.110 Typically, source zone concentrations of cis-1,2-dichloroethythene (DCE) are high because biodegradation of PCE and TCE has been occurring since the solvent release. BIOCHLOR also simulates different first-order decay rates in two different zones at a chlorinated solvent site. For example, BIOCHLOR is able to simulate a site with high dechlorination rates in a high-carbon area near the source that becomes a zone with low dechlorination rates downgradient when fermentation substrates have been depleted. In addition to the model, a database of chlorinated solvent sites is currently being analyzed to develop empirical rules for predicting first-order coefficients that can be used in BIOCHLOR. For example, at sites with evidence of considerable halorespiration, the use of higher first order decay coefficients will be recommended. Indicators of high rates of halorespiration may include: i) high concentrations of fermentation substrates such as BTEX at the site, ii) high methane concentrations, which indicate high rates of fermentation, and iii) large ratios of progeny products to parent compounds, and iv) high concentrations of source zone chloride compared to background chloride concentrations. The BIOCHLOR model was used to reproduce the movement of the Cape Canaveral plume from 1965 to 1998. The Cape Canaveral site (Figure 23.1.9) is located in Florida and exhibits a TCE plume which is approximately 1,200 ft long and 450 ft wide. TCE concentrations as high as 15.8 mg/L have been measured recently at the site. The site characteristics used in the BIOCHLOR model are listed in Table 23.1.12. The hydraulic conductivity assumed in the model was 1.8x10-2 cm/sec and the hydraulic gradient was 0.0012. A porosity of 0.2 was assumed as well as the Xu and Eckstein model for longitudinal dispersivity.8 The lateral dispersivity was assumed to be 10% of the longitudinal dispersivity and vertical dispersion was neglected.
23.1 Natural attenuation of chlorinated solvents
1607
Figure 23.1.9. BIOCHLOR source zone assumptions (TCE as example), CCFTA-2, cape Canaveral Air Station, Florida [From reference 109].
1608
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
Table 23.1.12. BIOCHLOR example, Cape Canaveral Air Station, Florida. [From reference 109] Data type
Parameter
Value
Source Slug-tests results Static water level measurement Estimated
1.8 x 10-2 cm/s 0.0012 ft/ft 0.2
Hydrogeology
Hydraulic conductivity: Hydraulic gradient: Porosity:
Dispersion
Original Longitudinal dispersivity: Transverse dispersivity: Vertical dispersivity:
Adsorption
Individual retardation fac- PCE: 6.7 c-DCE:2.8 tors: ETH: 5.3 Common retardation factor: 5.3 1.6 kg/L Soil bulk density ρb: foc: 0.184% Koc: (L/kg) PCE: 398 c-DCE: 126 ETH: 302
Based on estimated plume length of 1450 ft. Note: No calibration was necessary to match observed plume length
varies with x varies with x 0 ft TCE: 2.8 VC: 5.6
TCE: 126 VC: 316
Calculated Median value Estimated Lab analysis Literature correlation using solubilities at 20°C
Biodegradation
Biodegradation rate coefficients (1/year): PCE → TCE 2.0 TCE → c-DCE 0.9 c-DCE → VC 0.6 C → ETH 0.4
Based on calibration to field data using a simulation time of 32 yr. Started with literature values and then adjusted model to fit field
General
Modeled area length Modeled area width Simulation time
1085 ft 700 ft 33 yrs
Based on area of affected groundwater plume from 1965 (first release) to 1998 (present)
Source thickness Source widths, ft
56 ft Zone 1 105
Zone 2 175
Zone 3 298
PCE TCE c-DCE VC ETH
Zone 1 0.056 15.8 98.5 3.080 0.030
Zone 2 0.007 0.318 1.0 0.089 0.013
Zone 3 0.001 0.01 0.01 0.009 0.003
Actual data
Distance from source, ft PCE concentration, mg/L TCE, mg/L c-DCE, mg/L VC, mg/L ETH, mg/L
560 <0.001 0.22 3.48 3.080 0.188
Output
Centerline concentration
see Figure 23.1.10
Source concentrations, mg/L Source data
650 930 ND <0.001 0.0165 0.0243 0.776 1.200 0.797 2.520 ND 0.107
1085 <0.001 0.019 0.556 5.024 0.150
Based on geologic logs and monitoring data. Source concentrations are aqueous concentrations
Based on observed concentration at site near centerline of plume
A median value for the retardation factor was used (R=5.3) since BIOCHLOR accepts only one value for this parameter. The site was modeled using one anaerobic zone with one set of rate coefficients as shown in Table 23.1.12. This is justified because the dissolved ox-
23.1 Natural attenuation of chlorinated solvents
1609
ygen readings at the site were less than 0.7 mg/L at all monitoring points. The rate coefficients were calculated by calibrating the model to the 1997 field data. The source zone was simulated as a spatially-variable source and the source concentrations ranged from 0.001 to 98.5 mg/L for the various compounds as shown in Table 23.1.12. The source thickness was estimated by using the deepest point in the aquifer where chlorinated solvents were deFigure 23.1.10. Centerline output. Cape Canaveral Air Force Base, tected. Florida. [From references 109]. Centerline concentrations for all five species (PCE, TCE, c-DCE, VC and ETH) predicted by the model are shown in Figure 23.1.10. Figure 23.1.10 shows the centerline predictions for each chlorinated solvent and a no degradation curve for all of the chlorinated solvents as well as field data. The data in Figure 23.1.10 indicate that TCE concentrations discharging into the ocean will be less than 0.001 mg/L. 23.1.6.3 RT3D numerical model RT3D (Reactive Transport in 3 Dimensions)111 is a FORTRAN 90-based model for simulating 3D multi-species, reactive transport in groundwater. This model is based on the 1997 version of MT3D (DOD Version 1.5), but has several extended reaction capabilities. RT3D can accommodate multiple sorbed and aqueous phase species with any reaction framework that the user needs to define. RT3D can simulate different scenarios, since a variety of pre-programmed reaction packages are already provided and the user has the ability to specify their own reaction kinetic expressions. This allows, for example, natural attenuation processes or an active remediation to be evaluated. Simulations can be applied to modeling contaminants such as heavy metals, explosives, petroleum hydrocarbons, and/or chlorinated solvents. RT3D’s pre-programmed reaction packages include: 1 Two species instantaneous reaction (hydrocarbon and oxygen). 2 Instantaneous hydrocarbon biodegradation using multiple electron acceptors (O2, NO3-, Fe2+, SO42-, CH4). 3 Kinetically limited hydrocarbon biodegradation using multiple electron acceptors (O2, NO3-, Fe2+, SO42-, CH4). 4 Kinetically limited reaction with bacterial transport (hydrocarbon, oxygen, and bacteria). 5 Non-equilibrium sorption/desorption. Can also be used for non-aqueous phase liquid dissolution). 6 Reductive, anaerobic biodegradation of PCE, TCE, DCE, and VC. 7 Reductive, anaerobic biodegradation of PCE, TCE, DCE, and VC combined with aerobic biodegradation of DCE and VC. 8 Combination of #3 and #7.
1610
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
Figure 23.1.11. Site layout, Plattsburg Air Force Base, New York. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
Figure 23.1.12. Hydrogeologic section, Plattsburg Air Force Base, New York. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
23.1 Natural attenuation of chlorinated solvents
1611
Figure 23.1.13. Chlorinated solvents and by-products, 1995, Plattsburg Air Force Base, New York. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
RT3D represents a remarkable breakthrough in the development and solving of optimization models of bioremediation design. It is a modular, three-dimensional simulator capable of predicting mufti-species (solutes and microbes), bio-reactive transport while understanding natural attenuation and active bioremediation processes. 23.1.6.4 CS case study - The Plattsburgh Air Force Base The Plattsburgh Air Force Base (AFB) in New York is a former fire training facility (site FT002). Activities at FT-002 (Figure 23.1.11) have caused contamination of shallow soils and groundwater with a mixture of chlorinated solvents and fuel hydrocarbons. Groundwater contaminants include TCE, cis-1,2-DCE, VC and BTEX. The site is underlain with 4 distinct stratigraphic units: sand, clay, till and carbonate bedrock. The depth to groundwater in the sand aquifer ranges from 45 ft below ground surface (BGS) on the west side of the site to zero on the east side of the runway (Figure 23.1.12). Groundwater flow is to the southeast and the average gradient is about 0.01 ft/ft.l Hydraulic conductivity values for the unconfined sand aquifer range from 0.059 to 90.7 ft/day. Wiedemeier et al.1 estimated an average velocity of 142 ft/yr for the sand aquifer. The extent of Light Non-Aqueous Phase (LNAPL) contamination at Plattsburgh is shown in Figure 23.1.13. The LNAPL is a mixture of jet fuel and waste solvents from which BTEX and TCE dissolve into the ground water (DCE and VC are not present in the waste mixture). The dissolved BTEX plume (Figure 23.1.13) extends approximately 2000 ft downgradient from the site and has a maximum width of about 500 ft. BTEX concentrations as high as 17 mg/L were measured in the source area. Historical data from FT-002 indicate that the dissolved BTEX plume has reached a quasi-steady state and is no longer expanding.
1612
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
Figure 23.1.14. BTEX and electron acceptors, 1995, Plattsburg Air Force Base, New York. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
Figure 23.1.15. BTEX and metabolic by-products, 1995, Plattsburg Air Force Base, New York. [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
The chlorinated solvent plumes (Figure 23.1.13) in groundwater extend about 4000 ft downgradient from FT-002. Concentrations of TCE, DCE and VC as high as 25, 51, and 1.5 mg/L, respectively, have been observed recently. Since DCE and VC were not measured in LNAPL samples from the source area, the presence of DCE and VC at the site can be attributed to dechlorination. The data in Figure 23.1.14 show the distribution of electron acceptor concentrations observed at the site including dissolved oxygen, nitrate and sulfate. Background concentrations for these compounds are 10, 10 and 25 mg/L, respectively and their absence within the contaminated zones is an indication of biodegradation of BTEX and chlorinated solvents at the site. Figure 23.1.15, on the other hand, shows the distribution of metabolic by-products of the biodegradation reactions including ferrous iron and methane. The presence of these by-products is further evidence of biological activity in the aquifer. Elevated chloride and ethene concentrations as shown in Figure 23.1.13 suggest that TCE, DCE and VC are being biodegraded. Wiedemeier et al.1 calculated apparent biodegradation constants for FT-002 using trimethylbenzene as a conservative tracer. Their results are shown in Table 23.1.13. The data in Table 23.1.13 indicate biodegradation rates of the chlorinated solvents at the site ranging between 0 and 1.27 per yr.
23.1 Natural attenuation of chlorinated solvents
1613
Table 23.1.13. Approximate first-order biodegradation rate constants Compound
Correction method
A-B 0-970 ft, year-1
B-C 970-1240 ft, year-1
C-E 1240-2560 ft, year-1
TCE
Chloride TMB Average
1.27 1.20 1.24
0.23 0.52 0.38
-0.30 NA -0.30
DCE
Chloride TMB Average
0.06 0.00 0.03
0.60 0.90 0.75
0.07 NA 0.07
VC
Chloride TMB Average
0.00 0.00 0.00
0.14 0.43 0.29
0.47 NA 0.47
BTEX
Chloride TMB Average
0.13 0.06 0.10
0.30 0.60 0.45
0.39 NA 0.39
a
NA, not analyzed [From T.H. Wiedemeier, H. S. Rifai, C. J. Newell and J.T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, after reference 88. Copyright © 1999 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.]
Available geochemical data were analyzed by Wiedemeier et al.1 and they concluded that the geochemistry of the ground water near the source area and for about 1500 ft downgradient is significantly different from the groundwater further downgradient from the source (between 1500 and 4000 ft downgradient). This led the authors to conclude that Plattsburgh exhibits Type 1 behavior near the source and Type 3 behavior within the leading edge of the plume (see Section 23.1.5.2). In the area extending to 1500 ft downgradient from the source, BTEX and TCE are comingled in the ground water. This area is characterized by anaerobic conditions that are strongly reducing. BTEX is being used as a primary substrate and TCE is being reductively dechlorinated to cis-1,2-DCE and VC. Between 1500 and 2000 ft downgradient from the source, however, the majority of the BTEX has been biodegraded and the system exhibits Type 3 behavior. These conditions are not optimal for reductive dechlorination, and it is likely that VC is being oxidized via ferric reduction or aerobic respiration. REFERENCES 1 2 3
4 5 6 7 8 9
T. H. Wiedemeier, H. S. Rifai, C. J. Newell and J. T. Wilson, Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface, John Wiley & Sons, Inc, New York, NY, 1999. C. J. Newell, L. P. Hopkins and P. B. Bedient, Ground Water, 28, 703-714 (1990). L. W. Gelhar, A. Montoglou, C. Welty and K. R. Rehfeldt, A Review of Field Scale Physical Solute Transport Processes in Saturated and Unsaturated Porous Media: Final Project Report, EPRI-EA-4190, Electric Power Research Institute, Palo Alto, CA, 1985. L. W. Gelhar, L. Welty and K. R. Rehfeldt, Water Resour. Res., 28, 1955-1974 (1992). P. Lallemand-Barres and P. Peaudecerf, Etude Bibliographique Bulletin, Sec., 3/4, 277-287 (1978). J. F. Pickens and G. E. Grisak, Water Resour. Res., 17, 1191-1211 (1981). K. Spitz and J. Moreno, A Practical Guide to Groundwater and Solute Transport Modeling, Wiley, Inc., New York, New York, 1996. M. Xu and Y. Eckstein, Journal of Ground Water, 33, 905-908 (1995). American Society for Testing and Materials (ASTM), Emergency Standard Guide for Risk-Based Corrective Action Applied at Petroleum Release Sites, ASTM E-1739, Philadelphia, 1995.
1614
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
37 38 39 40 41 42 43 44
45
46 47
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
U.S. Environmental Protection Agency, Background Document for the Ground-Water Screening Procedure to Support 40 CFR Part 269: Land Disposal, EPA/530-SW-86-047, Washington, D.C., 1986. W. J. Lyman, P. J. Reidy and B. Levy, Mobility and Degradation of Organic Contaminants in Subsurface Environments, Ed., C. K. Smoley, CRC, Boca Raton, FL, 1992. J. Dragun, The Soil Chemistry of Hazardous Materials, Ed., Hazardous Materials Control Research Institute, Silver Spring, MD, 1988. G. W. Bailey and J. L. White in Residue Reviews, F. A. Gunther and J. D. Gunther, Ed., Springer Verlag, New York, 1970, pp. 29-92. S. W. Karickhoff, D. S. Brown and T. A. Scott, Water Resour. Res., 13, 241-248 (1979). E. E. Kenaga and C. A. Goring, ASTM Special Technical Publication, 707, ASTM, Philadelphia, 1980. D. S. Brown and E. W. Flagg, J. Environ. Qual., 10, 382-386 (1981). R. P. Shwarzenbach and J. Westall, Environ. Sci. Tech., G15, 1360-1367 (1981). J. J. Hassett, W. L. Banwart and R. A. Griffin in Environment and Solid Wastes, C. W. Francis and S. I. Auerbach, Ed., Butterworth, Boston, 1983, pp. 161-178. C. T. Chiou, P. E. Porter and D. W. Schmedding, Environ. Sci. Tech., 17, 227-231 (1983). R. A. Freeze and J. A. Cherry, Groundwater, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1979. R. A. Larson and E. J. Weber, Reaction Mechanisms in Environmental Organic Chemistry, Lewis Publishers, Boca Raton, 1994. U. S. E. P. Agency, Hazardous Waste Treatment, Storage, and Disposal Facilities (TSDF), EPA-450/3-87-026, USEPA, OAQPS, Air Emission Models, 1989. K. Verschueren, Handbook of Environmental Data on Organic Chemicals, Second Ed., Van Nostrand Reinhold Company Inc., New York, 1983. NIOSH, Pocket Guide to Chemical Hazards, Ed., U.S. Dept. ofHealth & Human Services, Public Health Service, Centers for Disease Control, National Institute for Occupational Safety and Health, 1990. W. J. Lyman, Handbook of Chemical Property Estimation Methods, McGraw-Hill, New York, 1982. D. M. DiToro, Chemosphere, 14, 1505-1538 (1985). J. H. Montgomery, Groundwater Chemicals Desk Reference, 3rd Ed., Lewis Publishers, Chelsea, MI, 1990. T. M. Vogel, C. S. Criddle and P. L. McCarty, Environ. Sci. Tech., 21, 722-736 (1987). T. M. Vogel in Handbook of Bioremediation, R. D. Norris et al., Ed., Lewis Publishers, Boca Raton, Florida, 1994, pp. 201-225. R. C. Knox, D. A. Sabatini and L. W. Canter, Subsurface Transport and Fate Processes, Lewis Publishers, Boca Raton, Florida, 1993. R. L. Johnson, C. D. Palmer and W. Fish in Fate and Transport of Contaminants in the Subsurface, U.S. EPA, Cincinnati, OH and Ada, OK, 1989, pp. 41-56. P. L. McCarty, Symposium on Natural Attenuation of Chlorinated Organics in Ground Water, Dallas, TX, September 11-13, 1996, U.S. EPA, EPA/540/R-96/509, 1996, pp. 5-9. W. Mabey and T. Mill, J. Phys. Chem. Ref. Data, 7, 383-415 (1978). T. M. Vogel and M. Reinhard, Environ. Sci. Tech., 20, 992-997 (1986). B. J. Butler and J. F. Barker in Dense Chlorinated Solvents and other DNAPLs in Ground Water: History, Behavior, and Remediation, J. F. Pankow and J. A. Cherry, Ed., Waterloo Press, Waterloo, Canada, 1996, pp. P. M. Jeffers, L. M. Ward, L. M. Woytowitch and N. L. Wolfe, Environ. Sci. Tech., 23, 965-969 (1989). T. Vogel and P. L. McCarty, Environ. Sci. Tech., 21, 1208-1213 (1987). W. J. Cooper, M. Mehran, D. J. Riusech and J. A. Joens, Environ. Sci. Tech., 21, 1112-1114 (1987). W. L. Dilling, N. B. Tfertiller and G. J. Kallos, Environ. Sci. Tech., 9, 833-838 (1975). J. March, Advanced Organic Chemistry, 3rd ed. Ed., Wiley, New York, NY, 1985. C. S. Criddle, P. L. McCarty, M. C. Elliott and J. F. Barker, J. Contam. Hydrol., 1, 133-142 (1986). C. T. Jafvert and N. L. Wolfe, Environ. Toxicol. Chem., 6, 827-837 (1987). M. Reinhard, G. P. Curtis and M. R. Kriegman, Abiotic Reductive Dechlorination of Carbon Tetrachloride and Hexachloroethane by Environmental Reductants: Project Summary, EPA/600/S2-90/040, U.S. EPA, Washington, DC., 1990. D. W. Acton, Enhanced In Situ Biodegradation of Aromatic and Chlorinated Aliphatic Hydrocarbons in Anaerobic, Leachate-impacted Groundwaters, M. Sc. thesis University of Waterloo, Waterloo, Canada, 1990. R. W. Gillham and S. F. O’Hannesin, Ground Water, 32, 958-967 (1994). T. C. Wang and C. K. Tan, Bull. Environ. Contam. Toxicol., 45, 149-156 (1990).
23.1 Natural attenuation of chlorinated solvents
48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
1615
J. M. Gossett and S. H. Zinder, Symposium on Natural Attenuation of Chlorinated Organics in Groundwater, Dallas, TX, Sept. 11-13, 1996, U.S. EPA, EPA/540/R-96/509, 1996, pp. C. Holliger and W. Schumacher, Antonie Leeuwenhoek, 66, 239-246 (1994). C. Holliger, G. Schraa, A. J. Stams and A. J. Zehnder, Appl. Environ. Microbiol., 59, (1993). C. R. Smatlak, J. M. Gossett and S. Zinder, Environ. Sci. Tech., 30, 2850-2858 (1996). B. S. Ballapragada, H. D. Stensel, J. A. Puhakka and J. F. Ferguson, Environ. Sci. Tech., 31, 1728-1734 (1997). P. L. McCarty and L. Semprini in Handbook of Bioremediation, R. D. N. e. al, Ed., Lewis Publishers, Boca Raton, Florida, 1994, pp. 87-116. G. W. Sewell and S. A. Gibson, Environ. Sci. Tech., 25, 982-984 (1991). D. E. Fennell, J. M. Gossett and S. H. Zindler, Environ. Sci. Tech., 31, 918-926 (1997). F. H. Chapelle, P. B. McMahon, N. M. Dubrovsky, R. F. Fujii, E. T. Oaksford and D. A. Vroblesky, Water Resour. Res., 31, 359-371 (1995). C. Carr and J. B. Hughes, Environ. Sci. Tech., 30, 1817-1824 (1998). D. R. Lovley and S. Goodwin, Geochim. Cosmochim. Acta, 52, 2993-3003 (1988). D. R. Lovley, F. H. Chapelle and J. C. Woodward, Environ. Sci. Tech., 28, 1205-1210 (1994). P. L. McCarty, (1997), personal communication. E. Cox, E. Edwards, L. Lehmicke and D. Major in Intrinsic Bioremediation, R. E. Hinchee, J. T. Wilson and D. C. Downey, Ed., Battelle Press, Columbus, OH, 1995, pp. 223-231. C. Holliger, G. Schraa, A. J. Stains and A. J. Zehnder, Appl. Environ. Microbiol., 58, (1992). K. Ramanand, M. T. Balba and J. Duffy, Appl. Environ. Microbiol., 59, (1993). J. M. Suflita and G. T. Townsend in Microbial Transformation and Degradation of Toxic Organic Chemicals, L. Y. Young and C. E. Cerniglia, Ed., Wiley-Liss, New York, 1995, pp. 654 pp. S. Hartmans, J. A. M. de Bont, J. Tamper and K. C. A. M. Luyben, Biotechnol. Lett., 7, 383-388 (1985). S. Hartmans and J. A. M. de Bont, Appl. Environ. Microbiol., 58, 1220-1226 (1992). J. W. Davis and C. L. Carpenter, Appl. Environ. Microbiol., 56, 3878-3880 (1990). P. M. Bradley and F. H. Chapelle, Environ. Sci. Tech., 30, 553-557 (1998). G. Stucki, U. Krebser and T. Leisinger, Experientia, 39, 1271-1273 (1983). D. B. Janssen, A. Scheper, L. Dijkhuizen and B. Witholt, Appl. Environ. Microbiol., 49, 673-677 (1985). W. Reineke and H. J. Knackmuss, Eur. J. Appl. Microbiol. Biotechnol., 47, 395-402 (1984). G. Schraa, M. L. Boone, M. S. M. Jetten, A. R. W. van Neerven, P. J. Colberg and A. J. B. Zehnder, Appl. Environ. Microbiol., 52, 1374-1381 (1986). J. P. Spain and S. F. Nishino, Appl. Environ. Microbiol., 53, 1010-1019 (1987). J. A. M. de Bont, M. J. W. Vorage, S. Hartmans and W. J. J. van den Tweel, Appl. Environ. Microbiol., 52, 677-680 (1986). B. E. Haigler, S. F. Nishino and J. C. Spain, Appl. Environ. Microbiol., 54, 294-301 (1988). J. R. van der Meer, W. Roelofsen, G. Schraa and A. J. B. Zehnder, FEMS Microbiol. Lett., 45, 333-341 (1987). P. Sander, R. M. Wittaich, P. Fortnagel, H. Wilkes and W. Francke, Appl. Environ. Microbiol., 57, 1430-1440 (1991). S. F. Nishino, J. C. Spain and C. A. Pettigrew, Environ. Toxicol. Chem., 13, 871-877 (1994). J. C. Spain, Symposium on Natural Attenuation of Chlorinated Organics in Groundwater, Dallas, TX, Sept. 11-13, 1996, U.S. EPA, EPA/540/R-96/509, Washington, D.C., 1996, pp. F. Chapelle, Groundwater Microbiology and Geochemistry, Wiley, New York, 1993. W. D. Murray and M. Richardson, Crit. Rev. Environ. Sci. Technol., 23, 195-217 (1993). P. Adriaens and T. M. Vogel in Microbial Transformation and Degradation of Toxic Organic Chemicals, L. Y. Young and C. E. Cerniglia, Ed., Wiley-Liss, New York, 1995, pp. 654 pp. R. E. Miller and F. P. Guengerich, Biochemistry, 21, 1090-1097 (1982). P. M. Bradley and F. H. Chapelle, Environ. Sci. Tech., 30, 2084-2086 (1996). P. M. Bradley and F. H. Chapelle, Environ. Sci. Tech., 31, 2692-2696 (1997). T. M. Vogel and P. L. McCarty, Appl. Environ. Microbiol., 49, 1080-1083 (1985). E. J. Bouwer in Handbook of Bioremediation, R. D. N. e. al., Ed., Lewis Publishers, Boca Raton, FL, 1994, pp. 149-175. Remediation Technologies Development Forum, Natural Attenuation of Chlorinated Solvents in Groundwater Seminar, Class notes Ed., RTDF, 1997. J. W. Weaver, J. T. Wilson and D. H. Kampbell, EPA Project Summary, EPA/600/SV-95/001, U.S. EPA, Washington, D.C., 1995. M. P. Suarez and H. S. Rifai, Biorem. J., 3, 337-362 (1999).
1616
91 92 93 94 95 96 97 98
99
100 101 102
103 104 105 106
107
108 109
110 111
Hanadi S. Rifai, Charles J. Newell, Todd H. Wiedemeier
R. Oldenhuis, J. Y. Oedzes, J. van der Waarde and D. B. Janssen, Appl. Environ. Microbiol., 57, 7-14 (1991). J. E. Anderson and P. L. McCarty, Environ. Sci. Tech., 30, 3517-3524 (1996). A. R. Bielefeldt, H. D. Stensel and S. E. Strand in Bioremediation of Chlorinated Solvents, R. E. Hinchee, A. Leeson and L. Semprini, Ed., Battelle Press, Columbus, Ohio, 1994, pp. 237-244. B. Z. Fathepure and J. M. Tiedje, Environ. Sci. Tech., 28, 746-752 (1994). J. Gao, R. S. Skeen and B. S. Hooker in Bioremediation of Chlorinated Solvents, R. E. Hinchee, A. Leeson and L. Semprini, Ed., Battelle Press, Columbus, Ohio, 1995, pp. 53-59. L. Alvarez-Cohen and P. L. McCarty, Environ. Sci. Tech., 25, 1381-1387 (1991). U. Jenal-Wanner and P. L. McCarty, Environ. Sci. Tech., 31, 2915-2922 (1997). J. E. Keenan, S. E. Strand and H. D. Stensel in Bioremediation of Chlorinated and Polycyclic Aromatic Hydrocarbon Compounds, R. E. Hinchee, A. Leeson, L. Semprini and S. K. Ong, Ed., Lewis Publishers, Boca Raton, Florida, 1994, pp. 1-13. S.-C. Koh, J. P. Bowman and G. S. Sayler in Bioremediation of Chlorinated and Polycyclic Aromatic Hydrocarbon Compounds, R. E. Hinchee, A. Leeson, L. Semprini and S. K. Ong, Ed., Lewis Publishers, Boca Raton, Florida, 1994, pp. 327-332. S. E. Strand, G. A. Walter and H. D. Stensel in Bioremediation of Chlorinated Solvents, R. E. Hinchee, A. Leeson and L. Semprini, Ed., Battelle Press, Columbus, Ohio, 1994, pp. 161-167. M. E. Dolan and P. L. McCarty, Environ. Sci. Tech., 29, 1892-1897 (1995). D. Aronson and P. H. Howard, Anaerobic Biodegradation of Organic Chemicals in Groundwater: A Summary of Field and Laboratory Studies, SRC TR-97-0223F, American Petroleum Institute, Washington, D.C., 1997. F. Parsons, P. R. Wood and J. DeMarco, J. Am. Water Works Assoc., 76, 56-59 (1984). G. Barrio-Lage, F. Z. Parsons, R. S. Nassar and P. A. Lorenzo, Environ. Toxicol. Chem., 6, 571578 (1987). F. Parsons and G. Barrio-Lage, J. Am. Water Works Assoc., 77, 52-59 (1985). T. H. Wiedemeier, M. A. Swanson, D. E. Montoux, J. T. Wilson, D. H. Kampbell, J. E. Hansen and P. Haas, Symposium on Natural Attenuation of Chlorinated Organics in Ground Water, Dallas, TX, September 11-13, 1996, EPA/540/R-96/509, 1996, pp. 35-59. T. H. Wiedemeier, M. A. Swanson, D. E. Moutoux, E. K. Gordon, J. T. Wilson, B. H. Wilson, D. H. Kampbell, J. E. Hansen, P. Haas and F. H. Chapelle, Technical Protocol for Evaluating Natural Attenuation of Chlorinated Solvents in Groundwater, Air Force Center for Environmental Excellence, San Antonio, Texas, 1996. T. E. Buscheck and C. M. Alcantar in Intrinsic Remediation, R. E. Hinchee, J. T. Wilson and D. C. Downey, Ed., Battelle Press, Columbus, Ohio, 1995, pp. 109-116. C. E. Aziz, C. J. Newell, A. R. Gonzales, P. Haas, T. P. Clement and Y. Sun, BIOCHLOR Natural Attenuation Decision Support System User’s Manual, prepared for the Air Force Center for Environmental Excellence, Brooks, AFB, San Antonio, 1999. Y. Sun and T. P. Clement, Transport in Porous Media J., (1998). Y. Sun, J. N. Peterson, T. P. Clement and B. S. Hooker, Symposium on Natural Attenuation of Chlorinated Organics in Ground Water, Dallas, 1996, U. S. EPA, EPA/540/R-96/509, 1996, pp. 169.
23.2 Remediation technologies
1617
23.2 REMEDIATION TECHNOLOGIES AND APPROACHES FOR MANAGING SITES IMPACTED BY HYDROCARBONS Barry J. Spargo U.S. Naval Research Laboratory, Washington, DC, USA
James G. Mueller URS/Dames & Moore, Chicago, IL, USA
23.2.1 INTRODUCTION Subsurface contamination of soils and aquifers by chlorinated hydrocarbons (CHC) and non-chlorinated hydrocarbons (HC) is likely the largest environmental issue in industrialized nations worldwide. Decades without controlled disposal practices, inadequate storage and distribution systems, and accidental releases have resulted in a large number of contaminated drinking water and aquifer systems. In addition, an untold number of ecosystems are subject to future contamination by impinging hydrocarbon plumes. The extent of potential contributors ranges from neighborhood facilities, such as laundries or gas stations, to major fuel refineries, industrial operations and chemical manufacturing facilities. While characterization, control, and cleanup of these impacted areas may seem daunting, it is clear that not every impacted or potentially impacted area requires extensive remedial efforts. In fact, many impacts do not represent a significant risk to human health or the environment. In other areas, natural attenuation processes are effective in controlling the migration of the dissolved-phase plume (see Chapter 23.1). If, however, the presence of HCs elicits undesirable effects, then a number of strategies and developing remedial technologies can be used. This chapter will discuss these technologies and strategies and present a number of case studies documenting their effective implementation. 23.2.1.1 Understanding HC and CHC in the environment As summarized by Mueller et al.,1 hydrocarbons have been produced in the environment throughout geological time. Likewise CHCs are ubiquitous and are also of ancient ancestry.2 It follows, therefore, that microorganisms have developed mechanisms for utilizing these compounds as growth substrates. Depending on the inherent recalcitrance of the HC, biodegradation mechanisms may be associated with various abiotic degradation processes. While catabolic interactions are often complex and not fully elucidated, a more thorough understanding of integrated processes allows today’s scientists and engineers to design and implement more effective systems to mitigate situations where the concentration of HC in a given environment exceeds a desirable value. 23.2.1.2 Sources of HC in the environment Hydrocarbons found in the environment are of diverse structure and are widely distributed in the biosphere, predominantly as surface waxes of leaves, plant oils, cuticles of insects, and the lipids of microorganisms.3 Straight-chain HC, or alkanes, with carbon number maxima in the range of C17 to C21 are typically produced by aquatic algae. Conversely terrestrial plants typically produce alkanes with C25 to C33 maxima.4 Plants also synthesize aromatic HC such as carotenoids, lignin, alkenoids, terpenes, and flavenoids.5 Polycyclic
1618
Barry J. Spargo, James G. Mueller
aromatic hydrocarbons (PAH) are also of biogeochemical origin, formed whenever organic substances are exposed to high temperature via a process called pyrolysis. Here, the compounds formed are generally more stable than their precursors, usually alkylated benzene rings.6 The alkyl groups can be of sufficient length to allow cyclization and then, with time, these cyclic moieties become aromatized. The temperature at which this process occurs determines the degree of alkyl substitution. It follows, therefore, that fossil fuels such as coal and oil provide the largest source of mononuclear and polynuclear aromatic HC. Contemporary anthropogenic sources of HC in the environment thus originate from two primary sources: i) point source releases such as spills at industrial facilities which utilize large volume of fossil fuels, or ii) chronic lower-level inputs such as from atmospheric deposition. 23.2.1.3 Sources of CHC in the environment Chlorinated hydrocarbons are also abundant in nature.2 For example, it is estimated that 5×109 Kg of chloromethane are produced annually, mainly by soil fungi.7 However, CHCs of industrial origin perhaps represent an even greater contribution of CHC to the environment. These compounds include a variety of alkanes, alkenes, and aromatic compounds used principally as solvents and synthetic catalysts or intermediates. Reisch8 estimated that approximately 18 billion pounds of 1,2-dichloromethane are produced in the United States annually. More information on the production and distribution of CHCs can be found herein (see Chapter 3). While a majority of the CHC is used in a safe and conscientious manner, some material results in environmental contamination. This is often a result of accidental release, although improper disposal is also a common problem. Unfortunately, as a group, CHCs represent the most problematic of the environmental contaminants. This classification is based on their toxicity and environmental persistence. Thus effective means of remediating environments potentially impacted by CHCs is often necessary. 23.2.2 IN SITU BIOTREATMENT When the degree of impact or nature of contamination exceeds safe or acceptable conditions then environmental remediation may be proposed. When the contamination is confined and physically accessible then relatively quick and simple remedial efforts can be implemented. For example, impacted soils can be excavated and disposed in a safe manner. Other related remedial efforts have been reviewed and discussed previously.9-10 However, many CHC impacts are not easy to remediate because the point of impact, volume of release, and/or the magnitude of the problem are often not known. Moreover, the physical nature of CHC is such that they often concentrate in areas as non-aqueous phases. In these situations, a variety of in situ remediation and source management strategies have been developed. The potential benefits of these in situ approaches are many, with the more important features being that they are: non-invasive, applicable to large areas of impact, and usually represent the most cost-efficient remedial alternative. Various remedial approaches are presented below along with case studies summarizing their effective implementation. 23.2.2.1 Microbial-enhanced natural attenuation/bioremediation Remediation by monitored natural attenuation has been thoroughly reviewed in Chapter 23.1. In the environment, HCs are susceptible to a variety of physical, chemical, and microbiological transformation processes. Specifically, they can undergo biotransformation reac-
23.2 Remediation technologies
1619
tions under aerobic (presence of oxygen), hypoxic (low oxygen), and anaerobic (absence of oxygen) conditions. Examples of recognized biogeochemical reaction sequences are summarized in Figures 23.2.1 and 23.2.2. Some of these biological reactions are co-metabolic meaning that the microbes that catalyze them do not gain carbon or energy for growth and must therefore have a primary carbon source available to drive the processes. Our ability to understand and capitalize on biological processes, such as biotransformation, in in situ and ex situ strategies, often result in a low cost, simple alternative to conventional treatment strategies. Several methods to better understand these processes, including direct measure of microbial activity,11 transformation of tracer compounds into CO2 or metabolic intermediates,12 and microbial Figure 23.2.1. Aerobic degradation of fluorene. A typical aerobic utilization of specific carbon biodegradation processes, using fluorene as the example where the sources through stable isotopes compound is completely mineralized. Here, fluorene is utilized as a measurements13 have been develsole source of carbon and energy for microbial growth. Two pathoped and applied to a number of ways for fluorene degradation by Arthrobacter sp. Strain 101 as sugsites. gested by Casellas et al. [M. Casellas, M. Grifoll, J.M. Bayona, and hydrocarbon-impacted Other indirect measurements used A.M. Solanas, Appl Environ Microbiol, 63(3), 819 (1997)]. in the 1970’s and 1980’s such as plate counts have led investigators to believe that biodegradation was occurring, where direct measurements of biodegradation were not conducted to support those conclusions. Furthermore, the use of plate counts grossly underestimates the population of catabolically relevant biomass.1 Biodegradation likely occurs in most systems, however the level of biodegradation may be insufficient to expect reasonable cleanup to target levels in the desired time frames. The fact that HCs are amenable to aerobic biological treatment has been fully and convincingly established in the scientific literature (see also recent review1). In the absence of oxygen as an electron acceptor, microbially catalyzed reductive dehalogenation of CHC has been documented.14-19 Recently, Yang and McCarthy20 have demonstrated the reductive dechlorination of chlorinated ethenes at H2 tension too low to sustain competitive growth of hydrogenotrophic methanogens. Anaerobic biotransformation of non-chlorinated HCs typ-
1620
Barry J. Spargo, James G. Mueller
Figure 23.2.2. Anaerobic degradation of carbon tetrachloride. An example of anaerobic dehalogenation, using carbon tetrachloride as the model compound. In many cases, these reactions occur under cometabolic conditions meaning that an alternative growth substrate must be present to serve as an electron donor to drive the reduction reactions whereby carbon tetrachloride is used as the electron acceptor. Three known pathways for microbial degradation of carbon tetrachloride have been identified [U.E. Krone, R.K. Thauer, H.P. Hogenkamp, and K. Steinbach, Biochemistry, 30(10), 2713 (1991); C.H. Lee, T.A. Lewis, A. Paszczynski, and R.L. Crawford, Biochem Biophys Res Commun, 261(3), 562 (1999)]. These pathways are not enzymatically driven but rely on corrinoid and corrinoid-like molecules to catalyze these reactions.
ical of those prevalent in the dissolved phase, including PAH constituents, has also been demonstrated.21-28 It is therefore accepted that aerobic and anaerobic biodegradation of organic compounds occurs through the action of natural, indigenous microflora. As a result of these natural in situ microbial processes, many sites with elevated concentrations of biodegradable organics exhibit highly reducing and anaerobic conditions in areas containing elevated concentrations (i.e., suspected source areas). Moving outward laterally and down-gradient within the plume, the aquifers tend to become more oxidizing as a result of lower constituent levels, infiltration, and recharge with oxygenated water. 23.2.2.1.1 Case study - Cooper River Watershed, Charleston, SC, USA The Cooper River Watershed empties into the Charleston Harbor on the southern Atlantic coast of the United States. In the lower reaches, the Cooper River is a highly industrialized and urbanized watershed with storm sewer and surface run-off impact. The river supports industries such as a wood pulp processing plant, a former naval shipyard, and a chromium mining/processing facility. In addition a number of fossil fuel refineries, storage facilities,
23.2 Remediation technologies
1621
Figure 23.2.3. Bacterial productivity, a measure of bacterial activity, was compared with the level of contaminants at a number of site in the Cooper River watershed, South Carolina (USA). Where productivity is low and PAH concentration is high, the concentration of PAH or other mitigating factors suggest that the potential for PAH biodegradation may be limited in these systems compared to the other sampling points in this watershed. [Figure adapted from M. T. Montgomery, B. J. Spargo, and T.J. Boyd, Naval Research Laboratory, Washington, DC, NRL/MR/6115—98-8140, (1998)].
Figure 23.2.4. Microbial communities have the remarkable ability to adapt to utilize a number of carbon sources. Montgomery et al. suggest that the proportion of PAH degraders (shown as pie fraction) in a population can be reflected by their overall activity (measured by protein production) and their ability to mineralize the contaminant to CO2. [Figure adapted from: Montgomery, M. T. , T. J. Boyd, J. K. Steele, D. M. Ward, D. C. Smith, B. J. Spargo, R. B. Coffin, J. W. Pohlman, M. Slenska, and J. G. Mueller, International Conference on Wetlands & Remediation, Salt Lake City, UT, November 16-17, 1999.]
and commercial shipping and recreational docks are scattered along the river. The physical characteristics of the Cooper River have been documented.29 In 1997 a study was initiated to examine the capacity of the Cooper River to “self-remediate” if source input was reduced or eliminated from the system. Specifically, pressure was placed on the Charleston Naval Shipyard to dredge the sediments in the area adjacent to the Navy property. A site study was conducted in this region and showed an elevated number of hydrocarbons, heavy metals and other regulated compounds. The study was expanded to include the larger watershed and to place the contaminant levels in the context reflective of regional inputs of contaminants. Concomitantly, the “bio-capacity” of the sediments, overlying
1622
Barry J. Spargo, James G. Mueller
boundary layer, and water column of the Cooper River were examined. Figures 23.2.3-23.2.4 illustrate the overall biological activity of the ecosystem and the capacity to degrade specific HCs. In this case, an argument can be made that the system with limited non-point source HC input will self-remediate to acceptable levels in an reasonable time frame. However, several specific sites (Pulp Mill, Anchorage, Degas Plant, Shipyard Creek), where other contributing factors impede the microbial turnover rates of HC, are candidates for dredging or other technology. It was concluded that source control coupled with long term monitoring and strategic management of this ecosystem is a cost effective alternative to disruptive dredging or high-cost high technology approaches. The impact of these remediation approaches on adjacent, less-impacted ecosystems are compelling factors for avoiding their implementation. 23.2.2.2 Phytoremediation Phytoremediation is the use of higher plants in order to contain, sequester, reduce, or degrade soil and groundwater contaminants for the eventual closure of hazardous waste sites. This rapidly emerging technology can be applied to a diverse range of environmental conditions and contains many potential advantages over conventional remediation technologies; such as substantially lower costs, improved safety, better aesthetics, and wider public acceptance. There are at least three areas where phytoremediation per se can be utilized to treat soil or groundwater impacted by HCs and related compounds or co-constituents of interest such as heavy metals: 1) Rhizosphere-Enhanced Phytoremediation: Plants are used to stimulate the relevant catabolic activities of indigenous, root-colonizing soil microorganisms which results in enhanced remediation of soils (and potentially groundwater) impacted by HCs; 2) Phytoextraction: Specially selected plants are utilized to hyperaccumulate inorganic materials such as salts, heavy metals, trace elements, radionuclides, and naturally occurring radioactive materials (NORM), and; 3) Plant-Based Hydraulic Containment: Entails the use of the natural water uptake and transpiration ability of highly transpiring, specially selected trees or plants for either surficial or groundwater hydraulic control. As summarized in the schematic (Figure 23.2.5), these phytoremediation processes often occur simultaneously. Here, water uptake and evapotranspiration serves to transport CHC and other solutes through the plant as it partakes in the normal physiological mechanisms of plant life. Successful application of phytoremediation technology thus requires a thorough understanding of agronomy, plant physiology, biochemistry, and soil sciences. 23.2.2.2.1 Case study - phytoremediation for CHCs in groundwater at a chemical plant in Louisiana URS/Radian proposed the use of phytoremediation to treat groundwater contaminated with dissolved phase CHC at a chemical plant. The State of Louisiana regulatory agency recommended standard pump & treat technology. However, Radian was able to convince the State to use a new more cost effective remedy. Hybrid Poplar trees were planted to achieve hydraulic containment and phytoextraction of the entire dissolved plume. The shallow groundwater and long growing seasons were ideal for this remedial approach. Radian was eventually able to close the site and obtain a no further action letter from the State of Louisiana.
23.2 Remediation technologies
1623
Figure 23.2.5. Phytoremediation: transport modes for HC and CHC in plant systems. Transformation of HC and CHC can be found in the root system with root-associated microbes and tissues of many plant species. HC and CHC are transported by normal plant physiological processes, such as water uptake and evaporation and transpiration. [Schematic courtesy M.A. Bucaro.]
23.2.3 IN SITU TREATMENT TECHNOLOGIES 23.2.3.1 Product recovery via GCW technology Vertical groundwater circulation wells (GCW) create three-dimensional, in situ groundwater circulation cells to mobilize and transport dissolved phase constituents of interest from the aquifer to a central well for treatment via a number of biotic and/or abiotic processes. GCW technology relies on a positive- or negative-pressure stripping reactor in a specially adapted groundwater well. Often pressure is attained by an above-ground mounted blower and off-air treatment system, such as activated carbon. A generic GCW is shown in Figure 23.2.6. The basic principle of operation of one GCW technology depends on moving water within the well to a well screen area above the water table, where water cascades into the vadose zone surrounding the well. Water is drawn in at a screened area in the aquifer usually found below the contaminated zone creating vertical water circulation. Volatile organics are stripped from the dissolved phase by air-stripping and biodegradation is enhanced. For
1624
Barry J. Spargo, James G. Mueller
Figure 23.2.6. GCW process diagram. Effective hydrocarbon stripping in the water column is observed in these systems using a vacuum extraction. A circulation cell is created by directional flow of water in the vertical direction creating a capture zone extending several meters from the well. In addition a bioreactor (high surface area bacterial biofilm) can be used in the system to degrade low volatile contaminants [Adapted from Bernhartt et al., U.S. Patent 5,910,245, 1999]
more information on other GCW systems and their various modes of operation see Mueller et al.30 and Allmon et al.31 Modifications of the technology have evolved as promising solutions for in situ remediation, source management, and/or accelerated recovery of phase-separated hydrocarbons. In each of these applications, however, success requires that the GCW zone of influence (ZOI) be validated. Accurate information on the direction and velocity of groundwater flow is equally important because it determines the rate of transport of contaminants being released or treated in the sub-surface. Toward this end, pressure transducers, conservative dye tracers, and in situ permeable flow sensors have been used to validate theoretical predictions of flow fields and the corresponding ZOI. The resulting information is then used to reassess the model and make operational changes to the system in order to meet clean-up goals and objectives. 23.2.3.1.1 Case study - GCW recovery of creosote, Cabot/Kopper’s Superfund Site, Gainesville, FL This site is a former pine tar and charcoal generation facility and a active wood treatment facility. Creosote used in the wood treating operation is the primary HC. The soil is 93% sand with some silt and clay. Remedial investigation results indicated that ground water in the shallow aquifer (10 to 23 ft below ground) had been impacted. Horizontal and vertical con-
23.2 Remediation technologies
1625
ductivities were 9×10-3 cm/s and 9×10-4 cm/s, respectively, with a horizontal gradient of 0.006. Initial total concentrations of PAHs in the soil exceeded 700 mg/kg. Total concentrations of PAHs in groundwater for all wells tested ranged from 5-50 mg/L. A single GCW well was installed at a depth of 3 to 8.5 m immediately down gradient from a lagoon area that had been identified as a source of creosote constituents. The system was started in February 1995 and continues to operate to date, having been taken over by the client in 1998. Samples taken after 18 months of operation indicated total PAH concentrations of 10-35 mg/L in up gradient wells. Concentrations in down gradient wells were measured at 0.04-2 mg/L indicating a marked reduction in HC in the aquifer as groundwater moved down gradient through the GCW circulation cell. 23.2.3.2 Surfactant enhanced product recovery A possible remedy for CHC dense non-aqueous phase liquid (DNAPL) is to install an increased-efficiency pump-and-treat system based on introducing surfactants into the aquifer to increase the solubility of CHC and the rate at which it transfers into the water phase. In this type of system, groundwater is extracted, DNAPL is separated (if present), dissolved CHC is air-stripped or steam-stripped from the water, surfactant is added to the groundwater, and the surfactant-rich water is re-injected into the aquifer up-gradient of the suspected DNAPL deposit. As the surfactant-laden groundwater passes across the DNAPL zone it is capable of reaching a CHC saturation level that is many times the natural CHC solubility, thus removing DNAPL more efficiently. Surfactant-aided product recovery differs significantly from surfactant flushing approaches in that the goal of the efforts is to increase the dissolution of DNAPL into the aqueous phase to expedite its removal. The goal is not to physically mobilize DNAPL through the addition of high concentrations of surface-active agents (i.e., surfactant flushing). In general, surfactant flushing per se is not advocated unless the geophysical properties of the aquifer are extremely well characterized, and the nature and source of CHC impact are well defined. Since these requisites are rarely met, the more aggressive use of surfactants is rarely considered. 23.2.3.2.1 Case study - Surfactant-aided chlorinated HC DNAPL recovery, Hill Air Force Base, Ogden, Utah Hill Air Force Base at Ogden, Utah used large amounts of solvents as degreasing agents. From 1967 to 1975, unknown amounts of perchloroethene (PCE), trichloroethene (TCE), 1,1,1-trichloroethane (TCA), and dichloromethane (MC) were placed in shallow, unlined trenches as the means of disposal. In the mid-1980s, pools of CHC DNAPL were found at the base of the uppermost aquifer. These DNAPL pools were several feet thick in some areas and extended over 36 acres. As such, DNAPL was a significant and continuing source of off-site migration of dissolved-phase CHCs. As an interim remedial action, the Air Force designed and implemented conventional pump-and-treat technology to serve as a “Source Recovery System”. In its first year of operation, the system recovered over 23,000 gallons of DNAPL. After several years of operation, DNAPL recovery decreased but thousands of gallons of product remained in the form of residual saturation (on the order of 20 percent of the pore volume). This residual material was not recoverable by normal pump-and-treat methods. In an effort to enhance the recovery of residual CHC DNAPL, URS/Dames & Moore initiated a surfactant-aided DNAPL recovery system. About 8 percent sodium dehexyl sulfosuccinate (an anionic surfactant), 4 percent isopropanol and 7,000 mg/L NaCl
1626
Barry J. Spargo, James G. Mueller
were combined to create an average DNAPL solubility of 620,000 mg/L (compared with a TCE solubility of 1,100 mg/L in natural groundwater). The solution was injected over a period of time, in an amount equal to 2.4 pore volumes in the test portion of the aquifer. DNAPL removal was 99 percent (estimated) and surfactant recovery at the extraction wells was 94 percent. Based on these results, a full-scale system was designed and implemented. 23.2.3.3 Foam-enhanced product recovery The use of foams to remove heavy immiscible fluids such as DNAPL from soil was developed by the petroleum industry for crude oil production. Subsequently, The Gas Research Institute developed the use of foams to release and mobilized DNAPL contaminants in the subsurface. Coupled with in situ or ex situ bioremediation, foam-enhanced product recovery can, potentially, transport CHC contaminants upward in the groundwater, thus reducing the potential for driving the contamination to previously non-impacted areas. The use of foam for CHC DNAPL is currently viewed as experimental. The delivery of the foam, its sweep front, the foam stability, its ability to release CHC DNAPLs in the subsurface, and the resultant biodegradability of residuals can potentially be aided through the proper selection of foaming agents and nutrients. In theory, the technology can tailor the foam system to aerobic or anaerobic subsurface environments, depending on the selection of the carrier gas. This allows adequate biodegradation for the particular CHC in the foam-pollutant system. For example, DCE can be biodegraded aerobically, whereas PCE needs to be degraded anaerobically. 23.2.3.4 Thermal desorption - Six Phase Heating Six-Phase HeatingTM (SPH) is a polyphase electrical technology that uses in situ resistive heating and steam stripping to achieve subsurface remediation. The technology was developed by Battelle’s Pacific Northwest Laboratories for the U.S. Department of Energy to enhance the removal of volatile contaminants from low-permeability soils. The technology is also capable of enhancing the removal of DNAPLs from saturated zones. SPH uses conventional utility transformers to convert three-phase electricity from standard power lines into six electrical phases. These electrical phases are then delivered throughout the treatment zone by steel pipe electrodes inserted vertically using standard drilling techniques. Because the SPH electrodes are electrically out of phase with each other, electricity flows from each electrode to the adjacent out-of-phase electrodes. In situ heating is caused by resistance of the subsurface to this current movement. In this manner, a volume of subsurface surrounded by electrodes is saturated with electrical current moving between the electrodes and heated. By increasing subsurface temperatures to the boiling point of water, SPH speeds the removal of contaminants such as CHCs via three primary mechanisms: increased volatilization, steam stripping, and enhanced residual mobility toward extraction wells via viscosity reduction. Once subsurface soil and groundwater reach the boiling point of water, the in situ production of steam begins. Through preferential heating, SPH creates steam from within silt and clay stringers and lenses. As this steam moves towards the surface, it strips contaminants such as CHCs from both groundwater and soil matrix. Released steam can act as a carrier gas, sweeping CHC out of the subsurface and to extraction wells. However, it can also cause constituent migration and CHC displacement if the steam is allowed to condense prior to extraction.
23.2 Remediation technologies
1627
23.2.3.4.1 Case study - Six-Phase Heating removal of CHC at a manufacturing facility near Chicago, IL SPH has been employed to remove TCE and TCA from the subsurface at a former manufacturing facility near Chicago, Illinois.32 Since 1991 combined steam injection with both ground water and soil vapor extraction had resulted in significant mass removal, but had left behind three hot spot areas after seven years of operation. These areas, which contained DNAPL in tight heterogeneous soil, were treated in less than four months by SPH. Site lithology consisted of heterogeneous sandy silts to 18 ft below grade (bg) and a dense silty clay till from 18-55 ft bg. A shallow groundwater table was encountered at 7 ft bg and hydraulic conductivity through the remediation zone ranges from10-4 - 10-5 cm/sec. A network of 107 electrodes covering two-thirds of an acre was established. To treat beneath a warehouse, 85 of those electrodes were constructed directly through the floor of the building. Electrically conductive from 11-21 ft bg, the electrodes actively heated the depth interval from 5-24 ft bg. Once subsurface temperatures reach boiling, steam laden with chlorinated solvents was collected by a network of 37 soil vapor extraction wells screened to 5 ft bg. SPH operations began on June 4, 1998. Within 60 days, temperatures throughout the entire 24,000 cubic yard treatment volume had reached the boiling point of water. With another 70 days of heating, separate phase DNAPL had been removed and TCE/TCA groundwater concentrations reduced to below the risk based target cleanup levels. Cleanup results are shown in Table 23.2.1. Table 23.2.1. Summary of groundwater cleanup results Well
Compound
Jun. ‘98, µg/l
Oct. ‘98, µg/l
Reduction, %
B-3
TCE TCA
58,000 82,000
790 non detect
98.6 >99.4
Da2
TCE TCA
370,000 94,000
8,800 290
97.6 99.7
F13
TCE TCA
2,800 150,000
280 non detect
90.0 >99.9
In 100 days of heating, 23,000 cubic yards of DNAPL impacted subsurface were remediated to the site cleanup goals set by a State RBCA Tier III evaluation. Based upon these results, the site owner has elected to continue SPH to reach the lower cleanup goals to lessen long term liability. SPH preferentially heats subsurface zones with higher electrical conductivity. At this site, these zones included clay-rich soil lenses and areas with elevated chloride ion concentrations. DNAPL are typically trapped in silt and clay-rich stringers and lenses, while locations of elevated chloride ion concentrations, created from the biological dehalogenation of chlorinated solvents, also correspond to locations of elevated DNAPL concentrations. Thus, SPH targeted the specific subsurface locations of the DNAPL mass. Calculations of costs included project permitting, preparation of work plans, installation and operations of the SPH, vapor extraction, air abatement, and condensate treatment systems, electrical use, waste disposal, and interim sampling and reporting. Final demobilization, sampling, and reporting were not in the costing calculations. As of 20 November 1998, remedial costs of SPH were estimated at $32 per cubic yard of treatment area. At this time 1,775 MW-hr of electrical energy were consumed, representing an electrical usage rate
1628
Barry J. Spargo, James G. Mueller
of $14,000 per month plus $40 per MW-hr for an electrical cost of $148,000 or $6.41 per cubic yard of treatment volume (personal communication, Greg Smith, URS/Radian). 23.2.3.5 In situ steam enhanced extraction (Dynamic Underground Stripping) Dynamic Underground Stripping (DUS), developed by Lawrence Livermore National Laboratory (LLNL) in Livermore, California and the College of Engineering at the University of California at Berkeley, is a combination of the following technologies: 1) Steam injection at the periphery of the contaminated area to heat permeable zone soils, vaporize volatile compounds bound to the soil, and drive the contaminants to centrally located vapor/groundwater extraction wells; 2) Electrical heating of less permeable clays and fine-grained sediments to vaporize contaminants and drive them into the steam zone; 3) Underground imaging, primarily Electrical Resistance Tomography (ERT) and temperature monitoring, to delineate the heated area and track the steam fronts to insure plume control and total cleanup; and 4) Vapor and steam extraction followed by treatment of effluent vapors, NAPL, and impacted groundwater before discharge. DUS is potentially effective for material above and below the water table, and is also potentially suited for sites with interbedded sands and clay layers. DUS raises the temperature of the soil and groundwater leading to rapid removal of the contaminants due to the thermodynamic processes discussed above. 23.2.3.6 In situ permeable reactive barriers (funnel and gate) In situ permeable reactive barriers are used to convert CHC to less toxic and biodegradable intermediates using zero-valent metals such as iron. Permeable reactive barriers, primarily developed at the University of Waterloo, Groundwater Research Center, Canada and EnviroMetals Technologies, Inc. offer a unique cleanup option which does not require transport of contaminated materials (e.g., soil or groundwater) to the surface. Groundwater in the contaminate site can be directed to a permeable barrier region (usually through the use of non-permeable barriers) which is the reactive cell composed granular zero-valent iron. The thickness of the cell is based on the retention time (resident time of water within the cell, based on horizontal flow velocities), the ratio of granular iron to sand/pea gravel, and the types of contaminants. However, in the case study described below, 100% granular iron was used as an added safety factor to ensure complete transformation of the CHCs. Degradation of CHC occurs through a reduction of iron. This is fundamentally an iron metal corrosion event, where elemental iron is converted to ferrous iron in the presence of water and hydroxyl ions. When dissolved oxygen or the oxygen tension of the surrounding groundwater is low, reactive hydrogen is produced, resulting in reductive dehalogenation of the CHC species, as shown: Feo + X-Cl + H2O → X-H + Cl- + OHUsing permeable reactive barriers, investigators have shown virtually complete dechlorination of CHCs, such as TCE to ethene or ethane (for review see33). 23.2.3.6.1 Case study - CHC remediation using an in situ permeable reactive barrier at Naval Air Station Moffett Field, CA In late 1995, an in situ permeable reactive barrier demonstration at Naval Air Station Moffett Field near Mountain View, CA was constructed by URS/Dames & Moore. The pri-
23.2 Remediation technologies
1629
Figure 23.2.7. Schematic of a permeable reactive barrier. [From Permeable Reactive Barrier Technologies for Contaminant Remediation, EPA/600/R-98/125, 1998].
mary groundwater contaminants were TCE, PCE and cis-1,2-DCE (cDCE). The lithology of the site has been characterized as alluvial-fluvial clay, silt, sand and gravel, with an aquifer extending 5 to 60 ft below ground surface. The lithology is complex in this region and separates the aquifer into two zones with a discontinuous semi-confining aquitard. The mixed TCE, PCE, cDCE plume was ca. 10,000 ft by 5,000 ft extending along the direction of groundwater flow. Prior to installation of the permeable barrier system, TCE levels exceeded 5,000 µg/L, PCE 1,000 µg/L. A permeable reactive barrier was constructed with the dimensions 10 ft wide, 6 ft long, and 25 ft deep with a 2 ft pea gravel layer in front of and behind the cell. The cell was filled with granular zero-valent iron and a steel corrugate wall was constructed on each of the sides of the cell to form a funnel redirecting groundwater flow in that area through the iron cell. After nearly 4 years of operation, monitoring wells down gradient of the permeable barrier continue to shown non-detect for CHCs. 23.2.4 CONCLUSIONS There are a number of technologies and strategies for managing hydrocarbon remediation. We have reviewed a number of the more promising, simple, and perhaps higher-technology solutions. However, a number of other demonstrated (pump and treat, air sparging) and emerging technologies do exist.34 The basic themes of preventing migration passively, and removal of contaminant via degradation (biotic and abiotic) are seen in all the technologies. The proper use of particular technologies and strategies is very dependent on the extent and type of contamination, the site characteristics (hydrology, lithology, etc), the cleanup goals, and applicable regulations to name a few. Unfortunately, there are no hard and fast rules for the use of a particular technology; past experience suggests that the more that is known about site characteristics the greater the success of technology application.
1630
Barry J. Spargo, James G. Mueller
REFERENCES 1 2 3 4 5 6 7 8 9 10 11
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
J.G. Mueller, C.E. Cerniglia, and P.H. Pritchard. in Bioremediation: Principles and Applications, R.L. Crawford and D.L. Crawford, Ed., Cambridge University Press, Cambridge UK, 1996, pp. 125. L.P. Wackett, in Bioremediation: Principles and Applications, R.L. Crawford and D.L. Crawford, Ed., Cambridge University Press, Cambridge UK, 1996, pp. 300. F.J. Millero and M.L. Sohn, Chemical Oceanography, CRC Press, Boca Raton, FL, 1991. M. Nishmura and E.W. Baker, Geochemica et Cosmochimica Acta, 50, 299 (1986). D.J. Hopper, in Developments in Biodegradation of Hydrocarbons, R.J. Watkinson, Ed., Applied Science Publishers, London., 1978, pp. 85. M. Blumer, Scientific American, 234, 35, (1976). R.A. Rasmussen, M.A.R. Kahlil, and R.W. Dalluage, J. of Geophysical Research, 85, 7350, (1980). M.S. Reisch, Chemical and Engineering News, 72(15), 12, (1994). R.D. Norris, Handbook of Bioremediation, Lewis Publishers, Boca Raton, FL, 1994. P.E. Flathman, E.J. Douglas and J.H. Exner, Bioremediation: Field Experience, Lewis Publishers, Boca Raton, FL, 1994. M.T. Montgomery, T.J. Boyd, B.J. Spargo, J.G. Mueller, R.B. Coffin, and D.C. Smith, in In Situ Bioremediation and Efficacy Monitoring, B.J. Spargo, Ed., Naval Research Laboratory, Washington, DC, NRL/PU/6115—96-317, 1996, pp. 123. T.J. Boyd, B.J. Spargo, and M.T. Mongtomery, in In Situ Bioremediation and Efficacy Monitoring, B.J. Spargo,Ed., Naval Research Laboratory, Washington, DC, NRL/PU/6115—96-317, 1996, pp. 113. C.A. Kelley, B.A. Trust, and R.B. Coffin, Envir. Sci. Technol., 31, 2469 (1997). D.F. Berry, A.J. Francis and J.-M. Bollag, Microbiol. Rev., 51, 43, (1987) E. Arvin, B. Jensen, E.M. Godsy and D. Grbic-Galic, in: International Conference on Physiochemical and Biological Detoxification of Hazardous Wastes, Y.C. Wu (Ed.), 1988, p828. M.D. Mikesell and S.A. Boyd, Appl. Env. Microbiol., 52, 861, (1986). S.L Woods, J.F. Ferguson, and M.M. Benjamin, Env. Sci. Technol., 23, 62, (1989). A.J. Frisbie and L. Nies, Bioremediation Journal, 1, 65, (1997). J. Dolfing and J.M. Tiedje, FEMS Microbiol. Ecol., 38, 293, (1986). Y.Yang and P.L. McCarthy, Environ Sci. Technol., 32, 3591, (1999). E.M. Godsy and D.F. Goerlitz. U.S. Geological Survey Water-Supply Paper 2285, (1986). J.R. Mihelcic, and R. G. Luthy, Appl. Environ. Microbiol., 54, 1182, (1988). J.R. Mihelcic and R.G. Luthy, Appl. Environ. Microbiol., 54, 1188, (1988). W.R. Mahaffey, UPRR, In Situ Treatment Process Development Program, Milestone II Report, Volume 3/ 4, (1988). B. Al-Bashir, T. Cseh, R. Leduc, and R. Samson, Appl. Microbiol. Biotechnology, 34, 414, (1990). J. Thierren, G.B. Davis, C. Barber, B.M. Patterson, F. Pribac, T.R. Power and M. Lambert, Groundwater, 33, 469, (1995). A.A.M. Langenhoff, A.J.B. Zehnder and G. Schraa, Biodegradation, 7, 267, (1996). D. L. McNally, J.R. Mihelcic and D. R. Lueking, Environ. Sci. Technol., 32(17), 2633, (1998). R.F. VanDolah, P.H. Wendt, and E.L. Wenner, South Carolina Coastal Council, NA87AA-D-CZ068. (1990) J. Mueller, M. Ohr, B. Wardwell, and F. Lakhwala, Soil & Groundwater Cleanup, Oct/Nov, 8, (1999). W.E. Allmon, L.G. Everett, A.T. Lightner, B. Alleman, T.J. Boyd, and B.J. Spargo, Naval Research Laboratory, Washington, DC, NRL/PU/6115—99-384, (1999). B. Trowbridge, V. Jurka, G. Beyke and G. Smith, in Second International Symposium on Remediation of Chlorinated and Recalcitrant Compounds, Battelle Press, Columbus, OH, 2000. U.S. EPA, In Situ Remediation Technology Status Report: Treatment Walla, EPA 542-K-94-004, Washington D.C., 1995. S.K. Sikdar and R.L. Irvine, Bioremediation: Principles and Practice, Battelle Press, Columbus OH, 1997.
24
Protection George Wypych ChemTec Laboratories, Inc., Toronto, Canada
24.1 GLOVES Many types of gloves are manufactured for different purposes. Proper gloves selection is important to ensure that the solvent or other chemical compound cannot penetrate through the glove where its subsequent evaporation would be prevented by the glove itself. Exposure to solvent entrapped between the glove and the skin is more severe than the exposure to solvent without glove. The first requirement in the selection of a glove is that the materials from which gloves are produced are not dissolved or swollen by solvent. This alone is not sufficient. The other important requirement in glove selection is permeability of all of the materials in its structure. Permeability is measured according to ASTM Standard F 739 or European Standard EN 374. Permeability results are usually expressed in mg of permeate which penetrates 1 m2 of material per minute. Good materials for gloves should have permeability below 1 mg m-2 min-1. Materials having permeability in the range from 1 to 10 mg m-2 min-1 may still be considered as suitable materials. Relative to permeability, materials are rated by a five point scale of the permeation index number. The best choice of glove material has a permeation index number = 0. If the material has index 1 (permeability in the range from 1 to 10 mg m-2 min-1), the glove can be eventually accepted for harmful chemicals. Any higher index number makes justification for its use more difficult and in plain language such a glove should not be used with harmful chemicals. Breakthrough time is another measure of glove fitness for purpose. This parameter measures the time in minutes during which a measurable amount of a particular chemical compound penetrated through a particular garment. Specialized monographs give data on breakthrough times for a large number of gloves and other protective clothing and numerous penetrants.1,2 These sources1,2 are the best collections of data which should be considered in glove selection. There is a relationship between permeation rate (and thus permeation index number) and breakthrough time. For practical purposes gloves are selected based on the assumption that they should resist penetrant breakthrough for more than eight hours. Such gloves are generally considered to have permeability index number = 0. In addition to the properties of protective clothing materials, the toxicity of the penetrant should be considered. For less toxic solvents, materials with a permeation index
1632
George Wypych
number of 1 or even 2 may still be cautiously considered but will be less acceptable for more toxic penetrants. Monitoring the quality of protection during use is not less important. Materials are not permanent. Frequent inspection of gloves is therefore very important. Any mechanical damage will compromise protection properties. The presence of cuts, cracks, or holes immediately renders the gloves useless. Gloves should be inspected for changes of color, texture, or hardening which are signs of material degradation. It should be borne in mind that protective clothing is not universal for all chemical materials. For this reason, protection against one penetrant does not suggest that the glove has protecting qualities against the other materials. Mixtures of solvents behave in a different manner than do pure components. The temperature at which exposure occurs is also an important factor. Generally as temperature increases, permeability increases. Table 24.1 gives general guidance regarding the applicability of different glove materials for different groups of solvents. Note that glove materials are not equal. They vary with the formulation used by a specific manufacturer and the thickness of the protective layer. For each selection, the manufacturer’s specification should be consulted to estimate previously discussed parameters. The best practice requires that gloves are tested for particular conditions of their use. The other good option is to consult results of measurement included in the specialized monographs.1,2 Table 24.1. General guidelines for selection of gloves for different groups of solvents Solvent group
Suitable glove materials
Manufacturer
Acids
butyl neoprene Saranex-23
North and many other Ansell, Best, and many other DuPont
Aldehydes
butyl PE/EVAl/PE*
Best, North and many other Safety4
Amines
PE/EVAl/PE*
Safety4
Esters
PE/EVAl/PE* PVAl
Safety4 Edmont
Ethers
PE/EVAl/PE* PVAl
Safety4 Edmont
Halogenated hydrocarbons
PE/EVAl/PE* PVAl Viton
Safety4 Edmont North
Hydrocarbons, aliphatic
nitrile PE/EVAl/PE* Viton
Best, Mapa-Pioneer, North Safety4 Best, North
Hydrocarbon, aromatic
PE/EVAl/PE* PVAl Viton
Safety4 Anselledmont, Comasec North
24 Protection
Solvent group
1633
Suitable glove materials
Manufacturer
Ketones
butyl PE/EVAl/PE* PVAl
Best, Comasec, Guardian, North Safety4 Anselledmont, Edmont
Monohydric alcohols
butyl nitrile neoprene
Best, Guardian, North Best, Comasec Ansell, Best, Mapa-Pioneer
Phenols
butyl neoprene Syranex-23
Best, Comasec, North Anselledmont, Mapa-Pioneer DuPont
Polyhydric alcohols
neoprene nitrile rubber
many manufacturers
*PE/EVAl/PE = polyethylene/ethylene vinyl alcohol/polyethylene
The above suggested selections are based on the test data included in the monograph.1 Only the gloves which have a permeability index number = 0 were included. However, since solvents in each group frequently have diverse physical properties, in some specific cases the selection for that group may not be the most suitable for a particular member of the group. It should be also considered that some manufacturers have several grades of gloves in the same generic group of polymer therefore their specification should always be consulted.
24.2 SUIT MATERIALS The selection of suit material is more complex because not only physical properties of the components but also the design of suit and structure of laminate used will determine its barrier properties. As a whole, these are usually proprietary and difficult to compare. The information in Table 24.1 can be used for general guidance but the industrial hygienist should make a full evaluation of the performance of the protective clothing in particular conditions and application.
24.3 RESPIRATORY PROTECTION The selection of a respirator, general suggestions for its use, the adsorption capacity of filters, and breakthrough time are reviewed below. The selection of respirator is based on the nature of respiratory hazards such as permissible exposure limits, established concentration immediately dangerous to life, actual concentration of hazardous substances, and amount of oxygen. When using filters, the ambient air must contain at least 20% oxygen at sea level. Filter protection should not be used for unknown contaminants and contaminants that are immediately dangerous to life and health. Permissible exposure limits and actual concentration determine (among other factors) the breakthrough time as discussed below. Only approved respirators should be used. Each country has a body which can approve respirators for use. In the USA, respirators are approved jointly by the National Institute for Occupational Safety and Health (US Department of Health and Human Services) and the Mine Safety and Health Administration (Department of Labor).
1634
George Wypych
In addition to the the above parameters determining respirator selection, the following factors are also considered: nature of operation process, location of hazardous area (especially in respect to supply of respirable air), employee activity and intensity of effort required to perform the work (determines the amount of air which must be supplied to lungs), physical characteristics and limitations of various respirators. The selection of respirator must be done with a full consideration of these many factors which require specialized knowledge going beyond this discussion. Further information on this subject can be found in the specialized monograph.3 Respirators can be generally divided into two main groups: air-purifying respirators and atmosphere-supplying respirators. The first group is more common and for this reason will be discussed in more detail below. Air-purifying respirators are still divided into non-powered and powered and these are further divided into gas/vapor and particle removing. Our main interest here is given to vapor-removing, non-powered respirators which are the most common in industry and laboratories in solvent applications. Two additional subjects are integral part of good protection: respirator fit testing and training of users. Non-powered, vapor-removing respirators are also termed as negative pressure respirators. This means that respirator is under positive pressure during exhalation and under negative pressure during inhalation. If the respirator does not fit the user properly, some air from the surroundings will be drawn into the respirator during inhalation because of leaks. This will result in a diminishing efficiency of protection. For this reason, each user should be given help from the employer in selecting the most suitable type and size of respirator for the particular individual. In addition, the reasons for fitting should be clearly explained so that they are fully understood by the employee. It is only the employee who may later assure that the respirator is used properly and this will depend on his/her full understanding of the principles. The first matter of importance is the rate of breathing under various conditions.The amount of air we breathe depends on our energy requirements or more precisely on the intensity with which we expand energy. If no work is performed, the average human being requires about 10 l of air per minute. The rate of breathing increases with the intensity of work (light work 20-30, heavy work 70-100 l/min). The use of a filter reduces air flow rate due to the pressure drop and many respirators cannot cope with high rates of flow which becomes the one important limitation of negative pressure respirators that must be evaluated during the selection process. The pressure drop in the filter depends on its design. Chemical cartridges used for solvent adsorption are filled with activated carbon or activated charcoal which are amorphous forms of carbon characterized by their ability to rapidly absorb many gases and vapors. The carbon is obtained by pyrolysis of wood, nutshells, animal bones, petroleum products, and other carbonaceous materials. Activated carbon for respirators usually comes from coconut shells or coal after activation at temperature in the range from 800 to 900oC. It has a porous structure with a surface area of 10,000 m2/g. On the one hand, maximizing surface contributes to increased capacity on the other it may lead to an increased pressure drop during breathing (depending on particles shapes and sizes). It is thus important to consider both capacity of the filter and its pressure drop. The adsorption of vapor is a physical process which can be reversed. Desorption of vapor does not require a high energy and is equilibrium driven (equilibrium relative to the concentration of vapor in surrounding atmosphere and distribution of adsorbed vapor in
24 Protection
1635
activated carbon). Desorption during storage or non-use time can result in migration of previously adsorbed molecules. This phenomenon is very important for effective use of respirators. On one hand, it precludes the possibility of cartridge recycling. On the other hand, it imposes restriction on cartridge use. For good performance the mask should be worn 100% of the time. It is estimated that removal of respirator for 5 min in an 8-hour day causes a loss of 50% of the expected protection. It is also important to limit filter use to one shift even if it were not fully exhausted in that time. The absorption capacity is calculated from the following equation: 2 BT 2 ps W = ρW o exp − log p β
[24.1]
where: W ρ Wo B T β ps p
adsorption capacity per unit weight of carbon solvent density total volume of adsorption space microporosity constant of carbon temperature affinity coefficient of solvent vapor for carbon saturated vapor pressure of solvent at temperature T equilibrium partial pressure of the solvent
This equation shows that both solvent type and type of carbon affect performance. On the solvent side, its density, affinity, saturated vapor pressure and its actual concentration determine adsorption. On the carbon side, its porous structure and surface area available for adsorption determine the capacity of a particular filter. The data for the parameters of the equation can be found in chemical handbooks and therefore the equation can be used for predicting the adsorption capacity of a filter. The capacity of commercial filters is normally rated by experimental method in which amount of solvent adsorbed by filter is determined under conditions which specify concentration of solvent, rate of flow, and time. For example, a filter which contains 70 g of activated carbon was found to adsorb 20-25 g carbon tetrachloride present in concentration of 1000 ppm. Under conditions of use it is important to predict how long a particular cartridge will last under real conditions. This is done by estimation of the breakthrough time from the following equation: t=
Wρ c An 1 z + QCo acρ c
dG η
0. 41
η ρ D a
0. 67
C ln b Co
where: W ρc A n Q Co z ac
adsorption capacity per unit weight of carbon carbon density cross-sectional area of adsorbent bed number of cartridges tested flow rate concentration of solvent V/A where V is carbon volume specific surface area
[24.2]
1636
George Wypych
d G η ρa D Cb
diameter of granule mass velocity through the cartridge viscosity of air-vapor stream density of air-vapor stream diffusion coefficient breakthrough concentration
This equation gives information on the relevance of major parameters of performance of filters. Breakthrough time increases with the increase of the following parameters: adsorption capacity of carbon, its density, its volume, and the cross-sectional area of the adsorbent bed and size of the granules. It decreases with increased flow rate, concentration of solvent, and an increase in the value of the diffusion coefficient. The toxicity of solvent plays a role here as well because with the increase of solvent toxicity, the breakthrough concentration is decreased which decreases breakthrough time. Breakthrough time for 3M cartridges can be calculated using available software by providing the type of solvent, its concentration, and type of work (light, medium, heavy). In addition, relative humidity is accounted for in the calculations. The adsorption of vapors is not affected by humidity below 50% but decreases rapidly as the relative humidity increases above 50%. Adsorption temperature is also an essential factor. The above discussion shows that, although there is generally one type of cartridge used for organic vapor, all cartridges are not the same in terms of performance. The conditions of use of these respirators determine if they can perform specific protective functions. Considering that one cartridge, typically disposed after one day work may contain 40 g of solvent which would otherwise be inhaled, the selection and use of respirators is not a trivial matter and should be given serious attention.
REFERENCES 1 2 3
K Forsberg, L H Keith, Chemical Protective Closing. Performance Index. 2nd Ed., John Wiley & Sons, New York, 1999. K Forsberg, S Z Mansdorf, Quick Selection Guide to Chemical Protective Clothing. 3rd Ed., John Wiley & Sons, 1997. W H Revoir, C-T Bien, Respiratory Protection Handbook, Lewis Publishers, Boca Raton, 1997.
25
New Trends Based on Patent Literature George Wypych ChemTec Laboratories, Inc., Toronto, Canada
Changes in production methods and the increased emphasis on environmental protection have produced continuous changes in solvent use and in the technology of solvent processing. The aim of this chapter is to analyze existing patent literature to illustrate the trends in • the development of new solvents • the use of solvents in various branches of industry • the new methods of solvent processing with emphasis on solvent recycling.
25.1 NEW SOLVENTS The developments and introduction of new solvents are discussed in Chapter 21 (supercritical fluids and ionic liquids) and Chapter 3 (terpenes). These three generic groups of materials are emerging technologies studied to address environmental concerns. Cleaning solvents are used in many industrial applications. These solvents have typically been highly volatile, toxic, and flammable. Dichloromethane, methyl ethyl ketone, toluene, xylene are frequently used for cleaning. They have all negative features contributing to potential hazards such as high volatility, flammability, and toxicity, and they influence the natural balance of the environment. If they are to be replaced, their replacements must have equal performance − a difficult task considering their excellent cleaning properties. Two recent patents1,2 suggest alternative solutions. 1-methyl-2-pyrrolidone is a suitable substitute for dichloromethane and methyl ethyl ketone in terms of its solvent abilities. It is less flammable and less volatile but its high cost prohibits its use in many applications. When it is blended with an alicyclic carbonate, a polar solvent, and terpene, a non-polar component, cost reaches an acceptable level and the blend can be used for cleaning paint equipment, reclamation of silkscreens, cleaning offset print rollers, printing surfaces, and as a general purpose cleaner.1,2 The removal of soiling and graffiti from exterior coatings requires the use of substantial quantities of solvents. Paint removers such as dichloromethane, benzene or toluene all of which which are hazardous solvents are commonly used. Several applications are needed before complete removal is achieved. Anti-graffiti coatings which are based on
1638
George Wypych
wax-containing compositions help to reduce solvent use. Such coating does not contain a high concentration of solvent and can be cleaned using water heated to temperatures from 120 to 190oC under elevated pressure. This cleans the graffiti but because the cleaning temperature must be higher than the melting temperature of wax, the coating is also removed and must be re-coated. An alternative method is based on the use of environmentally friendly 1-methyl-2-pyrrolidone which cleans graffiti without removing protective coating.3 Terpene, d-limonene, was selected for cleaning oil spills and waste petroleum products (e.g., tar or grease).4 The combination of 1-methyl-2-pyrrolidone with d-limonene makes a good paint stripper.5 It eliminates dichloromethane and other undesirable solvents. Replacement of fluorine-containing solvents is yet another goal of recent inventions. A non-toxic solvent was developed for cleaning electrical parts and motors.6 Such solvents must have high dielectric properties to prevent damage to equipment, electrocution, and/or fire. Chlorinated and fluorinated solvents which met these perfomance requirements can be replaced by a formulation according to the invention6 which uses a mixture of aliphatic hydrocarbons and adipate, salicylate, and succinate esters. A similar composition is also proposed7 as the solvents in the production of PVC foams. Medical implants must have roughened surfaces or a porous coating to promote bone and tissue integration. During the manufacture of such implants, loosely adhering coating particulates and debris collect on the surface and must be removed. The conventional cleaning method involves surface blasting with particles of hard material such as alumina, glass beads, etc. This may produce cross-contamination which will adversely affect implant performance. Other methods include polishing with ice particles, dry ice particles, and chemical cleaning. An invention8 which includes blasting of the surface with soluble solids (e.g., water soluble inorganic salts) which are then removed in ultrasonic water bath offers reduced contamination and eliminates the use of organic solvents. A few general examples show that solvents can be optimized for better performance by mixing. Substantial amounts of solvents can be eliminated by the careful tailoring of the process and properties to the application. Post cleaning with solvent based materials can be minimized or eliminated.
25.2 ADHESIVES When the impact of a particular adhesive technology is being assessed, all relevant factors must be considered. These include the method of adhesive production, solvents present in the adhesive itself and the cleaners and primers required for application. The current patent literature provides many examples which reduce pollution without compromising the quality of the joint. Regulations are primary drivers for this new technology with California regulations having the greatest impact. These regulations include the VOC limits imposed by the South Coast Air Quality Management District which reduced VOC limits from actual 750 to 850 g/l for some solvent contact adhesives to below 250 g/l. These limits depend on the technology and on the adhesive type. A patent9 provides a means to eliminate solvent from the production process of a pressure sensitive adhesive. The typical method of producing pressure sensitive adhesive involves mastication of the elastomer, dissolving the elastomer and additives in a hydrocarbon solvent, coating the solution onto a backing, and drying to remove solvent. The phenolic tackifying resin of the invention eliminates the use of solvent process.
25 New trends based on patent literature
1639
But often solvents must be used in adhesives in order to lower their viscosity so that product can be applied without the use of extensive force and so that the adhesive can flow into small openings in the substrate surface to promote mechanical adhesion by anchoring. The easiest formulating technique is to add solvent. This is simple and there is no need to develop new polymeric material. This is particularly true of polyurethane based adhesives which require either a special blending techniques or manufacture of the polymer with low viscosity polyols to reduce its viscosity. However, formulation of polyurethanes without solvents is technically possible as shown in patents for an automotive adhesive10 and a general purpose adhesive.11 In these areas additional research will bring dividends but the current effort lags behind the needs. In some adhesives, the total elimination of solvents is not possible. The best example is that of a solvent contact adhesive which is composed of solvent and a rheological additive. The solvent performs the essential function of swelling the adjacent surfaces of the polymeric material surfaces which are to be bonded. Until recently it seemed that coping with California regulations with such adhesives systems would not be possible. However, a recent patent12 shows that not only can the amount of volatile solvent be reduced to fulfil California requirements but also the toxicity of the solvent can be reduced and its flash point increased without compromising the performance characteristics of the adhesive. In earlier concepts of contact adhesives very volatile solvents, such as tetrahydrofuran and methyl ethyl ketone were the ones most frequently used and this was based on the premiss that such solvents would penetrate the polymer rapidly and be removed from the material immediately after swelling occurred. This type of adhesive is particularly well suited when used in plastic piping systems where the assembled system must be tested before it is enclosed in partitions. This approach eliminated the use of very volatile solvents which evaporated to become air pollutants and, to provide adequate open time, were used in excess which generally led to dripping from the joint and consequent waste of adhesive. In the invention, 1-methyl-2-pyrrolidone is used in combination with dimethyl adipate and some aromatic hydrocarbon to produce an adhesive known for its ease of use, long term durability, application with no special equipment, low cost, fast cure, and low concentration of total solvent. In other invention13 concentration of solvent in a fumigation adhesive used for bonding polyethylene to polyethylene in agricultural applications was also reduced by the reformulation of the copolymer used in the adhesive. For cleaning/priming operations there are also new developments. One trend is to apply the primer in a water based form.14 In this invention, a water based primer was developed using water reducible polyurethanes. The primer is for general use but has been optimized for use with construction sealants where solvent based primers are most commonly used today. Another approach is to use environment friendly solvents such as terpenes for both cleaning and priming. But some attempts are misguided. In one case, it was proposed to replace the primer with an isocyanate, which does not evaporate readily therefore fulfils a requirement of the regulation, but exposes applicators to extremely toxic material. Attempts of this kind must be discouraged because regulations are not designed to cope with such a gross misuse of knowledge of chemistry. One invention15 consists of a multipurpose cleaner/primer for roofing membranes. These elastomeric sheets are normally powdered with talc to prevent interlayer sticking in the roll. Instead of using separate cleaning solvent followed by a primer application the
1640
George Wypych
product of invention was able to bond the talc and other contaminants to the membrane and provide a tacky surface to promote bonding to the roof substrate. These examples show that there are many opportunities to reduce or eliminate solvents from adhesives. These usually require a more intensive R&D effort but lead to safer and higher performance products.
25.3 AEROSPACE Information on inventions related to solvents used in the aerospace industry is sparse (13 patents in last 25 years compared with 935 for adhesives in the same period). The reason is the secretive nature of the industry with respect to new technology and the reliance on proprietary technology to maintain a technological advantage. There is certainly a move towards limiting the solvent use in aerospace industry as discussed in Chapter 14 but the primary objective is always the performance of the applied technology. Aerospace materials must have high solvent resistance.16-20 Patents16-20 which demonstrate this also stress the need for high quality and uniformity in the materials used. For these reasons, solvents were used to extract low molecular weight components which could be detrimental in space where materials vaporize or degrade extremely rapidly. Also, solvent removal after processing is emphasized to eliminate odor and remove components which may cause degradation of material exposed to highly a degrading environment.
25.4 AGRICULTURE Recent patents describe the use of solvents to improve the properties of agricultural chemicals. In one invention, a carrier was developed from an agglomerated composition of plant fibers and mineral filler. The purpose of the carrier is to absorb and hold a large quantity of pesticide until it is delivered to the application site. The pesticide must be in a form of low melting liquid. In order to reduce its melting point, solvents selected from aromatic hydrocarbons are used to dissolve pesticide.21 In a water dispersible composition of insecticide, solvent is used to convert insecticide to a liquid form at room temperature.22 Solvents proposed for this application are from a group of alkyl aromatic hydrocarbons, methyl esters of alkanoic acids, and ester mixtures derived from plant oils.
25.5 ASPHALT In various asphalt processing methods, the rheological properties of the compounded asphalt are of primary importance. One invention23 shows how, by blending different fractions of crude oil, the processing properties can be optimized for pavement applications. In another recent invention,24 solvent-free polyurethane was compounded with asphalt. Such products have always contained substantial amounts of solvents to reduce viscosity to the level that allows the material to be cold applied with simple techniques. Using organoclay as compatibilizer for polyurethane and asphalt and a suitable plasticizer, elimination of solvent becomes possible. In addition to environmental safety, these products can be used safely on substrates which would otherwise be affected by solvents the formulation (e.g., polystyrene insulation boards). This is one example which shows that the benefits of solvent free systems go beyond the elimination of air pollutants.
25 New trends based on patent literature
1641
25.6 AUTOMOTIVE APPLICATIONS Cleaning products contain the majority of solvents used in automotive maintenance. New equipment has been designed for solvent recycling.25,26 In one such design, solvent is dispensed from a holding reservoir to a wash basin. After the parts are washed, the spent solvent is directed to a distillation chamber and condensate is returned to the holding reservoir.25 The equipment is designed in such a way as to prevent vapor from escaping to the atmosphere. In another development a biodegradable aqueous solution of detergent is used for cleaning.26 The cleaning solution is periodically filtered to remove solid contamination then returned for reuse. Cleaning operations are conducted under elevated temperature to increase cleaning efficiency. In automotive refinishing shops, various cleaning liquids are used most of which are based on organic solvents. In addition to the potential of pollution, such cleaning liquids are not suitable for removal of the water soluble dirt frequently found on car body parts (e.g., tree sap, bird droppings, etc.). A cleaning composition has been designed which can clean both solvent soluble and water soluble dirt.27 It is composed of a water emulsion of hydrocarbon solvent, an alkyl ester, and a blend of surfactants. Extensive efforts are known to be in progress which may lead to a reduction and even the elimination of solvents used in automotive painting operations. Each coating layer is applied by a different process which may be either a water based coating or a powder coating. An example of a water based coating for an automotive base coat can be found in a recent patent.28 The resin is derived from propoxylated allyl alcohol containing both hydroxyl and carboxyl groups and can be cured by a variety of crosslinkers. This technology is also applicable to industrial appliance coatings and wood finishes.
25.7 COIL COATING Coil coating is performed in an enclosed system. Generally less emphasis is placed on the elimination of solvents and more to the quality of the coating. There are technological barriers to the reduction of solvent use due to the stringent rheological requirements of the coatings. A primer layer is applied after the metal surface has been cleaned and baked. Primer plays an important role in preventing metal corrosion, therefore several performance requirements must be met. The primer layer must have good flexibility to prevent mechanical fractures and loss of adhesion of the top coats. It must not have imperfections such as blisters, craters, pops, and voids any of those will conduct water to metal surface. The transport of water to the interface depends on the properties of the primer layer and its thickness. It is preferable that primer layers are thick and can be applied in one operation. This can be fulfilled if the primer is formulated to release solvent quickly during curing. After curing, the open structure of the film should harden to a uniform, non-permeable layer. In summary, the primer should form a flexible, uniform, and thick layer. This has been achieved by the application of a hydroxy-terminated polybutadiene component formulated with a large number of solvents from the group of petroleum distillates, ethers, ketones, and alcohols.29 The permeability of the resin has a prevailing influence on primer properties. There are on-going efforts to reduce the VOC content of coil coatings. High-solids coatings typically contain 50-70% solid resin. Coatings higher than 70 wt% solids are more difficult to design. The coatings termed “extra high solids” contain less than 20 wt% solvent. These extra high solids coatings were made possible by the design of the resin and the
1642
George Wypych
selection of solvents which, in combination, give low viscosity, good flow, and a rapid evaporation rate.30 Solvents useful in this invention are xylenes, cyclohexanone, methyl ethyl ketone, 2-butoxyethanol, ethyl-3-ethoxypropionate, toluene, and n-butanol. Like primers, top coatings have followed a similar development path. The goal is to obtain a thick coating in one pass. Several polymers are used for coil coatings, such as polyesters, polyester-polyurethane, PVC plastisols and organosols, polyvinylidene fluoride, silicone modified polyesters, epoxy resins, and acrylic resins. A typical film thickness with all resin systems except PVC plastisols is up to about 30 µm. PVC plastisol coatings are applied to give a film thickness in the range of 80-200 µm. Because of disposal issues, chlorinated polymers such as PVC are less desirable today and have been replaced by solvent based systems. It is possible to use hot melt coatings but the high cost of equipment limits their use. In one recent invention31 a halogen-free, heat curing coating composition was developed based on a polyester solution. The solvent combination was such that one solvent would have a boiling point of at least 160oC and preferably 220oC to allow for curing process to be fast without causing a non-uniform film. High boiling solvents can be selected from the group of aromatic hydrocarbons, esters, glycol esters, and alcohols. The concentration of solvents in this coating depends on formulation. In pigmented formulations, solvents constitute 25-40 wt% whereas in unpigmented formulations 30-45 wt%. Liquid, single component stoving enamels were developed based on polyurethanes in which polyisocyanate component was partially or completely blocked.32 Solvents constituted 30-60 wt% and should be mixtures of two or more components. In still another development, a retroreflective coating composition was developed.33 It is designed to reflect the incident light beam in the direction of light source. The major application is in road signs and automobile licence plates. These had not been previously produced by coil coating. In addition to the binder components the product contains more than 50 wt% reflective glass microspheres. The solvent content is over 30 wt% of the entire composition. Another invention deals with the role of a crosslinker.34 The crosslinking mechanism plays an important role in the performance of coating. If the crosslinker acts too rapidly, it prevents the formation of an uniform film. But if the crosslinker acts too slowly the coater must be slowed to an inefficient rate (typically belt speeds are about 200 m/min).
25.8 COMPOSITES AND LAMINATES The performance of composites and laminates depends heavily on interfacial adhesion between fibrous component and the binding layer. Traditional technologies combine layers of prefabricated textile or fibrous component which are adhered by layers of polymeric matrix. The most recent trend involves preparation of composites on a molecular level which increases the contact surface area between both components and thus performance of the fabricated materials. One such process makes a glass/polymer composites by mixing glass and polymer on molecular level. It employs a compatible solvent to disperse materials on molecular level. The preparation of the molecular phase composite involves four steps. The first is a selection of compatible solvents and solutes. The second involves mixing a solution of glass with polymer to form a homogeneous mixture at the molecular level. The third involves removing solvent while retaining a homogeneous molecular mixture. The fourth involves forming bonds between the glass and polymer. Polar solvents are used in this application with preference given to water. Consequently, polymers used are water soluble.
25 New trends based on patent literature
1643
The traditional polymer/glass composites (and other combinations of fibrous material with polymer) are made either by dispersion of glass fibers in the polymer carrier matrix or by intercalation of the polymer phase through a glass carrier matrix. In these processes, it is important to limit the formation of air pockets and to increase adhesion because both affect the performance of composites. For this reason, the rheological properties of the polymer matrix must be carefully controlled through the use of appropriate solvents. In composites produced from cores, the strength of the bond between facing and core must be maximized. Improved bonding has been achieved by wetting the core with solvent. The solvent serves both as a wetting agent and the adhesive solvent.36 Due to this wetting, compressive and shear modulus have been improved by a factor of 1.4. Depending on the adhesive and the core properties, an appropriate solvent was selected from a group including water, acetone, petroleum ether, xylene, methyl ethyl ketone, methyl isobutyl ketone, and some other solvents. In another invention,37 a decorative laminate was produced from a fabric coated on both sides with PETG copolyester. The coating contains very little volatiles.37 It was found38 that a selection of solvent has effect on film thickness which is related to the boiling point of solvent. Fast evaporating solvents such as n-butyl acetate and o-xylene gave thicker coatings. Thinner films were obtained when mesitylene or diglyme were used. A solvent-free adhesive was used for laminating vinyl sheet to wood.39 A new technology of recycling waste carpets and textiles was developed based on coating individual yarn fibers by an adhesive containing solvent.40 The high surface contact area produced materials with excellent mechanical properties.
25.9 COSMETICS Two groups of solvents are used in cosmetics industry. These are process solvents which are removed after the material is prepared and solvents which are retained in product. Preparing polyaminoacid particles for encapsulation of cosmetics requires a process solvent.41 This particular process has several stages. First, an emulsion of aminoacids is prepared which is then polymerized and the organic solvent is removed. Hollow particles can be obtained by adding a poor solvent. Such solvents belong to one of the following groups: halogenated aliphatic hydrocarbons, halogenated aromatic hydrocarbons, esters, ethers, hydrocarbons. The preferred solvents are dichloromethane, chlorobenzene, and o-chlorobenzene. Microsphere reservoirs for controlled release applications in medicine, cosmetics and fragrances are obtained evaporating solvent from an oil-in-water emulsion. The microcapsules have diameters of from 3 to 300 µm.42 Many polymers are suitable for use in this invention which is based on the knowledge that the inclusion of plasticizer renders porous and spongy structure as opposed to the hollow core and relatively solid surface which results when no plasticizers are used. Even smaller, nanometer size particles, can be prepared from cyclodextrin. These particles are useful as a carrier for pharmaceuticals and cosmetics. First, a solution of cyclodextrin in an organic solvent mixture is prepared, followed by the preparation of a water dispersion of surfactant. When the two components are mixed together a colloidal dispersion of microspheres is produced.43 Particles sizes range from 90 to 900 nm. Typical solvents are methanol, ethanol, isopropanol, and acetone. Solvents are used in the purification of xanthan gum to lower ash and to obtain a product with no traces of solvents.44 Lower alcohol is used as the solvent. Solvents have also been used to decolor fatty acid esters.45 Crude oils extracted by pressing or with solvents cannot be used in cosmetic products. Re-
1644
George Wypych
fining processes include “chromatographic” decolorization which involves diluting the oil in an apolar solvent and absorbing impurities by contacting the resulting solution with a solid absorbent in a column. Food quality n-hexane is used as a solvent. In some cases, solvents do remain in the final product. One such example occurs in the preparation of liquid vanillin composition used in food and cosmetics production.46 The preparation of such a solution is complex. The solution must be pourable at room temperature, have high solids concentration (50-70%), be mechanically and chemically stable, be easy to dilute, be transparent, be stable to bacteria, and inexpensive. The solvents include water, ethanol, and propylene glycol. Polymeric liquid crystals are prepared by dispersing polysaccharide in water.47 These liquid crystals are used for perfumes. Xanthan gum is also in use for thickening cosmetics.48
25.10 CLEANING In this broad range of applications, recent work on household and electronic cleaners and metal degreasers have produced the most patents. Several patents have been selected to show progress in household cleaners. Soap scum removal and hard water deposit removal is done using a multi-component mixture containing hydrophobic cleaning solvents from the following group: dipropylene glycol mono-butyl ether, tripropylene glycol monobutyl ether, and ethylene glycol monohexyl ether.49 In addition to the solvent, surfactants, polycarboxylic acids, hydrogen peroxide, a thickener, and water are used. A terpene-free microemulsion composition was developed based on combination of nonionic and anionic surfactants, methyl esters of C6 to C14 carboxylic acids, a co-surfactant and water.50 This is VOC-free composition for cleaning hard surfaces such as metal, glass, ceramic, plastic, and linoleum. An aerosol cleaning composition contains surfactant, a water-soluble or dispersible solvent and a chelating agent (potassium EDTA), a propellant, and water.51 This composition is specifically designed for the hard surfaces found in a bathroom. The composition forms foam which collapses on the stained surface to deliver the cleaning agent. The novelty is in the use of the chelating agent which enhances soil and scum removal. A furniture cleaning and polishing compound provides protection against water.52 This is achieved by the use of a halofluoro polymer in combination with mineral oil, a hydrocarbon solvent, silicon oil, a detergent, and water. A thickened aqueous cleaning compound was developed for cleaning hard surfaces, disinfection, and drain cleaning.53 The thickened product increases the residence time and cleaning effectiveness especially on non-horizontal surfaces. A firearm cleaning agent has been designed to remove the fouling residue which contains carbon and metal residue.54 This is achieved by a combination of butyl or amyl lactate which dissolves the carbon residue and detaches the metal residue, and an electrical neutralizing agent, namely a citrus distillate, which bonds with the positive ions of the metal residue. There are other examples of efforts to eliminate or reduce solvents but the general trend is towards optimization of formulations to perform well in a specialized application. A cleaning liquid for semiconductor devices comprises 1-5% R4NF (where R is a hydrogen atom or an alkyl group), 72-75% solvent, and water.55 Solvents used in this application must be water soluble. Suitable solvents are dimethylformamide, dimethylacetamide, 1-methyl-2-pyrrolidone, and dimethylsulfoxide. Resist residues, which are formed during dry etching by the reaction of the photoresist with a reactive gas containing chlorine, must
25 New trends based on patent literature
1645
be removed because they may cause disconnections or abnormal wiring. In past, alkaline cleaning liquids were used which do not perform well under severe etching conditions. A flux-removing cleaning composition is a water-based composition containing an alkaline salt component and a surfactant.56,57 This formulation is designed to eliminate solvent based products. A degreasing and cleaning compound for electronics has been developed based on a silicon containing cleaning agent (polyorganosiloxane), an isoparaffin (C3-C15), a surfactant, and a hydrophilic solvent (ethanol or acetone).58 This formulation is intended to replace chlorinated solvents such as trichloroethane, trichloroethylene, tetrachloroethylene, and carbon tetrachloride which have been used up to now. In another invention,59 a molecular level cleaning compound is proposed. It includes a specific solvent (n-propyl bromide), low boiling solvents (selected from the group containing nitromethane, 1,2-butylene oxide, 1,3-dioxolane, sec-butanol, ethanol, methanol), a saturated terpene, and a defluxing or ionic removing additive. The compound is designed for use in a vapor degreaser. It is suggested that this composition reduces the environmental impact because the solvents are non-flammable, non-corrosive, non-hazardous, and have very low ozone depletion potential. Superior cleaning capability of the compound is attributed to the strong solvent power of the mixture, bringing the material being cleaned into a contact with this fresh condensing solvent, and the solvent flashing action during its downward flow. In still another invention,60 a degreasing and cleaning compound is based on non-polar hydrocarbons (C9-C12 hydrocarbons or dehydrogenated terpenes), an organic polar component (alcohol, amine, ketone, or ester), water and a surfactant. In this method, residual solvent is removed from the hot article by contacting it with an aqueous medium or low boiling solvent which are subsequently removed by evaporation. An aqueous, solvent-free degreaser contains a blend of polyoxyalkylene block copolymers.61 This compound is intended for cleaning oils and greases from a variety of surfaces but specifically printed electronic circuit boards and automotive parts. The formulation is designed to compete with cleaning products which contain solvents. It is formulated with nonionic polymeric surfactant dispersed in water. In another development,62 a water-immiscible hydrocarbon is used in the first cleaning step. This is followed by treatment with an aqueous displacement solution which contains surfactant and a pH modifier. The displacement solution displaces the hydrophobic solvent residue from the surface of the substrate and prevents its redeposition. The invention is intended to replace ozone depleting CFCs and chlorine containing solvents. These patents indicate that, although there are attempts to replace solvents from formulation or to use more environmentally friendly solvents, the main requirement remains the performance of the cleaner and this is best served by using solvents. The major push has been to formulate high performance, specialized materials and processes which are carried out in an enclosed environment to prevent pollution. It should be underlined that in these applications quality of cleaning is critical and any compromise in cleaning effectiveness would contribute to an even higher environmental impact because it would lead to production of non-performing products which would have to be discarded.
25.11 FIBERS Solvents are used in the production of fibers and in their modification and recovery from wastes. Production of fibers with optoelectronic properties for optoelectronic modulators
1646
George Wypych
requires several steps.63 In the first step, a metal hydrate or a hydrated metal compound (based on Pb, La, Zr, Ti) is dispersed in solvent (ethanol, n-propanol, isopropanol, n-butanol, isobutanol, 2-methoxyethanol, or 2-ethoxyethanol), the resulting dispersion is then heated to polymerize the material and stretched to gel the fiber. The final fiber formation is achieved by heating the gelatinized fiber. Fibers containing intrinsically conductive polymer are manufactured from a composition containing an organic acid salt of an intrinsically conductive polymer, a matrix polymer, and a spinning solvent.64 Polyaniline is the intrinsically conductive polymer. Polyaniline alone cannot be processed into fiber because of its low solubility (standard wet spinning methods require a polymer concentration of 15-20%). Many polymers can be used as the matrix polymer. For this application, polyacrylonitrile was selected. A variety of solvents can be used but dimethylacetamide was found to be the most useful. There are many reasons for surface treatment of the fibers. In one process,65 introduction of functional groups is described. The equipment designed for this process includes a cleaning vessel, a vacuum drier, and a plasma treatment vessel. The fiber is first treated with a solvent, which is subsequently removed in the vacuum drier to remove all residual solvent. The surface is then modified by a plasma treatment. Cleaning removes dirt from the natural fibers and impurities from man-made fibers. Water, hydrocarbons, and halogenated hydrocarbons are used in an enclosed system. Surface etching and cleaning techniques which were used in the past released solvents and other materials to environment, especially because the fibers were not sufficiently dried. Fiber glass is modified by the application of a thermosetting resin in two pass process. Thus treated, these fibers meet the dust free requirement for use in microchip substrate.66 Epoxy resin is applied from a solution in methyl ethyl ketone. In another invention,67 cellulose fibers are treated with an aqueous solution to prevent fibrillation. Fibrillation gives a hairy appearance and occurs due to the mechanical abrasion of fibers when they are processed in a wet and swollen state. Polyester can be recovered from contaminated polyester waste which contains non-polyester components.68 The waste may be polyester blended with cotton or other fibers, polyester magnetic tapes, or coated polyester films and engineering resins. The waste is dissolved in dimethylterephthalate, methyl-p-toluate or dimethylisophthalate. The contaminants are filtered from solution and polyester is recovered by crystallization or used as a feedstock in a methanolysis process which forms dimethylterephthalate and alkylene glycol.
25.12 FURNITURE AND WOOD COATINGS A water based wood copolymer coating was developed to accentuate the color and the natural grain structure of the original substrate.69 This enables solvent based system such as nitrocellulose lacquers to be replaced by waterborne coating. In another invention,70 a water based composition was developed for covering scratches, blemishes and other damage on finished wood articles such as furniture, wood trim and wooden flooring. The inventors claim that the product has performance characteristics similar to solvent based systems. In still another invention,71 a repair kit was developed using soluble colorants. This kit contains solvent based product. An aqueous dispersion of polyurethane is proposed for coating wood substrates such as wood floorings, furniture, and marine surfaces.72 This product is intended to replace solvent based system, although the emulsion does contain 7% of a polar
25 New trends based on patent literature
1647
coalescing solvent (selected from the group of 1-methyl-2-pyrrolidone, n-butyl acetate, N,N-dimethylformamide, toluene, methoxypropanol acetate, dimethyl sulfoxide, and others). There is also solvent based system (solvent selected from a group of ketones, tertiary alcohols, ethers, esters, hydrocarbons, chlorocarbons). Process solvents are later stripped in the manufacturing process. There is an obvious move in the wood finishing industry to replace solvent based systems with water based systems. This is a necessary development given that finishing processes are often carried out in homes or in small plants or workshops which lack adequate ventilation.
25.13 PAPER Solvents are relevant in pulping processes and waste paper processing. Solvent based coatings and inks are applied to paper and paper must be designed to accept such materials without absorbing an excessive amount of solvent. The traditional pulping process uses various inorganic chemicals to separate cellulose from lignin and other components. A major problem associated with this process is a difficulty in recovering or destroying these inorganic chemicals before they become major environmental pollutants. Solvents may be part of the solution to this problem. A process was developed in which wood is digested in a single-phase mixture of alcohol, water, and a water-immiscible organic solvent such as a ketone.73 After digestion and adjustment of the proportion between water and solvents, lignin is present in the organic solvent, cellulose in the solid pulp phase, and hemicelluloses and dissolved sugars are in the aqueous phase. Acceptance of the solvent pulping method has been slow due to difficulties in solvent recovery. However, apparatus and technology have been developed which allow recovery of almost all the solvent.74 In another development,75 it was found that pulp bleaching can be performed more effectively in the presence of a solvent, either methanol or ethanol. Alcohol protects cellulose during the ozone bleaching stage better than does water. Better quality cellulose is the result. Wax paper is recycled by immersing it in solvent, preferably n-hexane.76 The wax is dissolved in the solvent and separated from the paper. Tipping paper used to produce cigarette filters is made with special grades of calcium carbonate which reduce the wear of cutting knifes and limit absorption of solvent from ink.77 Paper stock was developed for release paper with a reduced tendency to absorb solvent from the coating. This allows thinner coats to be applied which retain good release performance.78
25.14 PRINTING A rotogravure or flexographic printing ink has been developed which reduces wear on printing equipment.79 Solvents used in these inks include aliphatic, aromatic, and naphthenic hydrocarbons, alcohols, glycol esters, ketones, esters, and water. With water, glycol esters are used as cosolvents. A lithographic ink has been developed based on an ester-modified carboxyl-containing copolymer dissolved in a solvent system selected from a group which includes aliphatic and aromatic hydrocarbons.80 The invention gives better pigment dispersion and color development. A printing ink for gravure printing is based on a polyurethane oligomer dissolved in organic solvents selected from aliphatic, aromatic, and naphthenic hydrocarbons, alcohols, ethers, and esters.81 Here, again, the driving force behind the invention was to in-
1648
George Wypych
crease color strength. A novel method of screen printing was developed to apply surface layers of materials with properties such as phosphorescence, electrical conductivity, mechanical adhesion, dielectric properties, etc.82 The choice of binder and solvent depends on the substrate involved. A metallic appearance on plastic products can be obtained by applying an ink coating.83 Typical solvents include aromatic hydrocarbons, cyclohexanone, butyl acetate, diacetone alcohol, an glycol ether. The concentration of solvents is in the range from 75 to 90%. The method of processing is UV curing. The choice of solvent depends on the substrate. The method was developed as an alternative to hot foil stamping which has numerous limitations. In another invention,84 a device for cleaning screen plates was developed. The process used a solvent and cleans with increased precision. A water based ink was developed for writing, drawing and marking using fountain and ball pens.85 An organic solvent based ink was developed for ink jet systems.86 The ink contains 40 to 98% solvent. Typical solvents include ethanol, isopropanol, acetone and methyl ethyl ketone. In still another development,87 a water based ink jet composition was developed. These patents show that solvent replacement is not a primary activity in this field. There is substantial interest in developing new technology for printing industry but it is driven mostly by a desire to improve quality and to satisfy needs of emerging printing technologies.
25.15 STONE AND CONCRETE A limited number of inventions were patented in this very broad and important field of construction technology. A solvent based coating for a containment structure is based on chlorosulfonated polyethylene.88 The coating is intended for applications in trenches, tank interiors, containment lining, flooring systems and joint overlays. The aim is to improve adhesion of the coating to the substrates. Solvents are selected from acetone, xylene, methyl isobutyl ketone, diacetone alcohol, and several other ketones and esters. A water reducible coating for concrete sealing was developed based on an acrylic resin.89 The coating is intended to compete with solvent based coatings. It contains butyl carbinol as the dispersing solvent. A waterproofing coating was developed based on a solution of styrene polymer.90 Solvents are chosen from among dichloromethane, ethylene chloride, trichloroethane, chlorobenzene, acetone, ethyl acetate, propyl acetate, butyl acetate, benzene, toluene, xylene, ethyl benzene, and cyclohexanone. Efflorescence of masonry, brick, concrete, and mortar can be eliminated by coating with a polyvinylalcohol solution in water.91 A solvent addition during silica glass production prevents cracking.92 The most effective solvents are methanol, ethanol, n-propanol, isopropanol, N,N-dimethylacetamide, N,N-dimethylformamide, methoxyethanol, and tetrafurfuryl alcohol. Oily stains on stonework can be cleaned with petroleum hydrocarbons.93
25.16 WAX This section includes a discussion of formulated products which contain wax. A fluorinated copolymer was used in a wax composition for car body refinishing.94 A mixture of ethyl acetate, toluene and heptane is used as a solvent. A cleaning and polishing composition for automobiles is also based on a fluorinated polymer and contains hydrocarbon solvent at up to 70%.95 Universal auto lotion is a water/solvent emulsion of wax.96 It contains isoparaffinic
25 New trends based on patent literature
1649
solvents. A car polish formulation contains 10-35% mineral spirits.97 A semiconductor polishing composition contains alcohols, acid and water.98
25.17 SUMMARY The patents evaluated show that the inventions address the needs of the industry. All industries seek to improve or optimize product performance. Elimination of solvents in these inventions is secondary. This does not mean that there is a lack of awareness of environmental impact. In some industries, such as, for example, the aerospace industry, quality is so important that the best solution must be found even if it is not the least polluting. The best opportunity for solvent replacement are in industries which are known to have limited capabilities to protect employees by the use of protective equipment, such as the furniture industry. There are industries which need to be more concerned about pollution but do not seem to have developed new products with this in mind (at least by the account of existing patents). This applies to the metal stamping and processing industries, industries supplying DIY markets, asphalt processing industries, concrete coatings, etc. There is little emphasis placed on pollution by these industries because they are made up of small, frequently, one-man operations and thus they are very difficult to assess. This also brings up the issue of the investment required to improve the environmental record. For example, automotive industry was able and did invest in environmental protection and the results are clearly beneficial. It is interesting to ask the question “will solvents eventually disappear from the industrial environment?” The above discussion shows that many new technological processes are developed based on solvents. There are more new products containing solvents than there are inventions aimed at their elimination. So, in spite of a strong desire to eliminate solvents their production volume continues to increase slightly. One might ask if this approach is legally, politically and morally acceptable given that the general public see solvent elimination as a benefit. Or are the efforts being made simply directed at finding the least harmful means of delivering the product which the public demand. Above examples show that, although sometimes solvents are introduced in new processes which previously did not use them, the overall objective of having a lower impact on the environment seems to have been met. The paper industry example shows that the use of solvents in a well designed plant can be less harmful for the environment than traditional pulping. Technologies aimed at reprocessing wastes provide more insight. A large number of studies on polymer blending have shown that reprocessing of mixed plastics, usually conducted without solvents, results in products which are not useful. In order to make recycling work effectively (especially of products which are multi-component mixtures such as composites, laminates, coated fabrics, coextruded films, etc.) some means of material separation must be found and the use of solvents offers just such a possibility. This brings several questions: which solution is better? should we dispose of these materials in the soil and wait for 50-100 years until they are digested? or should we responsibly use solvent under controlled conditions which does not pollute water, air and soil and does not endanger workers and customers? Solvent use can meet public needs and modern technology can find ways to ensure that solvent emissions are not released to the external environment. Chapter 14 has several examples that show that many solvent releases can be avoided by elimination of negligence and sloppiness. For example the majority of solvents emissions from the petroleum industry can be avoided if valve leaks are repaired. This does not require extensive research but adoption of simple engineering principles to rectify the situa-
1650
George Wypych
tion. The patents discussed above teach us that solvents can be eliminated or their amount reduced if resins are reformulated to reduce their viscosity thus eliminating the need to add solvents. In many cases where products are reformulated, this option is completely neglected due to the fact that the formulator finds it easier to use solvents since they usually do not affect other process characteristics or material performance and provide the simplest tool to adjust application properties. More knowledge must be acquired and more research is required than is currently being expended in product development. Solvents are inexpensive and this tends to encourage their use to lower the cost of the product. If situation develops that allows manufacturers to be able to profit from solvent-reduced or solvent-free formulations many products will be reformulated accordingly. But this requires that the environmentally aware customers be willing to pay premium prices for products which are solvent-free. The other option for future improvements, which is certainly being exercised as is shown in the above patents, is a rationalization of solvent composition. Here, regulations have played a progressive role in eliminating or reducing some very harmful materials. New inventions are still needed which aim to find more benign solvent mixtures or to replace resins which require aggressive or harmful solvents by ones which allow for more environmentally friendly compositions. There are many technological processes which will continue to require solvents at the highest levels of technological development. In these instances, good engineering practices will take the place of solvent elimination. There is nothing wrong with using solvents in a responsible manner. Scientific and engineering evaluations of all options and an assessment of the total social costs and benefits will show the best ways to optimize products and processes.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
A J Lucas, Z E Halar, US Patent 5,665,690, Inland Technology Inc., 1997. A J Lucas, Z E Halar, US Patent 5,449,474, Inland Technology Inc., 1995. D Perlman, R H Black, US Patent 5,773,091, Brandeis University, 1998. W L Chandler, US Patent 5,549,839, 1996. E Stevens, US Patent 5,288,335, Stevens Sciences Corp., 1994. T Mancini, US Patent 5,186,863, US Polychemical Corporation, 1993. T Mancini, US Patent 5,158,706, US Polychemical Corporation, 1992. J A Davidson, A K Mishra, US Patent 5,344,494, Smith & Nephew, Inc., 1994. D C Munson, A C Lottes, A Psellas, D A Brisson, US Patent 5,914,157, 3M, 1999. H Khalil, W Majewski, G Nickel, G Wypych, J D van Heumen, US Patent 5,288,797, Tremco Ltd, 1994. G Krawczyk, V Dreja, US Patent 5,824,743, Morton International GmbH, 1998. C D Congelio, A M Olah, US Patent 5,962,560, BFGoodrich, 1999. J L Troska, L M Kegley, W B Dances, US Patent 5,932,648, Shell Oil Company, 1999. Y Tominaga, A Tsuchida, H Shiraki, US Patent 5,866,657, Takeda Chemical Industries Ltd., 1999. R L Senderling, L E Gish, US Patent 5,849,133, Carlisle Companies Inc., 1998. H R Lubowitz, C H Sheppard, US Patent 5,446,120, The Boeing Company, 1995. R R Savin, US Patent 5,098,938, 1992. R F Sutton, US Patent 4,906,731, DuPont, 1990. C H Sheppad, H R Lubowitz, US Patent 4,851,495, The Boeing Company, 1989. C E Sroog, US Patent 4,687,611, DuPont, 1987. A D Lindsay, B A Omilinsky, US Patent 5,843,203, GranTec, Inc., 1998. D L Miles, US Patent 5,427,794, Rhone-Poulenc, 1995. R E Hayner, P K Doolin, J F Hoffman, R H Wombles, US Patent 5,961,709, Marathon Ashland Petroleum, 1999. R J Janoski, US Patent 5,421,876, Tremco, Inc., 1995.
25 New trends based on patent literature
25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
1651
P G Mansur, US Patent 5,549,128, Mansur Industries Inc., 1996. D Hartnell, US Patent 5,464,033, Major Industrial Technology, Inc., 1995. J C Larson, G L Horton, US Patent 5,230,821, DuPont, 1993. S-H Guo, US Patent 5,959,035, ARCO, 1999. M T Keck, R J Lewarchik, J C Allman, US Patent 5,688598, Morton International Corporation, 1997. T Kuo, US Patent 5,922,474, Eastman Chemical Company, 1999. D. Keogler, M Schmitthenner, US Patent 5,916,979, Huels Aktiengesellschaft, 1999. R Braunstein, F Schmitt, E Wolf, US Patent 5,756,634, Huels Aktiengesellschaft, 1998. G L Crocker, R L Beam, US Patent 5,736,602, 1998. W T Brown, US Patent 5,936,043, Rohm and Haas Company, 1999. W D Samuels, G J Exarhos, US Patent 5,422,384, Battelle Memorial Institute, 19995. K-H Holtz, J Luke, W Riederer, US Patent 5,575,882, Hoechst Aktiengesellschaft, 1996. M D Eckart, R L Goodson, US Patent 5,958,539, Eastman Chemical Company, 1999. D A Babb, W F Richey, K Clement, E R Peterson, A P Kennedy, Z Jezic, L D Bratton, E Lan, D J Perettie, US Patent 5,730,922, The Dow Chemical Company, 1998. G R Magrum, US Patent 5,837,089, Ashland Inc., 1998. A M Kotliar, S Michielsen, US Patent 5,912,062, Georgia Tech Research Corporation, 1999. K Makino, S Fukuhara, K Kuroda, T Hayashi, US Patent 5,852,109, JSR Corporation, 1998. R J R Lo, US Patent 5,725,869, Zeneca Limited, 1998. M Skiba, D Wouessidjewe, A Coleman, H Fessi, J-P Devissaguet, D Duchene, F Puisieux, US Patent 5,718,905, Centre National de la Recherche Scientifique, 1998. A Haze, K Ito, N Okutani, US Patent 5,473,062, Takeda Chemical Industries, Ltd., 1995. P Gonus, H-J Wille, US Patent 5,401,862, Nestec S.A., 1995. F Fournet, F Truchet, US Patent 5,895,657, Rhone-Poulenc Chimie, 1999. M El-Nokaly, US Patent 5,599,555, The Procter & Gamble Company, 1997. M H Yeh, US Patent 5,552,462, Rhone-Poulenc Inc., 1996. A E Sherry, J L Flora, J M Knight, US Patent 5,962,388, The Procter & Gamble Company, 1999. S F Gross, M J Barabash, J F Frederick, US Patent 5,952,287, Henkel Corporation, 1999. M Ochomogo, T Brandtjen, S C Mills, J C Julian, M H Robbins, US Patent 5,948,741, The Clorox Company, 1999. K J Flanagan, US Patent 5,925,607, Sara Lee Corporation, 1999. C K Choy, US Patent 5,705,467, 1998. P R Pomp, US Patent 5,935,918, 1999. T Maruyama, R Hasemi, H Ikeda, T Aoyama, US Patent 5,962,385, Mitsubishi Gas Chemical Company, Inc., 1999. F R Cala, R A Reynolds, US Patent 5,958,144, Church & Dwight, 1999. F R Cala, R A Reynolds, US Patent 5,932,021, 1999. M Inada, K Kabuki, Y Imajo, N Yagi, N Saitoh, US Patent 5,888,312, Toshiba Silicone Co. Ltd., 1999. L A Clark, J L Priest, US Patent 5,938,859, Lawrence Industries Inc., 1999. G J Ferber, G J Smith, US Patent 5,300,154, Bush Boake Allen Limited, 1994. M C Welch, C O Kerobo, S M Gessner, S J Patterson, US Patent 5,958,859, BASF, 1999. S B Awad, US Patent 5,464,477, Crest Ultrasonic Corporation, 1995. K Kitaoka, US Patent 5,911,944, Minolta Co. Ltd., 1999. P J Kinlen, B G Frushour, US Patent 5,911,930, Monsanto Company, 1999. S Straemke, US Patent 5,960,648, 1999. B K Appelt, W T Fotorny, R M Japp, K Papathomas, M D Poliks, US Patent 5,928,970, IBM, 1999. C D Potter, US Patent 5,882,356, Courtaulds Fibers Limited, 1999. W D Everhart, K M Makar, R G Rudolph, US Patent 5,866,622, DuPont, 1999. A J Swartz, M E Curry-Nkansah, R P Lauer, M S Gebhard, US Patent 5,922,410, Rohm and Haas Company, 1999. I J Barlow, US Patent 5,849,838, S C Johnson & Son, Inc., 1998. R L Setzinger, US Patent 5,590,785, 1997. K C Frisch, B H Edwards, A Sengupta, L W Holland, R G Hansen, I R Owen, US Patent 5,554,686, 3M, 1996. S K Black, B R Hames, M D Myers, US Patent 5,730,837, Midwest Research Institute, 1998. J H Lora, J P Maley, B F Greenwood, J R Phillips, D J Lebel, US Patent 5,681,427, Alcell Technologies Inc., 1997.
1652
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
George Wypych
M Solinas, T H Murphy, A R P van Heiningen, Y Ni, US Patent 5,685,953, MacMillan Bloedel Limited, 1997. R Vemula, US Patent 5,891,303, 1999. L Snow, K Mahone, I Baccoli, US Patent 5,830,318, Schweitzer-Mauduit International Inc., 1998. B Reinhardt, V Viehmeyer, M Hottentrager, US Patent 5,807,781, Kammerer GmbH, 1998. J S Perz, S L Rotz, US Patent 5,958,124, The Lubrizol Corporation, 1999. R H Boutier, B K McEuen, M F Heilman, US Patent 5,948,843, Elf Atochem North America, Inc., 1999. P G Harris, P D Moore, US Patent 5,886,091, Milliken & Company, 1999. D-A Chang, J-Y Lu, US Patent 5,843,534, Industrial Technology Research Institute, 1998. B W Bechly, US Patent 5,961,706, Technigraph Corporation, 1999. O Tani, US Patent 5,901,405, Tani Electronics Industry Co., Ltd., 1999. R Fraas, US Patent 5,961,703, J S Staedtler GmbH & Co., 1999 S Engel, A Badewitz, US Patent 5,935,310, J S Staedtler GmbH & Co., 1999. B A Lent, A M Loria, US Patent 5,929,134, Videojet Systems Interantional, Inc., 1999. T G Priest, R P Chmiel, R L Iazzetti, E G Brugel, US Patent 5,814,693, Forty Ten LLC, 1998. P A Smith, US Patent 5,777,071, ChemMasters, 1998. J H Gaveske, US Patent 5,736,197, Poly-Wall International, Inc., 1998. T Beckenhauer, US Patent 5,681,385, 1997. K Takei, Y Machii, T Shimazaki, H Teresaki, H Banno, Y Honda, N Takane, US Patent 5,871,558, Hitachi Chemical Company, 1999. J E Kerze, US Patent 4,956,021, 1990. M R Wollner, US Patent 5,962,074, 3M, 1999. P A Burke, K J Flanagan, A Mansur, US Patent 5,782,962, Sara Lee corporation, 1998. R L Fausnight, D A Lupyan, US Patent 5,700,312, Blue Coral, Inc., 1997. S Howard, R Frazer, US Patent 5,288,314, Crescent Manufacturing, 1994. D D J Allman, W J Crosby, J A Mailo, US Patent 5,861,055, LSI Logic Corporation, 1999.
Acknowledgments
1653
ACKNOWLEDGMENTS This following section contains acknowledgments included in the various sections of the book which were combined to form one section. For better identification, individual acknowledgments follow the reference to the title and authors of the book section. Preface George Wypych, ChemTec Laboratories, Inc., Toronto, Canada I would like to thank Dr. Robert Fox and John Paterson who made all efforts that the language used in this book is simple to understand and the book is read with pleasure. 4.2 Polar solvation dynamics: Theory and simulations Abraham Nitzan, School of Chemistry, the Sackler Faculty of Sciences, Tel Aviv University, Tel Aviv, 69978, Israel This work was supported by Israel Science Foundation. I thank my E. Neria, R. Olender and P. Graf who collaborated with me on some of the works described in this report. 4.4 Methods for the measurement of solvent activity of polymer solutions Christian Wohlfarth, Martin-Luther-University Halle-Wittenberg, Institute of Physical Chemistry, Geusaer Straße, D-06217 Merseburg, Germany Thanks are given to G. Sadowski (TU Berlin) for providing Figure 4.4-7(b), B. A Wolf (Univ. Mainz) for providing Figure 4.4-13, and G. Maurer (Univ. Kaiserslautern) for providing Figure 4.4-6. Furthermore, I wish to thank M. D. Lechner (Univ. Osnabrück) and G. Sadowski for many helpful comments and discussions about this manuscript. 5.4 Mixed solvents, a way to change the polymer solubility Ligia Gargallo and Deodato Radic, Facultad de Quimica, Pontificia Universidad Catolica de Chile, Casilla 306, Santiago 22, Chile The authors wish to express their appreciation to Mrs. Viviana Ulloa for her technical assistance in this work and to publishers and authors for permission to reproduce figures and tables from their publications as indicated specifically in the legends of the figures and tables. 6.1 Modern views on kinetics of swelling of crosslinked elastomers in solvents E. Ya. Denisyuk, Institute of Continuous Media Mechanics; V. V. Tereshatov, Institute of Technical Chemistry, Ural Branch of Russian Academy of Sciences, Perm, Russia This work was supported by a grant from Russian Fund of Fundamental Research (grant No 98-03-33333).
1654
Acknowledgments
10.3 Solvent effects based on pure solvent scales Javier Catalán, Departamento de Química Fisíca Aplicada, Universidad Autónoma de Madrid, Cantoblanco, E-28049, Madrid, Spain The author wishes to thank all those who contributed to the development of our solvent scales (C. Díaz, P. Pérez, V. López, J.L. G de Paz, R. Martín-Villamil, J.G. González , J. Palomar, and F. García-Blanco) and also Spain’s DGICyT (Project PB98-0063) for funding this work. 12.2 Chain conformations of polysaccharides in different solvents Ranieri Urbani and Attilio Cesàro, Department of Biochemistry, Biophysics and Macromolecular Chemistry, University of Trieste, Italy The paper has been prepared with financial support of University of Trieste and of Progetto Coordinato “Proprietà dinamiche di oligo e polisaccaridi”, Grant CT97-02765.03 of the National Research Council of Italy (Rome). The authors wish also to thank dr. Paola Sist for patient technical assistance. 13.2 Solvent Effects on Free Radical Polymerization Michelle L. Coote and Thomas P. Davis, Centre for Advanced Macromolecular Design, School of Chemical Engineering & Industrial Chemistry, The University of New South Wales, Sydney, Australia We acknowledge the publishers Marcel Dekker for allowing us to reproduce sections of an earlier review, “A Mechanistic Perspective on Solvent Effects in Free Radical Polymerization”.128 MLC acknowledges the receipt of an Australian Postgraduate Award. 14.19.2 Recent advances in coalescing solvents for waterborne coatings David Randall, Chemoxy International pcl, Cleveland, United Kingdom I would like to acknowledge with much gratitude the help given by Mr R J Foster of Harco for his help in assembling the MFFT data for the presentation. I must also thank my colleagues at Chemoxy, Ms Carol White, who assembled much of the data used in this paper, and Miss Tracy McGough, who helped me produce the OHPs. Finally, I must acknowledge the assistance given by Mr T J P Thomas, who has acted as a consultant to Chemoxy International in this whole area. I am indebted to Bob Foster at Harco, who kindly carried out some comparative formulations using Coasol, Di-isopropyl AGS and Di-isopropyl Adipate in comparison with a Monoester of Pentane Diol. 14.21.1 Use of solvents in the manufacture of drug substances (DS) and drug products (DP) 16.2 Residual solvents in pharmaceutical substances Michel Bauer, International Analytical Department, Sanofi-Synthélabo, Toulouse, France; Christine Barthélémy, Laboratoire de Pharmacie Galénique et Biopharmacie, Faculté des Sciences Pharmaceutiques et Biologiques, Université de Lille 2, Lille, France The authors thank Nick Anderson, Steve Byard, Juliette de Miras and Susan Richardson for their participation in the elaboration of this document.
Acknowledgments
1655
15.2.2 A simple test to determine toxicity using bacteria James L. Botsford, Department of Biology, New Mexico State University, Las Cruces, NM, USA This work has been supported by the principal investigator’s participation in several programs to assist ethnic minorities in the sciences. Many students have helped with this work. 20.3 Pregnancy outcome following maternal organic solvent exposure Kristen I. McMartin and Gideon Koren, The Motherisk Program, Division of Clinical Pharmacology and Toxicology, Hospital for Sick Children, Toronto, Canada Supported by grants from Imperial Oil Limited, Physician Services Incorporated, The Medical Research Council of Canada, and the CIBC Global Market Children’s Miracle Foundation Chair in Child Health Research, The University of Toronto. 20.4 Industrial solvents and kidney disease 20.6 Chromosomal aberrations and sister chromatoid exchanges 20.7 Hepatotoxicity Nachman Brautbar, University of Southern California, School of Medicine, Department of Medicine, Los Angeles, CA, USA The author wishes to thank Ms. S. Loomis for her tireless work in transcribing this manuscript. 21.1 Supercritical solvents Aydin K. Sunol and Sermin G. Sunol, Department of Chemical Engineering, University of South Florida, Tampa, FL, USA Assistance of both Dr. John P. Kosky of MEI Corporation and Irmak E. Serifoglu with editing and typesetting are appreciated.
Index A ab initio 39,424,426,430,455,470,485,655,661, 711 abiotic attenuation 1579 degradation 1175 process 1151 absorbance 343 absorption capacity 1635 acceptor number 566 acetaldehyde 1131 acetic acid 9,506,829 acetone 341-342,640,829,847,849,854,881,906, 959,1149,1508 acetonitrile 2,13 acetophenone 819 acid/base character 580 interaction 3,565,570,584,683,689 properties 617 acidic groups 348 acidity 585,806,1053 acids 574 acoustic dissipation 364 wave admittance analysis 342 acrylics 848 acrylonitrile 787 activated carbon 1509,1513,1543,1634 service life 1531 activation energy 778,807 theory 362 activity coefficient 177,282 at infinite dilution 164 adhesion 14,339,570 adhesives 581,847,849,851,1638 anaerobic 847 chemically reactive 847 polychloroprene 849 pressure-sensitive 851 UV-curable 847 adsorbent 1509 inner surface area 1510 adsorption 570,1151 capacity 1510 enthalpy 1509 fundamentals 1509
heat 1526 isotherm 1510 kinetics 1509,1511 preferential 274 adsorption-desorption 1160 adsorptive power 1510 aerosolization 354 aerospace 852,1640 AFM 569,706 agar 710 aggregate mass 689 vesicular 690 aggregates 506,580,697 aggregation 565,570,578,689,697 agitation speed 798 agriculture 1640 air 1149 contaminants, standard 925 velocity 1526 airflow 353-354 airways 1082 Alagona approach 444 alcohol consumption 1396 alcohols 127,137,806,828,880 alkahest 7 alkalinity 1053 alkanes 127 N-alkylpyridinium 1468 allergic effects 1319 potential 1320 Allerhand-Schleyer scale 588 p-allylanisole 826 allylation 811 altitude 1191 alumina 1513 aluminum stearate 693 amine 510 aminoacids 23 ammonia 16 amobarbital 286 amorphous phase 351 structure 348 amphiphiles 691 amphotericity 577 n-amyl acetate 848-849 n-amyl ketone 849 analytical ultracentrifuge 186 anemia 1376
1658
Index
anhydrous form 282 aniline 9 point 102,1056 anisole 819 annealing 254 anthraquinone 701 antifungal properties 881 antigens 1361 antiparallel orientation 12 Antoine equation 389 apolar 875,884 approximation, two-body 37 aprotic 762-763,798,811,813,829 aquifer 1572 Arrhenius description 573 asphalt 855 processing 1640 association heteromolecular 505-507 number 692,689 self 350 asthma 1328 atomic surface tension 482 ATP 866 attenuation capacity 1163 autoignition temperature 1054 automotive 851 finishing 1641 industry 958 maintenance 1641 painting operations 1641 autoprotolysis 524 Axilrod-Teller term 453
B bacteria Gram-positive 862 solvent-tolerant 859 Bader charges 444 barbituric acid 286 Barker-Henderson approach 749 barotropic effect 1425 barrier properties 693 basicity 585 bathochromic shift 590,592,602 battery electrolytes 1472 Beer law 353 Benard cells 412 benzene 1,39,105,127,342,697-698,812,817, 829,848,882,952,957,1127,1133,1334 exposure 1365 benzonitrile 794 benzyl alcohol 778,854 benzyl cyanide 815 BET 1510 binding energy 754 binodal 131,148,184,191-192,400,408
bioavailability 1130 biocatalysis 875 biodegradation 1151 first-order 1153 potential 1054 biofilm 1624 biological membrane 866,870 monitoring 1080 biomarker 1375 bioreactor 1624 biosphere 856 biotic process 1151 biotransformation 859,1133,1618 biphenyls 287,301 bleaching 344 blends 184,694 blistering 956 blisters 1131 Blodgett-Langmuir film 496 blood 1081 blue shift 639 blush resistor 956 boiling point 1055 temperature 15 Boltzmann factor 476,655 bond covalent 16 energy 16 intermolecular 345 strength 849 Bondi approach 126 Bootstrap model 791 Born approach 482 equation 769 formula 677 model 769 radius 771 term 652 theory 133,738 Born-Kirkwood-Onsager expansion 665 Born-Oppenheimer approximation 422 Bose-Einstein statistics 429 boson 429 boundary condition 312 surface 419 Brannon-Peppas equation 348 breakdown voltage 1059 breakthrough time 342,1631,1633,1635 breath monitoring 1079 breathing 1634 Brillouin theorem 659 Bristow-Watson method 246 bromine index 1055
Handbook of Solvents
bromobenzene 691-692,784 1-bromobutane 811 Brönsted equation 587 Brooker scale 590 Brownian dynamics 472 bubble collapse 360 dynamics 356,366 formation 354 growth 360 small 363 surface 364 Buckhingam formula 682 potential 459 burning, sustained 1069 butabarbital 286 1-butanol 545,694,854 2-butanol 694 t-butanol 694,1397 2-butanone 794,854 butyl acetate 848,881 butyl glycol 880 t-butyl methyl ether 2 2-butylhexanol 848 γ-butyrolactone 696
C calorific value 1056 calorimeter 243,1056 calorimetric study 540 capillary condensation 1510,1524 effect 379 force 365 carbon dioxide 2,39,688,897 carbon tetrachloride 591,648,697,754,827,1133, 1382,1396 carbonyl frequency 698 group 842 stretching frequency 567 stretching mode 349 carcinogen 1367 classification 1322 carcinogenic 2,1133 carcinogenicity 1132,1321 Car-Parrinello simulation 470 carrageenan 710 cartridge 1636 catalysis, triphase 830 catalyst 1467 phase 826 phase transfer 824 separation 828 catalytic oxidation 1508
1659
reaction 2 catheters 956 cation radius 771 cavitation 371 free energy 11,17 cavity 284,366 boundary 498 collapsing 366 ellipsoidal 680 field 682 formation 292,648,663 molecular 480 radius 717,769 shape 482,652,662,665 surface 21,284,484 surface area 284 volume 648 cell membrane 860 multiplication 1176 physiology 866 toluene-adapted 862 cellophane 1535 cellulose 127,694 acetate butyrate 881 chain dimension 706,711 flexibility 127,273 rigid 127 segment 345 separation 128 topology 709,711 transfer 779 chair conformation 711 Chako formula 680 charge density 709 distribution 132 transfer 463,586,780,782 complex 571 state 641 Charnahan-Starling equation 762 chemical 344 composition 419 equilibrium 1 potential 101,149,159,198,307,341,344-345, 575 reaction rate 132 reactivity 1,737,739,741,743,745,747,749, 751,753,755,757,759,761,763,765, 767,769,771,773,775 resistance 339 shift 591 chemisorption 1509 chlorinated hydrocarbons 827 chlorobenzene 506,794,810,812,830 1-chlorobutane 811
1660
1-chlorooctane 826 chlorofluorocarbons 853,1189 chromatographic separation 2 chromophore 589,592 chromosomal changes 1375 chromosomes 1376 circulation cell 1625 clathrates 1130 Clausius-Clapeyron equation 244 Clausius-Mosotti relation 680 theory 738 Clean Air Act 929 Clean Water Act 932 cleaners 1644 cleaning 853,894,957 processes 1508 cleavage 702 climate forcing 1192 clothing 339 materials 1631 protective 342 cloud point 101,184,189-190,271,273,408,695 Cluff-Gladding method 334 coagulation bath 695 coal 952 tar 855 coatings 386,565,963,1169,1555,1557 environmental performance 965 multilayer 35 waterborne 352-353 coefficient hydrodynamic interaction 685 mutual 340-341 of diffusion 354 of volumetric absorption 353 self-diffusion 340-341,344 coffee 1441 cohesion energy 115,261 coil conformation 709 dimension 273 expansion 268 coil coating 880,1641 collisions 419 color 1056 colorimetric analysis 1056 combustion 1056 cometabolism 1582 compatibility 318 compatibilizer 693 complex formation 556 complexation 780,782,824 composites 1642 compressibility 126,200 factor 244 isothermal 648
Index
Compton scattering 757 condensation 852 conductivity electrical 516,1126 thermal 354 configuration 18 interaction 431,658 conformation 18,25,707,710,788 conformational state 706 conformer 506,711 energy 530 equilibrium 530 form 532 stability 531 transformation energy 531 conformic equilibrium 506 construction industry 1648 products 855 contact angle 565 containers 693 continuum dielectric theory 133 continuum models 479 convective movement 1151 conversion 798 copolymer 686 copolymerization 779 model 781 corrosion 339,852,1057 cosmetics 881,1319,1643 COSMO 481,484 cosolvency 267,793 criterion 320 cosolvent 282,284-286,288,867,1175 power 269 Coulomb contribution 444 hole 430 integral 448 law 430,875 term 459 Coulombic attraction 575 interaction 439,451,574,727 Coupled-Cluster theory 431 coupling reactivity 816 covalence 463 cranial nerve 1328 cresol 957 meta 590 crew-cut 690 critical behavior 191 molar volume 340 point 191-192,202 region 214 scattering, pulse-induced 184 solution temperature 129
Handbook of Solvents
temperature 193 crosslink density 128 crosslinked elastomer 130 crosslinks 128,331,701 distance 345 cross-propagation reaction 786,790 cryoscopy 147,149,188 crystal energy 282 lattice 1,1130 surface 1130 texture 1118 crystalline phase 350 crystallinity 351,694 degree 352 crystallization 348,567,694 crystals 694 curing time 343 cybotactic 585 cavity 585 region 584,608 cyclodextrin 294 cyclohexane 105,185,188,506,591,701,810, 826-827 cyclohexanone 698,704,813,956,1131,1165 cyclopentanone 956 cytochrome 1099,1112,1356,1379,1394
D Dalton law 364 Deborah number 339-340 Debye longitudinal relaxation time 136 model 134 relaxation model 137 relaxation time 134 Debye-Hückel approximation 728 decaffeination 1441 n-decane 810,812,830 Dee-Walsh equation 203 defect, stress-related 386 defects 388 deformation 344 history 360 degreasing 1227,1233 dehydrobromination 826 dehydrogenase 1356 dehydrohalogenation 827,1572 de-icing fluids 853 DelPhi 481 demixing 176,184,189 - 190,206,273,695 density 505,516,521,1057 density-functional theory 574 deprotonation 33 dermal exposure 1079 dermatitis 881
1661
des Cloizeaux Law 308 desorption 566 free energy 569 DFT 481 dialysis 956 dibromomethane 807 p-dichlorobenzene 699 1,2-dichloroethane 127,686,819,1133 dichloromethane 686,692,694-695,698,701,810, 812-813,815-817,819,827,826,829,956, 1396,1398 dichroism 592 dielectric constant 25,115,190,454,459,479,487, 652,738,754,763,781,829,875 continuum 584,649,663 response function 133 surface 494 continuum theory 674 enrichment 613 Diels-Alder 1470 diethyl ether 2 di-2-ethyl hexyl phthalate 686 diethylaniline 815 differential refractometer 158 diffusion 339 coefficient 38,310,351,388,1152,1157,1509 control 779 distance 346,348,1526 elastic 340 equation 307 gradient 339 molecular 1082 rate 344,350 time 340 translational 779 diffusive transport 340 diffusivity 341 dilute 282 dilution ratio 102,1058 dimeric interaction 425 2,3-dimethyl-butane 341 1,3-dimethyl-2-imidazolidinone 1136 N,N-dimethylacetamide 694,1135,1400 dimethyl-benzene 854 N,N-dimethylformamide 127,188,694-695,956, 1135,1170,1400 dimethylsulfoxide 302,507,694,709 Dimroth-Reichardt scale 588 value 281 1,4-dioxane 686,691,694,709,956,1401 dip coating 956 molding 956 diphenyl ether 2 diphenylhydantoin 286 dipolar moment 15
1662
term 750 dipole moment 12,298,506,585,641,653,754, 763,807,875 dipole-dipole forces 584 interaction 12,138,528,530-531 permanent 767 Dirac function 662 dispersion 13 forces 454,649 disproportionation 783 dissociation 508 dissolution spontaneous 127 dissolving power 1422 distillation units 962 diurnal 1195 DMF 754 DNA 454 dodecane 826 domain amorphous 350 crystalline 350 Dong-Winnick scale 589 donor complex 785 number 566,738 donor-acceptor approach 744 interaction 574,584 properties 527 reactions 571 Doppler broadening 644 double-liner system 1165 Drago scale 591 equation 569 driveway sealer 855 driving force 339 drug acceptability 1130 degradation 1131 dry cleaners 1091 dry cleaning 883,1189 dryer 353 - 354,387 drying 339,565 conditions 386,1118 convection 353 defects 409 rate 396 technology 386 time 351 DTA 190 Dubois-Bienvenue scale 590 dye reduction 1100 dyes 1169 dynamic light scattering 689
Index
E ebulliometry 147,149,167 ecosystem 1617 ecotoxicity 1170 ecotoxicological effects 1185 effluents 1149 efflux pump 863 eigenfunction 427 eigenvalue 420 Einstein’s equation 684 elastic linkage 361 elasticity potential 128 electron affinity 574 cloud 469 transfer reaction 765 transport 1099 electron beam curing 880 electronegativity 754 equalization 444 electronic nose 1222 state 462 transition 291,655 electronic industry 894,1644 electronic-vibrational spectra 639 electron-pair acceptors 3 donation 571 donors 3 electrophilic species 782 electrostatic forces 649 interaction 12,453,541,651-652 potential 481 electrostatics 480 electrosynthesis 1473 elongational flow 360 embryotoxicity 1133,1334 Emergency Planing Act 933 emission 1149 limit value 1214 enantiomeric ratio 877 enantioselectivity 858,877 encephalopathy 1327 energy consumption 388 minimum 18 of activation 341 potential 345 enhancement factor 1429 enthalpy 566-567,755 conformational 254 differential 195 evaporation 243,1509 exothermic 567 molar 345
Handbook of Solvents
of melting 346 entropic contribution 722 entropy 130 component 332 conformational 257 dilution 195 hydration 760 ideal 105 environmental fate 1149,1151,1153,1155,1157, 1159,1161,1170 laws 924 enzymatic factors 1317 reaction 857,874 enzymes 857,872,1395 epoxy 847 equidensity 21 equilibrium chemical process 528 constant 788 equipotential 21 ESR 2,700,790 ethanol 286,694,701,847,882 ethers 127 2-ethoxyhexyl acetate 848 ethyl acetate 342,694,816,827,881 ethyl benzene 341-342,854 N-ethyl carbazole 816 ethyl ether 810 ethyl formate 1131 ethylene glycol ether 1169 eutectic point 1465 eutectics 1466 EVA 347 evaporation rate 353-354,956,1059 cooling 354 Ewald approach 487 excimer 268 exciplex 752,766 - 767 excitation energy 641 excited state 2 explosive limits 388,1060 exposure parameters 1315 extraction 1058,1068,1442 efficiency 940 process 2 temperature 943 extrusion 956
F factor of chain expansion 685 fermentation 1584,1599 Fermi hole 430 Fermi-Dirac statistics 428 fermion 428 fetotoxicity 1338
1663
fiber 695,1535,1645 diffraction 708 fibrosis 1394 Fick law 1084,1151 Fickian kinetics 340 film 693 casting 956 thickness 354 filters 1633 Fixman theory 728 flammability limits 1060 flanges 962 flash point 1060 flocculation 14 Flory equation 308 Flory matrix 723 Flory-Huggins equation 365,389 interaction parameter 112,151,321,345,347, 569 theory 154,793,1125 Flory-Orwoll-Vrij model 199,206 Flory-Rehner equation 128,321,331 theory 344 fluorescence 342,610,689 intensity 692 recovery 344 fluorescent radiation 701 fluorination 693 fluorine 693 fluorocarbon resin 880 foam-enhanced remediation 1626 folding rate 350 food 709,923 industry 1128 safety 934 force field 451 repulsive 14 formaldehyde 1190 resin 881 formation heat 790 formic acid 591 fractionation 689 Franck-Condon principle 291,752 state 644 - 645,654,658 free energy of mixing 345 free volume 125 freedom, degree 199-200,433,470,715 free-volume theory 339,353 freezing point 1061 frequency shift 699 Freundlich equation 1152 isotherm 1510,1525,1575
1664
Index
Friedel-Crafts alkylation 1468 frontier orbital interaction 574 FTIR 2,342-343,396,565,570,698,756,759 fuel, sulfur 1056 fugacity 149, coefficient 170,195,207 furniture 1169,1205,1222-1223,1646 future trends 946,1637
G gas cleaning 1507 gas chromatography 1062,1064,1116,1135, 1228,1562 gas tank 693 gasoline 1169 Gaussian chain 685 Gay-Berne potential 458 GC-IR 1131 GC-MS 1131 gel 333,344,707,710 stable 194 structure 698 gelation 697 genetic component 1361 geometry change 450 Gibbs energy of mixing 124 free energy 101,148,152,191,705 Gibbs-Dugem-Margulis equation 157,553 glass transition temperature 176,254,340, 348-349,703 glassy state 350 gloss 704 gloves 1631-1632 surgical 956 glycerol 286 glycol ester 880 glycol ethers 1169 goniometer 689 grafting efficiency 703 green chemistry 1422 strength 849 greenhouse gases 1189 ground transportation 950 groundwater 1150,1163,1571,1599 well 1623 Grunwald-Winstein scale 587 Gutmann concept 738
H habit 1119 hair spray 881 halorespiration 1582
Hammond-Stokes relation 39 Hansen solubility 112 parameters 106,264,695 hard sphere diameter 746 potential 14,37 hard-core cell 201 hardness absolute 575 Hartree-Fock approximation 484 description 430 equation 665 method 655 procedure 423 theory 661,666 hazard communication standard 926 hazardous wastes 1548 health and safety laws 924 heart 1415 heat capacity 254,1460 conductivity 392 dissipation 363 flow 354 of vaporization 387 transport 1 Helmholtz free energy 493,566 hematopoietic effect 1365 hematoxicity 1365 Henry constant 173,177,206,213,1155 law 148,389,1125,1150,1578 diagram 112 hepatotoxicity 1337,1379,1385,1393 hepatoxins 1396 n-heptane 590,697,848,1126 herbicides 1102 hexamethylphosphoric triamide 2 n-hexane 2,127-128,323,590,697,812,828, 848-849,1127 Hildebrand theory 124 Hildebrand-Scatchard equation 104 hypothesis 345 hole 339-340 HOMO 571,573-574,789 homomorph 585-586,593,600 homopolymerization 777-778 homopropagation 778 homotactic interaction 722 Hook law 160 Hookean energy 732 hops 1441 HPLC 759 HSGC 154-155,167,178 Huggins equation 246 Hughes-Ingold theory 745
Handbook of Solvents
humidity 852 relative 353 Huyskens-Haulait-Pirson approach 103 hybridization 754 hydrate 282 hydrocarbons 102,941,1361,1617 hydrodynamic interaction 361 hydrodynamics 356 hydrogen bond 15,683,709,760 bonding 111,573,584,613,703,740,1126 hydrolase 858 hydroperoxide 700 hydrophilicity 830 hydrophobic 866 effect 16,285,758,760 interaction 11,16,282 hydrophobicity 758,760,830,875 parameter 286 hydroxyl group 806 radicals 1559 hypernetted chain approximation 461,465 hypsochromic sensitivity 607 shift 590,607,609 hypsochromism 588
I IGC 148-149,154-155,164,166,177,565-566 immune system 1356,1367 incineration 880 infrared absorption 111 ingestion 1079 inhalation 1079,1316,1355 initiation 777 injection molding 956 ink 1647 inorganic chemicals 950 interaction charge transfer 16 constant 566 dimeric 424 electrostatic 16 intermolecular 639 parameter 345 polymer-solvent 1125 potential 455 trimeric 424 interatomic force 470 interface 339,565 air-solvent 288 intermolecular bonding 755 distance 12 forces 1 interaction 37,261 internal gradient 386
1665
intersolute 282 effect 282,293 ion association 546 ion-dipole interaction 528,545 ionic associate 507 liquid 2,1459 melts 616 radius 768 ion-ion interaction 528 ionization 816 iron and steel industry 951 irradiation 1131 isobutanol 351 isomerization 506,826 equilibrium 530 iosoctane 347 isopiestic measurement 158 isopleth diagram 1194 isopropanol 302,848,851,881,906,1397 isotactic 578 isotropic liquid 1465
J Jackob number 376 joint filling 855 jump unit 339-340
K Kapustinskii equation 1461 Kauri number 101 Kelvin equation 1510 ketones 880 kidney disease 1355,1357,1359-1361,1655 function 1359 kinetic rate 807 Kirkwood factor 754,762 theory 738 Kirkwood-Riseman-Zimm model 361 Kitaura-Morokuma decomposition scheme 429 knots 691 Knudsen method 245 Kohlemann-Noll theory 360 Kohn-Sham equation 471 Kosower scale 589
L labeling 1062 laboratory hoods 962 laminar thickness 694 laminates 1642 land disposal 1149 Langevin dynamics 472 Langmuir isotherm 1510,1575
1666
Index
method 245 laser 689 lattice-hole theory 203 LCST 198,268,372,380,384,1424 leachate concentration 1165 leather 1222 Leffler-Grunwald operator 285 Lennard-Jones energy 749 interaction 138 potential 37,40,458,462,467,477 solvent 748 lethality 1176 leukemia 1365 Lewis acidity 740 basicity 738 definition 571,617 donor 705 number 376,392 parameter 565 lifetime 654,1190 Lifshitz-van der Waals interaction 565 light absorption 2,680 limonene 864 linear response 133 response theory 133 lipase 878 lipophilic contamination 898 liposomes 860 liquid compressibility 356 constrained 495 crystals 485 general condition 357 motion 365 Newtonian 356 surface 493 system 419 lithology 1627 liver 1379,1393 function 1394 London dispersion 469 equation 718 formula 447,672 Lorentz-Lorenz relation 680 LUMO 571,573-574,789 lungs 1082 Lux-Flood definition 573,618 lyophilization 874
M macroscopic level 1 Manning theory 729 Marangoni number 413
Mark-Houwink equation 363,728 Markov chain theory 475 Martin correlation 362,390 mass transfer 387,816 material degradation 1632 Maxwell element 361 equation 358 model 360 Mayo-Lewis model 780 mechanical equilibrium 307 toughness 339 medical applications 955 devices 955 melting 253 melts 200 membrane 689 cytoplasmatic 863 morphology 694 memory effect 357 Menschutkin reaction 1,9,30,742,745 menstruum 8 menstruum universale 7 Mercedes-Benz logo 761 mesomorphic solid 1465 mesophase 1465 mesopores 1513 metabolites 1317,1356 metal cation 450 metal casting 957 methane 208 methanol 188,286,341-342,547,694,826,829, 957,1131 4-methyl-2-pentanone 854 N-methyl-2-pyrrolidone 398,686,694-695,698, 778,841,847,956 2-methyl-pentane 341 methyl acetate 342,694,881,1131 methyl chloroform 1396 methyl ethyl ketone 127,129,687,842,847,849, 854,881,952,956 methyl glycol acetate 881 methyl isobutyl ketone 703,881 1-methyl naphthalene 688 methylcyclopentane 129 1-methylimidizole 1460 micelles 690 microbial membrane 860 microcoulometry 1065 microdomains 350 microfiltration 694 microflora 1620 microorganisms 857,865 growth 1176 solvent resistant 865
Handbook of Solvents
survival 1176 micropores 1513 microscopy 689 microstructure 386 Millington equation 1152 mineral spirits 847-848,855,950 mining 1169 miscibility gap 268,273 mitochondria 1112 mixing configurational entropy 124 entropy 124,292 free energy 104 model classical 419 continuum 17,20,25,133 electrostatic 9 free volume 196 iceberg 16 lattice 198 lattice-fluid 196 supermolecule 18,22 molar cohesive energy 103 volume 516,521 molecular arrangement 698 design 36 dynamics 18,472 ensemble design 36 mass 345 mechanics 450 model 698 motion 254 orbital 423 orbital theory 571 orientation 752 packing 752 sieves 1513-1514 theory 361 toxicity 860 weight 343,351-352,689,703 molecule-molecule interaction 475 Møller-Plesset perturbation 670 theory 431 treatment 671 molten salt 1459 Momany approach 444 monoclinic 1119 monomer emission ratio 268 feed 781 monomer-monomer complex 785 Monte Carlo 17,19,37,208,472,476,662,723,760 morphological feature 690
1667
structure 689 morphology 339 of skin 339 Mulliken charges 444 mutagenicity 1132
N nail polish 881 remover 881 nanofibers 691 nanospheres 690 nanostructures 689 naphtha 848,854,881,950,957,1233 naphthalene 288,302 natural attenuation 1571,1573,1575,1577,1579, 1581,1583,1585,1587,1589,1591,1593, 1595,1597,1599,1601,1603,1605,1607, 1609,1611,1613,1615 navy 1472 nephropathy 1359 nephrotoxicology 1356 Nernst layer 831 nervous system 1407 network 331 networks 710 neuropathy 1328 neuropsychological assessment 1329 neurotoxicity 1132 neutron scattering 185,698,756 small angle 147 Newton-Euler equation 37 Newtonian viscosity 361 nitrile rubber 848 nitrobenzene 641,819-820 nitrocellulose 848,881,1646 nitroethane 1400 nitromethane 754 nitroparaffins 1400 2-nitropropane 1400 NMR 2,342,351,388,705-706,713,759,788, 793,1432 1 H-NMR 1135 shift 738 nondipolar 754 nonpolar 827 surface area 286 non-solvents 177 Norrish type II 701 nucleic acids 16 nucleophilic agent 807 species 782 nucleophilicity 813 number of arms 689 Nusselt number 392
1668
Index
O octane 127,185 1-octanol 287-288 octupolar 754 odor 851,1219 hedonic tone 1221 in industrial plants 1222 in materials 1222 intensity 1220 testing 1063 unit 1220,1223 odorant concentration 1220 oils 923 Oldroyd equation 358,361 olfactometer 1223 Onsager cavity radius 654 dipolar term 653 model 651,667 reaction field 654 theory 655,715 Onsager-Bottcher theory 682 opalescence, critical 190 operation cost 1508 optical clarity 339 orbital radius 574 organic acids 16 organic chemical industry 962 orientational order 762 Ornstein-Zernike equation 464 theory 490 Orofino-Flory theory 726 orthorhombic 1119 Osawa concept 729 OSHA standards 925 osmometry 147,149,154,168,195,267 membrane 178,183 Ostvald-Freundlich equation 1491 oxidation degree 618 oxygen, singlet 702 ozone 1189,1192,1267 depletion 1192 precursor 1189 production 1196 tropospheric 1190
P pacemaker 956 packing density 746 Padé approximation 750 paints 354,581,709,1169,1235,1555,1557 emissions 1559 industry 1063 painting 950 paper 1647
Papirer’s equation 566 partial pressure 1150 partition coefficient 288,291,860,875,1082, 1149,1153 partitioning 2,867 patching 855 patent literature 1637 Pauli exclusion principle 427-430 repulsion 649 Pauling crystal radius 770 PDAC 445 Pekar factor 766 Peklet number 363 PEL 925,1079 penetration 1422 perchloroethylene 914,956,1057 Percus-Yevick approximation 461,465 perfluorohexane 2 performance 1422 perichromism 2 permeability 693,860,1631 temperature effect 1632 permeation flux 694 number 1631 permittivity 505-506,513,517-518,531,540,669 perturbation theory 436,461 pervaporation membrane 693 pesticides 1169 PET 688 pH 297,348,688,1064 pharmaceutical 709,1552 industry 1468 product 1119 substances 282,1129,1135 pharmacopoeia 1129 phase boundary 344 diagram 129-130 homogenization 190 separation 125,190,386 thermally-induced 694 toxicity 860 transfer, catalyst 824 transition 1420 phenol-formaldehyde resin 847 phenomenological model 288 phospholipids 860 photochemistry 1194 photodegradation 1151 photoefficiency 1559 photoionization 133 photolysis 701,1175,1192 photooxidation 700,1175 photoreactor 1563 photoresists 701,1546
Handbook of Solvents
photosensitizers 701 photostabilization 702 phytoremediation 1622 pickling 952,957 piezoelectric method 163 pinholes 956 Piola tensor 306 pipeline coatings 855 plasma cleaning 897 plasticizer 856 plasticizing action 701 plastisol 847 PLUME 1164 p-nitrophenol 699 Poisson equation 480 polar 750,754,760,807,824,842,897 bond 754 character 586 liquids 693 solvation 746 polarity 2,9,288,531,577,584-585,610,700,745, 803,813,815,875 effect 780,782 index 527 parameter 288 scale 584,588 polarity-polarizability scale 584 polarizability 298,584,655,740,749,785 polarizable continuum model 481 polarization 12,133,428,486 effect 459 orientational 669 poly(acrylic acid) 687 poly(amide-imide) 841 poly(ethylene glycol) 206 poly(ethylene oxide) 186,687,699 poly(vinyl methyl ether) 185 poly(vinylidene fluoride) 693 polyacrylonitrile 127,694 polyaddition 841,963 polyamide 694 polyamides 694,847 polyaniline 703,1126 polybutadiene 127 polybutylene 848 polycarbonate 694,700,820 polychloroprene 127,848-849 polycondensation 818,820,841,963 polycrystalline particles 1118 polydimethylsiloxane 185,206,704 polydispersity 154 polyelectrolyte 454 polyester 880-881 polyesters 694,847 polyether 818 polyetheretherketone 348-349,699 polyetherimide 694
1669
polyethersulfone 694 polyethylene 128-129,693,701-702 polyethylene glycol 808,817,826 polyethylmethacrylate 103 polyimide 692,694,702,841,847 polyisobutylene 685,848 polyisobutylmethacrylate 103 polyisoprene 127 polymer coil 271,779 drying 350 segment 339 semicrystalline 350 solubility 198 polymerization 842 free radical 777 polymers 253 polymethylmethacrylate 127,269,343,568,570, 578,686,1437 polymorph 755 polymorphic transition 1119 polymorphism 709 polyolefins 694 telechelic 694 polyphenylene 685 polyphenyleneoxide 694 polypropylene 693 polypyrrole 568 polysaccharides 706-707,709,711,713,715,717, 719,721,723,725,727,729,731,733,735,1654 polysiloxanes 273 polystyrene 127,129,186,188,269,342,344,686, 698 atactic 128 isotactic 128,697 syndiotactic 697-698 polysulfides 818,848,853 polysulfonate 818 polysulfone 694 polyurethanes 331,346,354,694,847-848,850, 855,956,1640 polyvinylacetate 185,351-352,848 polyvinylalcohol 127 polyvinylchloride 127,129,691-692,698,880 pore diameter 1509,1524 size 694 system 1513 porins 862 porosity 1118 porous matrices 1422 membrane 694 potential energy surface 422 potentiometric titration 619 powder coating 881 Poynting correction 171,1429
1670
Index
Prandtl number 392 Prausnitz theory 208 pregnancy 1333,1335,1337,1339,1341,1343, 1345,1347,1349,1351,1353,1655 pressure drop 1526 partial 353 pressure-temperature relationship 1420 Prigogine approximation 199 cell model 200 theory 125 Prigogine-Flory-Patterson equation 202 theory 199-200,207 primer 705,881 printing 1169,1647 probability function 430 process safety management 927 propagation 778 heat 787,790 pathway 785 radical 793 step 780,782 1-propanol 545,694 2-propanol 694 propyl acetate 342 propylene carbonate 591 propylene glycol ether 1169 protagonism 15 protection 1631 protective equipment 1080 proteins 16,754,866 protic 745,752,762,798,813,829 protolysis 524 proton influx 860 leakage 860 transfer 507 proton-donor function 510 pulmonary blood vessels 1082 pulping 1647 pump valves 962 pupil dilatation 1222 purity 1065 pycnometer 1057 pyridine 507
Q QM/MM 478 quadrupolar 754 forces 753 quantum description 419 level 18 number 430 yield 702
R radiant energy 353 radicals 700,782 lifetime 701 stability 790 radius of gyration 684-685,689,691 Raman 756 Raoult law 154,196,365,372,389,613,794 activity coefficient 877 Rayleigh factor 182 ratio 689 time 363 reaction endothermic 1 field 481,651-652 field approach 487 nucleophilic 9 rate 798,807,811,815-816 reactive barriers 1628 reactivity 700 reactors 962 recrystallization 2,1130 recycling 880 red shift 752 refractive index 158,190,740,1066 regional scale 1193 regulations 1267,1269,1271,1273,1275,1277, 1279,1281,1283,1285,1287,1289,1291, 1293,1295,1297,1299,1301,1303,1305, 1307,1309 in Europe 1311,1313 regulatory approaches 1199 Reichardt scale 740 reinforcing 339 relaxation enthalphy 254 segmental 346 spectrum 361 time 133,340,358,361,654 relief devices 962 reproductive toxicity 1169 toxicology 1353 reptation 344 repulsion 14 residence time 387 residual stress 339 resistance specific 1059 resistive heating 1626 resistivity 1058 resorcinol 699 Resource Conservation Act 932 respirator 1633 respirators 1371
Handbook of Solvents
respiratory physiology 1081 protection 1633 respirometer 1099 retardation time 358 retention volume 566 retroreflective coating 881 rewind station 387 Reynolds number 366,370 rheological equation 356 properties 683,704 Rider approach 122 rigid spheres 14 ring deformation 711 RISM 466 rod-like micelles 690 Röntgen model 755 rotamer 591 energy 531 rotation 711 barrier 532 freedom 1477 Rouse theory 361 rubber 847 elasticity 148 natural 342 nitrile 342 rubbery state 349 Runge-Kutta-Gill method 393
S sample collectors 962 Sanchez-Lacombe equation 201 theory 207 SANS 185-186,706 SAPT 440 SAXS 184-185,706 scaled-particle theory 461 Scatchard equation 263 Scatchard-Hildebrand theory 197 scattered intensity 689 scattering angle 689 Schlieren photography 186 Schrödinger equation 420,482,656-657,663,671, 674,676 sealants 847-849,851,855 seals 962 second viral coefficient 689 sedimentation velocity 186 segment fraction 200 size 200 statistical 361 selectivity 694 self-consistent field 423
1671
self-diffusion coefficient 38 self-remediation 1622 SEM 689 semiconductor 1546 shear modulus 347 Sherwood number 392 shoe industry 849 shrinkage 354 silica gel 1513 silicones 847-848 Simha-Somcynsky equation 203 similia similibus solvuntor 8 simulated-annealing 470 simulation 137 numerical 132 skeletal structure 806 skin 1631 absorption 1355 layer thickness 697 protective devices 1371 Slater determinant 669 small molecule 339 Small method 261 Smidsrød-Haug approach 729 Snyder’s polarity 851 Soave-Redlich-Kwong eqaution 213 soil 1149,1157 contamination 1617 solar cell 1476 solid matrix 339 solubility 16 behavior 3 demands 318 equilibrium 282 parameter 103,111,198,243,253,261,347 relative 285 supercritical pressure 1425 solute 684 electrostatic field 133 mass concentration 684 polarization 11 pure 282 surface area 286 solute-solute interaction 282,292,301 solute-solvent interaction 455,481 solution 768 dilute 362 free energy 285 inhomogeneous 148 lattice 125 processing 956 solvation 282,759-760,807 continuum 482 dynamics 133 effect 283 energy 133,505,528,639,657,667 free energy 132
1672
function 137 preferential 613 process 133 selective 538 shell 17,132,284,288,301,450 solvatochromic 288 analysis 609 dyes 660,705 shift 639,655,660,665,674,677,738 solvatochromism 2,590,593,601,610,751 negative 602 solvent acidity 607 activity 146,194,207 adsorbed on crystal faces 1116 allergen 1319 basicity 600 binary 505,513 cavity 288,291 chemical potential 125 choice 872-873,875,877,879 cleaning 952 concentration 388 cost 657 crystalline 346 definition 1201,1234 degradation 700 diffusion 344,353 donor 787 effect 281 emission 1232 evaporation time 697 expansibility 747 explicit effect 780 exposure 1078 extraction 923 extraction process 935 flux 342 good 569,685 hydrophilic 1152 implicit effect 780 interaction 1130 ionizing power 587 loss 931 maximum contaminant level 1164 maximum permissible concentration 1164 metabolism 1317 mixed 505 mixture 267,281,352,683 motions 133,137 movement 339 neurotoxicity 1327 occluded 1116 oxidation 1559 partitioning 1067,1152 penetration 1631 polarity 3,593,610,702,705
Index
polarity scale 588 poor 569 power 101 probe 585-586 quality 346 recovery 1507,1509,1511,1513,1515,1517, 1519,1521,1523,1525,1527,1529,1531, 1533,1535,1537,1539,1541,1543 reduction 847 removal 386 replacement 847 residual 349,1125,1129 retention 570 route of entry 1355 selection 1 separation 693 solubility 1150 state 345 symmetric 318 toxicity 1097,1132 traces 701 transport 3 transport phenomena 339 treated 1556 uptake 348 vapor phase 196 solvent activity data 198 solvents 1150 applications 1507 binary 318 chlorinated 1572 classification 1235 cleaning 1056 critical properties 1420 diffusion 305 new 1637 oxygenated 694 polar 683 residual 1067 spent 1555 theta 685 use 847 solvent-solute interaction 282 solvent-solvent interaction 282,284,455 solvent-tolerant bacteria 859 solvolysis 297 rate 584,587 solvophobic effect 285 sorption isotherm 177 sorption-desorption hysteresis 177 spatial ordering 419 scale 1191 specific gravity 1057 specification 1068 spectral band 639
Handbook of Solvents
broadening 643 line 639 shift 585,639 spectroscopy dielectric 581 UV 689 spillages 962,1149 spin 430 spinodal 131,148,183-184,190-192,400,408 spin-spin interaction 530 spontaneous abortion 1343 Spriggs law 362 stability 700 stabilization 700 starlike 690 Staverman relation 203 steam-enhanced extraction 1628 stereochemical evidence 789 stereocomplex 578 stereoregularity 578 steric properties 785 repulsion 448 Stockmayer fluid 138,767 model 462 Stockmayer-Casassa relation 180 stoichiometric equilibrium 283 Stokes shift 592,612 storage tanks 962 strain, solvent-tolerant 862 strands 691 stress cracking 1059 tensor 356 stripping 853,950 reactor 1623 styrene 1385 styrene-butadiene copolymer 848 styrene-butadiene-styrene 848 subsurface contamination 1617 suit materials 1633 sulfur dioxide 853 supercritical fluid 2,202 technology 1420 CO2 854 supermolecule approach 674 surface crazing 700 distribution 569 energy 566 free energy 565 inhomogeneity 566 isopotential 653 layers 339 lubricity 956 modifier 565
1673
solubility 1131 tension 14,284,288,301,352,364,567,570, 648,690,898,901-902,1460 surfactant-aided remediation 1625 surgical gloves 956 Svedberg equation 187 swell ratio 309 swelling 128,305,339,565 binary solvents 318 data 327 equilibrium 128,130,189,308,318,344-345, 348 extreme 320 kinetics 310,347 maximum 326 mode 312 pressure 195,346 stage 313 syndiotactic 578 synergistic effect 267 system free energy 132
T tableting 1552 tacticity 778 tautomeric equilibrium 506 tautomerism 1,33 Taylor diffusion coefficient 1432 tear strength 850 TEM 689 temperature 341,349,353,798,852 temperature-shift factor 362 teratogenic 1334 teratogenicity 1132,1336,1341 teratology 1337 termination 779 1,1,2,2-tetrachloroethane 1396,1398 tetrachloroethylene 348,1149,1334,1336,1388, 1396, 1399 tetrachloromethane 2,813 tetrahedric coordination 16 tetrahydrofuran 347,569,686,694,696,700-701, 754,956,1401 texture 1119 TGA 396,1134 theophylline 282 thermal capacity 126 desorption 1626 expansion 190,198 oxidation 1508 thermochromic coefficient 752 thermodynamic 360 equilibrium 148,157-158,176 work of adhesion 565 thermolysis 702 thickness 343,353
1674
Index
change 350 time scale 339 time-temperature superposition 362 titanium dioxide 1559 TLV 1079 toluene 127,185,188,334,341-342,344,347,365, 686,704,794,816,826-827,829-830,847-849, 851,854,881,952,957,959,1328,1334,1336, 1385,1388 tomography 1628 torsional angle 711 Toth isotherm 1437 toxcicodynamics 1315 toxic residues 339 Toxic Substances Control Act 933 toxicokinetics 1315 transesterification 877 transfer reaction 779 transition state 9,786,807 tribological properties 339 1,1,1-trichloroethane 854,914,959,1138,1398 1,1,2-trichloroethane 695,1396 trichloroethylene 854,914,952,1328,1387,1396, 1399 triethylamine 30 trinitrotoluene 1400 Trouton entropy 762 rule 762 tubes 956 TWA 925
U UCST 268-269,272,1424 Uhlig model 284 ultrafiltration 694 ultrasonic absorption 190 unimolecular heterolysis 1 unit cell 755 uptake dermal 1316 routes 1315 UV 788 curing 880 radiation 853 UV/Vis 2
V van der Waals dispersion 469 envelope 673 equation 467 equation of state 1423 forces 16,1477,1509 interaction 451,875 radius 663 repulsion 647
solid 653 sphere 662 surface 21,445,663,721 volume 206 van Oss-Good's expression 565 van’t Hoff parameters 558 vapor data 161 degreasing 1056, 1233 measurement 147,149,154 pressure 157,262,365,1069 solubility 1084 vaporization energy 198 entropy 762,765 heat 15,261 vapor-liquid equilibrium 198 varnishes 344,583,1169,1235 velocity field 356 vents 962 vibronic transitions 589 N-vinylpyrrolidone 703 viscoelasticity 357 viscosity 351,505,515,517,683,844,1069,1432, 1460 intrinsic 268,684,687 measurement 689 Newtonian 684 relative 685 specific 684,692 VOC 854,959,1196,1201,1546 calculation 1069 calculation methods 1205-1209 definition 1202,1241,1267 emission 388,848 exempt 1070 Vogel-Tammann-Fulcher equation 1474 volume excluded 273,689 fraction 308,345,347,350 free 274,340-341 hydrodynamic 272 molar 341-342,345 vorticity 360 Vrentas-Duda theory 353,388
W wash plant 1556 wastes 1163 water 15,137,1149 waterproofing 855 wave function 423 Wertheim pertubation theory 208,750 wettability 570 wetting 496,898,902 index 916 Wilke-Chang equation 38
Handbook of Solvents
wood coatings 1646 workplace monitoring 1079 regulations 925 wound dressing 956
X XPS 565,567,569 X-ray 708,756,770 small angle scattering 147 xylene 127,186,341,696,700-701,847-848,851, 854,881,952,959,1385
1675
Y yellowing 881 Yvon-Born-Green approximation 461
Z Z scale 589 zeolites 495,1514 Zimm plot 183 equation 689 zwitterion 1471
