Polymer Reviews, Volume 50, Issues 1-4 (2010)

Author: Elliot P. Douglas (Editor-in-Chief)

16 downloads 928 Views 20MB Size Report

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!

Report copyright / DMCA form

DOWNLOAD PDF

R Journal of Macromolecular Science , Part C: Polymer Reviews, 50:1–13, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583721003624859

Perspective Scattering from Polymers YVONNE A. AKPALU Department of Chemical & Materials Engineering, University of Cincinnati, Cincinnati, Ohio Knowledge and understanding from scattering provides a scientific basis of control of polymer properties. This understanding has fueled technological advances in synthetic polymers that have revolutionized our daily lives. This issue of Polymer Reviews focuses on recent advances in scattering instrumentation, data analysis and modeling, and its application to the structure-property characterization of polymers. We highlight research directions where the structure-property characterization by scattering measurements can enable polymer products and technologies that significantly reduce reliance on fossil feedstock and environmental pollution. Keywords X-ray scattering, neutron scattering, renewable resources, polyhydroxyalkanoates

1. Introduction In this special issue of Polymer Reviews, we present four articles that review recent advances in structure-property characterization of polymers by X-ray and neutron scattering. Scattering techniques have been employed since the beginning of polymer science to provide information on the structure and properties of polymers.1 As early as the 1920s C.W. Bunn used X-rays to determine the crystal structure of polyethylene via the Bragg law. nλ = 2D sin(θ/2),

(1)

where D is the distance between crystallographic planes, λ, is the wavelength of the radiation used, θ , is the angle of scatter and n is the (integer) order of reflection. The scattering angle θ , is determined by the spatial period of the Fourier component that is responsible for the scattering; thus, for each scattering angle there is a corresponding Bragg spacing, D, which is given by Eq. (1). The scattering intensity I(Q), measured as a function of the momentum transfer vector, Q, is related to θ via Q=

4π sin(θ/2), λ

(2)

Received January 11, 2010; accepted January 13, 2010. Address correspondence to Dr. Yvonne Akpalu, Chemical & Materials Engineering, 400 Rhodes Hall, Cincinnati, OH 45221-0012, United States. E-mail: [email protected]

1

2

Y. A. Akpalu

Combining equations (1) and (2) gives D 2π/Q

(3)

which indicates the distance scale probed by a measurement at a given value of Q. The Fourier or inverse relationship between the structure of a material in real space (r) and the scattering in Q-space, means that Eq. (3) can be applied to first order for all types of scattering. Experiments in the range 0.6 < Q < 15Å−1 , commonly referred to as wideangle scattering (WAS), contain most of the information for determining the unit cell dimensions of crystals.2–10 WAS probes a distance scale ∼ 0.4 < D < 10Å. The technique of small-angle scattering (SAS) is used to study the structure of the size on the order of 10Å or larger.11–16 Information such as the typical size, shape, and arrangement of the structure is contained in the intensity of the scattering X-rays, neutrons, and light at small angles. In general, data from the SAS measurements can provide information on the average size and distribution of the scattering unit or heterogeneity as long as the wavelength of the incident radiation is comparable to the size of the scattering unit or heterogeneity.17 Analysis of the scattering profiles can provide information on the nature of the interfaces, size, shape, and distribution of domains. Furthermore, contrast variation using isotopic substitutions allows one to distinguish between the shape and the spatial correlation of the different polymeric domains, ion-rich or ion-poor, crystalline or amorphous. Most polymer systems exhibit a large-scale structure that necessitates the use of multiple scattering techniques. To illustrate this complexity, we use a semicrystalline polymer as an example. In semicrystalline polymers, the macroscopic behavior is strongly dependent on the underlying microstructure consisting of molecules arranged in the unit cell (∼Å), lamellar crystals (∼10 nm), and the aggregation of these lamellae into fibrils (∼100 nm) and larger structures such as spherulites (∼µm) (Fig. 1). Quantitative relationships between microstructure and properties in these materials requires a knowledge of microstructural features on the scale of lamellar (∼10 nm), fibrils/lamellar stacks (∼100 nm) to spherulites (∼µm). The morphological characterization of crystalline polymers involves determining the unit cell dimensions and the average size of the crystal (lamella) from wide-angle scattering and interlamaller morphology from small-angle scattering.14, 18–21 Determining interfibrillar and interspherulitic parameters requires the use of ultra-small angle X-ray scattering (USAXS)22 and small-angle light scattering (SALS).23–25 Thus determining quantitative structure-property relationships from scattering studies of polymers with large-scale structures or hierarchical microstructures necessitates the use of multiple techniques to span all length scales of structure that influence the properties of the polymer. Many polymers self-organize into hierarchical structures with spatial heterogeneities in the range 10–100 nm. These polymers include block copolymers, ionomers, and liquid crystalline polymers. Scattering experiments give information on the time-averaged structure and conformation of polymer molecules and form the bulk of the large body of work undertaken to characterize the polymer structure, and understand the interrelationships among polymer properties, structure, and morphology. The review articles included in this special issue of Polymer Reviews provide a comprehensive treatment of the principles of small-angle X-ray and neutron scattering techniques as well as recent advances in instrumentation and data analysis and their application to structure-property characterization of polymers. The first contribution by Hammouda focuses on recent advances and applications of neutron scattering for polymer solutions, copolymers, polymer blends, branch or grafted polymers, polymer gels, polymer networks,

Scattering from Polymers

3

Figure 1. Schematic representation of the morphology of semicrystalline polymers and characteristic structural variables. Volume fraction of structures characteristic of each morphological level are the volume fraction of superstructures/spherulites (xs ), volume fraction of lamellar stacks (xL ), and fraction of crystals within lamellar stacks (xCL ). Interlamellar morphological variables are the average distance between crystals (L), the average crystal thickness (lc ), and the average amorphous thickness (la ). Interfibrillar morphological variables are the average size of the lamellar stack (ξ L ) and the interfibrillar amorphous regions (LD ). The crystalline fraction within spherulites (xcs = xL xCL ) is an averaged nanoscale quantity. The assumption here is that all crystals are within lamellar stacks and spherulites.

polymer micelles, polymeric nanomaterials, and polymer membranes. The prospects for the measurement capabilities that will allow probing of polymer structures from the near atomic scale to well into the optical (20 micrometers) size scale are described. Zhang and Ilavksy focus on the application of ultra-small angle scattering for probing polymers with structural heterogeneities in the size range of 1–1000 nm. The review focuses on USAXS structure-property characterization of polymer nanocomposites, polymer gels and solutions, polymer blends, polymer micelles, and microemulsions. New advances in instrumentation that support the wider use of USAXS for polymer research, including new capabilities for measuring the “complete” small-angle scattering curve for polymers are described. The next two contributions focus on recent progress in structure-property characterization of polymers with fiber symmetry. Stribeck provides a critical review of the experimental methods and data analysis required for monitoring fabrication processes, mechanical properties, and the resulting fluctuations in polymeric materials with fiber symmetry. Burger, Hsiao, and Chu provide a theoretical treatment of structural information to be determined in scattering from natural and synthetic polymer fiber systems. This review emphasizes the calculation of complete X-ray scattering patterns required for building structure-property relationships in natural and synthetic fiber polymers or polymers with self-assembled meso-structures exhibiting density and orientation fluctuations that can be described by fiber symmetry.

2. X-rays and Neutrons The physics of X-ray and neutron scattering from polymers are covered in several standard texts.17,26,27 Here, we borrow heavily from these texts to present aspects that are important

4

Y. A. Akpalu

for understanding the importance of X-ray and neutron scattering for characterizing polymers. For most applications in polymer science, the scattering of X-rays and neutrons is elastic, where the energies of the incident and scattered radiation have the same energy or wavelength. X-rays and neutrons are scattered by atomic centers at discrete angles represented as sinusoidal (Fourier) components of the electron density and nuclear scattering potential of the specimen, respectively. X-rays and neutrons have wavelengths comparable to interatomic distances in materials. The manner in which these types of electromagnetic radiation are scattered by a material depends on the mechanism of scattering from individual atoms and on their relative positions in space. X-rays are electromagnetic radiation with wavelength, λ = 10−2−102Å. X-rays used for the study of the structure of materials have typical wavelengths of 0.5–2.5 Å and are most typically generated by conventional anode generators that offer the advantage of inhouse capabilities found in individual laboratories worldwide, as well as the bright light of synchrotron sources that are available only at national facilities.14 Studies on polymers are performed mostly with Kα characteristic radiation from a copper target tube having a wavelength of 1.5418 Å, but occasional work is also done with Kα line of wavelength 0.7107 Å from a molybdenum target tube. X-rays of similar wavelength can also be selected by means of a monochromator from a broad spectrum emitted by a synchrotron radiation source.14 X-rays, like light, exhibit particle-wave duality. Certain properties of X-rays are better understood when a beam of X-rays is regarded as a stream of photons rather than a wave with wavelength λ and frequency ν. The energy of an X-ray photon is characterized by its energy E and momentum p, which are related to λ and ν by E = hν

(4)

h p= λ

(5)

c v

(6)

and λ=

where c is the speed of light (= 2.998 × 108 m/s), and h is Planck’s constant (= 6.626 × 10−34 J s). The flux of photons produced by X-ray synchrotrons is several orders of magnitude higher than the flux on a neutron beamline. The increased flux can be very beneficial when collimating the beam to a spot size of a few millimeters in diameter and for increasing the experimental throughput based on reduced exposure time needed.16, 28 A neutron is an uncharged elementary particle with a mass m = 1.675 × 10−24 g and spin 12 . It kinetic energy E and momemtum p are E=

1 2 mv 2

(7)

and p = mv

(8)

where ν is its velocity. Neutrons also exhibit wave-like behavior, with the wavelength given by the de Broglie relation λ=

h h = p mv

(9)

Scattering from Polymers

5

Table 1 Typical values of ν, E, and λ of neutrons from Cold, Thermal and Hot Sources17 Polymer

Cold

Thermal

Hot

T (K) v (m/s) E (meV) λ (Å)

25 642 2.16 6.16

330 2333 28.4 1.696

2000 5743 172 0.689

Table 1 shows the most probable velocity v in the Maxwell-Boltmann Distribution, given by Eq. (11). The corresponding kinetic energy E = mv 2 /2 = kT , and wavelength λ, are listed for the three typical moderator temperatures 25, 330, and 2000 K. Cold source neutrons emerge from a small volume (∼20 liters) of liquid deuterium maintained around 25 K while thermal neutrons are those moderated usually with heavy water D2 0 around 330 K. It is worthwhile to note that the wavelengths of cold, thermal, and hot neutrons are on the order of 1 Å, similar to X-rays. As a result, neutron scattering is also a useful tool for investigating the structure of materials. The way neutrons are produced determines the energy and wavelength. The source of neutrons for most scattering experiments is a nuclear reactor, although spallation sources have gained importance in recent years. Neutrons produced by a nuclear fission reaction in a reactor or by bombardment of high-energy protons onto a heavy metal in a spallation source are of very high velocities. For neutron scattering studies these high velocity neutrons are moderated, i.e. they are allowed to slow down by repeated collisions with atoms in a moderating material. Moderation produces neutrons with a Maxwell-Boltzmann velocity distribution, given by m 32 1 2 (10) v 2 exp − 2 mv kT f (v) = 4π 2π kT where f (v)dv is the fraction of gas molecules with velocities between v and v + dv and k is Boltzmann’s constant (1.381 × 10−23 J/K). The maximum function of f (v) or most probable velocity v occurs at v=

2kT m

12 (11)

In many ways the scattering behavior of neutrons is similar to those exhibited by X-rays, so that experimental and theoretical tools developed for X-rays can be applied to neutron scattering and vice versa. There are, however, some important differences between X-rays and neutrons, and these differences often make the two methods complementary to each other, providing the required information for characterizing polymer structure and its relation to properties. 2.1 Energy The difference in energy between X-rays and neutrons determine what kind of structure is probed. Whereas the energy of an X-ray photon is on the order of 10 keV, the kinetic energy of a thermal neutron is of the order of 10 meV. The average energy associated

6

Y. A. Akpalu

with the motion of atoms, arising from vibrational, rotational, and translational motions of molecules, is of the order of kT. At ambient temperatures, kT is about 20 meV. Thus, when X-rays are scattered by matter, even when there is an exchange of energies between the motions of atoms and the X-ray photon, the energy of the photon is scarcely affected. On the other hand, when neutrons are scattering inelastically, their energies can be modified to an appreciable extent that can be measured experimentally. This difference can be understood also from a slightly different viewpoint as follows. The time (τ = ν1 ) associated with one wave period is of the order of 10−19 s for X-rays and 10−13 s for thermal neutrons. Since a typical time period for atomic motions is 10−13 s, an X-ray, unlike a neutron, does not see a change in the atomic position. Measuring the inelastic scattering of neutrons is a very useful method for investigating the motions of atoms in materials,30,31 which is beyond the focus of this special issue. 2.2 Mechanism for Interaction of Radiation with Matter The differences in the mechanism by which incident neutrons and X-rays interact with a material leads to several important differences in how the experimental data is obtained and corrected. X-rays are scattered by the electron density of an atom or molecule, and the scattering cross-section of an atom increases in direct proportion to the square of the number of electrons or atomic number, Z; in the case of hydrocarbon polymers the X-rays “see” the electron clouds contributed by carbon’s six electrons better than the single electron attributable to hydrogen. X-rays probe atomic dimensions within an order of magnitude of the X-ray wavelengths, so that the radiation scattered by the electron cloud on opposite sides of the atom results in a different path length that gives rise to a shift in phase and decreasing the scattering power with increasing scattering angle. Neutrons interact directly with the nuclei within a molecule, and the strength of the scattering interaction varies irregularly with the atomic number, so that even isotopes of the same element do not have the same neutron scattering cross-section or scattering length.17 For example, the most significant isotopic variation occurs for hydrogen, which has a coherent scattering length of −3.74 fm, while for deuterium the scattering length is 6.67 fm. Neutrons are therefore sensitive to hydrogen and the differences between its isotopes, which permits observation and measurement of the hydrogen structural correlations in polymers, that are not easily obtainable by X-rays. Scattering experiments probe the differential scattering cross-section defined as the ratio of the scattering cross-section dσ scattered into the solid angle d about the scattering angle θ . This can be analyzed in terms of the first Born approximation32 N dσ = bi bj e(iQ·rij ) d i,j

(12)

where the sum is over the N nuclei (in the case of neutrons) or electrons (in the case of X-rays) in the sample, b is the scattering length for neutrons of a given element, while b is replaced by a Q-dependent form factor in the case of X-rays; the {rij } are the positions of nuclei, electrons, and heterogeneities larger than atomic dimensions; Q = 4π sin(θ/2)/λ is the momentum transfer for the elastic scattering process where λ is the wavelength; and the brackets correspond to a thermal average in Eq. (12). The specific form of Eq. (12) depends on (i) the scattering length of the heterogeneity, (ii) the relative size of the heterogeneity compared to the probe radiation wavelength, and (ii) the spatial arrangement of the heterogeneities.17 When the scatterers are numerous (e.g., electrons on every atom

Scattering from Polymers

7

for X-rays) and are more or less continuously dispersed in space in the sample, we replace the summation in Eq. (12) with an integral. In applying Eq. (12) and its simplified forms, it is required that various corrections to the measured scattering intensity, I (Q), has been made, accounting for effects such as incoherent scattering, beam polarization, multiple scattering, inelastic effects, container absorption, etc. Details on how to perform these corrections correctly for X-ray and neutron scattering experiments are given in standard texts.16,17,26 The articles included in this special issue of Polymer Reviews provide examples of simplified forms of Eq. 12 for different types of scattering from polymers. In a typical scattering experiment, the scattered radiation signal is captured by a detector, or detector element, of dimensions dx × dy positioned at some distance, L, and the scattering angle from the sample. This detector records the flux of radiation scattered . Single-point detectors have been used to collect into a solid angle element, d = dxdy Lz X-ray structural information from polymers, but area detectors offer several important advantages over single-point detectors, including the reduction of the background signal and greater statistics, a larger range of Q-space data collected at the same time, and the collection of several perspectives of the same data that provides an important benchmark for validation of the subsequent data processing and modeling. The use of charge-coupled device (CCD) area detectors for X-ray diffraction began around 1995 and has become increasingly popular.14 The detection of neutrons is typically accomplished through an array of individual detectors (although one- and two-dimensional linear and area detectors are sometimes used) composed of a gas of 3He, for example, or scintillator materials based on 6Li that detect the neutron as a charge produced from a nuclear reaction.17 The primary issue in devising a neutron detector is to create high sensitivity to neutrons while remaining insensitive to background events (such as γ -rays) and to minimize the loss of signal due to the “dead-time” of the detector. Gas detectors have the advantage of good discrimination against γ -rays, while scintillator detectors have better sensitivity relative to gas detectors, with a dead time on the order of hundreds of nanoseconds.17 Scattering yields measurements in reciprocal (Fourier Transform) space and depends therefore on data interpretation using models33 and not on real space imaging like microscopy. Electron microscopic imaging is in principle more powerful than small-angle scattering (SAS) for elucidating nanoscale structure and morphology. The main reason is that the phase information is lost in scattering, so one cannot uniquely determine structure. Although the loss of phase information can be viewed as a severe limitation, the loss can be beneficial34 for understanding the spatial dependence of fluctuations in polymers arising from heterogeneities in backbone structure of polymers and mesoscale morphologies. The articles included in this special issue of Polymer Reviews provide several examples of the unique scientific benefits of scattering from polymers.

3. Renewable Polymers with Controlled Properties Within the last few decades, synthetic polymers have revolutionized our daily lives.35 Globally, we use in excess of 260 million tons of plastic per year, accounting for about 8 percent of world oil production. The dwindling of fossil resources, coupled with increasing public preference for environmentally friendly plastics, has increased academic and industrial interest in biodegradable polymers prepared from renewable sources. Biopolymers differ from petroleum-based in that their feedstock is from renewable biomass rather than being oilbased. These polymers may be natural polymers (e.g., cellulose), synthetic polymers made from biomass monomers (e.g., polyactic acid), or synthetic polymers made from synthetic

8

Y. A. Akpalu

monomers derived from biomass (e.g., polyethylene derived from bioethanol).36 Although materials with functionality comparable to conventional plastics can now be produced on an industrial scale; they are more expensive than conventional polymers and account for less than 1% of plastics production.36 Below, we highlight key structure-property characterization challenges of recently developed cost-competitive renewable polymers. These renewable polymers have the potential to replace petroleum-based polymers on scales that can lead to significant reductions in reliance on fossil feedstock and environmental pollution. Recently, polyhydroxyalkanoates (PHAs), which are biodegradable and compostable thermoplastics polyesters synthesized by bacteria, were introduced to the market as competitors for polyethylene and polypropylene.37–39 These materials are particularly interesting as one looks forward to the next 10–50 years since they can replace polymers based on fossil feedstock without loss of performance. Further, adding small amounts of nanofillers (<5 wt.%) can enhance barrier, mechanical, and the thermal properties of polymers.40,41 Using such methods, it is possible to create renewable polymer nanocomposites with properties that rival available synthetic polymers.42,43 Environmental benefits include the reduction of solid packaging waste because PHAs fully degrade in aerobic (e.g., household compost) and anaerobic (e.g., marine, septic tanks, below the surface of rice paddies) environments37 which support the bioactivity of microbes. Thus commercial products made of PHA will not spontaneously biodegrade in the typical environment of store shelves. Several recent life cycle assessments show positive impacts of PHAs on the environment such as reduction in greenhouse gas emissions and lower total fossil energy requirement compared to petrochemical counterparts.44–46 When PHAs are produced from food and agricultural waste, 0.49 kg CO2 is emitted per kg of resin, compared with 2–3 kg CO2 per kg of resin of petrochemical counterparts. This difference represents about 80% reduction in greenhouse gas emissions. The fossil energy requirement per kg of PHA is about 50% lower than that of petrochemical counterparts (78–88 MJ/kg resin). Potential green applications abound—window coatings that block heat but not light, more efficient solar panels, and light-weight, fully biodegradable automobile parts and biofuels from wastewater or sludge that do not compete with food sources and arable land.39 While the potential of PHAs is recognized in the literature and has even been realized in some cases, the knowledge of these systems is decades behind that of synthetic polymers. Composites based on PHAs, furthermore, are just emerging in the research community.40,41 NodaxTM PHA copolymers (Fig. 2) originally developed by Procter & Gamble (P&G)38,47 are designed to achieve mechanical properties and processing characteristics comparable to those of thermoplastic polyolefins such as polyethylene and polypropylene. NodaxTM copolymers (Fig. 2) consist predominantly of 3-hydroxybutryate (3HB) and 3hydroxyalkanoate (3HA) co-monomer units. The secondary 3HA comonomer units must

Figure 2. Molecular structure of NodaxTM class PHA copolymer. The value of y (mcl-3HA units) is between 2 to 50%. Chemical structures of more familiar linear PHAs: PHB is poly(3-hydroxybutyrate) (y = 0) and PHBV is poly (3-hydroxybutyrate-co-3-hydroxyvalerate) (n = 1).

Scattering from Polymers

9

have side groups consisting of at least three carbon atoms. Examples of PHAs with mediumchain-length (mcl) side groups include poly (3-hydroxybutyrate-co-3- hydroxyhexanoate) (PHBHx, n = 2), poly (3-hydroxybutyrate-co-3- hydroxyoctanoate) (PHBO, n = 4), poly (3-hydroxybutyrate-co-3-hydroxydecanoate) (PHBD, n = 6), and poly (3-hydroxybutyrateco 3-hydroxyoctadecanoate) (PHBOd, n = 14). The architecture of NodaxTM copolymers is substantially different from that of more familiar types of PHAs, such as poly(3-hydroxybutyrate) homopolymer (PHB) or poly(3hydroxybutyrate-co-3-hydroxyvalerate) copolymer (PHBV). The size of the side groups in the conventional PHAs is limited to short-side-chain (scl) types with no more than two carbon atoms. While PHAs with only one or two carbon side groups may be viewed essentially as linear polymers, PHA copolymers with medium-chain-length (mcl) side groups are moderately branched polymers. The applications of PHB and PHBV are limited by brittle behavior caused by their high crystallinity, poor thermal stability, and narrow processing window. To improve the overall physical properties of PHB, PHAs containing over 125 types of monomers have been harvested from different microorganisms.48 The copolymers show a wide range of physical properties depending on the chemical structure of the comonomer units as well as comonomer composition. The inclusion of a small amount of mcl-3HA in PHB gives rise to polymers with useful attributes, including polyolefin-like thermo-mechanical properties, polyester-like physicochemical properties and interesting biological activity37,49–53 not achieved with PHB or PHBV. The incorporation of mcl-3HA units effectively lowers the crystallinity and melting temperature (Tm ). The Tm of mcl-3HA copolymers of PHA can be lowered well below the thermal decomposition temperature of PHB and PHBV to make this material much easier to process. The stiffness of PHA copolymers can be controlled by varying the fraction of the mcl-3HA comonomer units49 in the copolymer (Fig. 3).

Figure 3. Young’s modulus of PHA copolymers. The value of Young’s modulus of PHA spans between that of very stiff polymers, like polylactic acid (PLA) and PP, and much softer material, like low density polyethylene (LDPE). PHB-Hexanoate is PHBHx, PHB-Octanoate is PHBO and PHB-Octadecanoate is PHBOd. Bionolle 1001 and 3001 are PHBs from Biomer. Reproduced with permission from Reference 49.

10

Y. A. Akpalu

NodaxTM copolymers can be converted into various forms and products, such as films, sheets, nonwovens, molded articles, pulps, powders, coating materials, laminates, and composites, using conventional processing steps like extrusion and thermo-forming. Large-scale commercial production of the PHA copolymers (600 million pounds/year) is scheduled to start in this year. Both industrial and consumer product applications are anticipated. Factors important to the successful large-scale manufacture of PHAs for different applications include the 3HA composition of copolymers, average molecular weights, and molecular weight distributions, and the presence of processing aids like nucleating agents and plasticizers. The reduced crystallinity of mcl-3HA copolymers of PHA provides the ductility and toughness required for many practical applications.37–38 Although the copolymerization improves toughness, it also decreases stiffness. For example, the mechanical properties of PHB and PHBV are similar to isotactic polypropylene, but they show much less ductility and impact strength (low strain at break) to replace commercial PP effectively (Table 2). Incorporating 12 mol% Hx into PHB provides PP impact properties while eroding stiffness and strength. Matching ductility and impact properties of polyethylene without significantly decreasing the strength and stiffness also presents new research challenges. Biotechnology opens new paths for synthesizing biopolymers with functional properties.55–58 To date, over 120 poly(hydroxyalkanoate) (PHA) monomer units with different side chains have been reported.48 The side chains can be alkyl, aromatic, unsaturated, halogenated, expoxidized, and branched. Substituents in the side chains of PHAs can also be modified chemically, for instance by cross-linking unsaturated bonds5,9–60 Variation in the length and the composition of side chains and the ability to modify their reactive substituents is the basis for the diversity of the PHA polymers and their vast array of potential applications. Structure-property relations from scattering measurements are required to achieve control of properties at costs competitive with petroleum-based polymers. Thus we argue that widespread adoption of renewable polymer based materials can only be achieved through knowledge derived from complete characterization of the spatial dependence of fluctuations that arise from heterogeneities in the backbone structure and mesoscale morphologies of PHAs, not only the mean distances between heterogeneities but also the spatial and orientational distribution of these elements. Thus control of properties of renewable polymers and the development of cost-effective manufacturing technologies will require information on the nanoscale (1–100 nm), but also on larger scales

Table 2 Mechanical properties of PHAs and petroleum based polymers: PP is isotactic polypropylene, HDPE is high density polyethylene and LDPE is low density polyethylene. Source Polymer Handbook Polymer PHB PHBV (7 mol% HV) PHBHx (∼12 mol% HV) PP HDPE LDPE

Young’s Modulus (GPa)

Tensile Strength (MPa)

Strain at break (%)

2.5 1.4 0.27 0.6–1.2 0.7–1.2 0.15–0.45

36 24 / 22 25–32 15–20

2.5 2.8 11.5 12–20 600–900 600

Scattering from Polymers

11

(>100 nm) readily assessable by the combined use of small-angle and ultra-small angle scattering techniques.

Acknowledgements This research is supported by the Department of Energy Basic Energy Science under contract DE-SC0002253. The author is indebted to Prof. Elliot Douglas and the editorial staff at Polymer Reviews. Without their guidance and patience this Special Issue of Polymer Reviews would not have been possible.

References 1. Flory, P. J. Principles of Polymer Chemistry; Cornell University Press: Ithaca, 1953. 2. Alexander, L. E. X-ray Diffraction Methods in Polymer Science; Wiley-Interscience: New York, 1969. 3. Kakudo, M.; Kasai, N. X-ray Diffraction by Polymers; Elsevier: Amsterdam, 1972. 4. Klug, H. P. X-ray Diffraction Procedures for Polycrystalline and Amorphous Materials, 2nd Edition; Wiley: New York, 1974. 5. Balt´a-Calleja, F. J. X-ray Scattering of Synthetic Polymers; Elsevier: Amsterdam, 1989. 6. Glusker, J. P. Crystal Structure Analysis: A Primer, 2nd Edition; Oxford University Press: New York 1985. 7. Glusker, J. P. Crystal Structure Analysis for Chemists and Biologists; VCH: New York, 1994. 8. Tadokoro, H. Structure of Crystalline Polymers; Wiley: New York, 1979. 9. Hosemann, R.; Bagchi, S. N. Direct Analysis of Diffraction by Matter; North-Holland Amsterdam, 1962. 10. Vainshtein, B. K. Diffraction of X-rays by Chain Molecules; Translation rev. and enl. by the author ed.; Elsevier: New York, 1966. 11. Guinier, A.; Fournet, G. Small-Angle Scattering of X-rays; Wiley: London, 1955. 12. Glatter, O.; Kratky, O. Small-Angle X-ray Scattering; Academic Press: New York, 1982. 13. Higgins, J. S.; Benoit, H. Polymers and Neutron Scattering; Clarendon Press, 1994. 14. Chu, B.; Hsiao, B. S. “Small-angle X-ray scattering of polymers,” Chem. Rev. 2001, 101, 1727–1761. 15. Melnichenko, Y. B.; Wignall, G. D. “Small-angle neutron scattering in materials science: Recent practical applications,” Journal of Applied Physics 2007, 102, 021101–24. 16. Stribeck, N. X-ray Scattering of Soft Matter; Springer-Verlag: Berlin Heidelberg, 2007. 17. Roe, R.-J. Methods of X-Ray and Neutron Scattering in Polymer Science; Oxford University Press: Oxford, UK, 2000. 18. Jokela, K.; Vaananen, A.; Torkkeli, M.; Starck, P.; Serimaa, R.; Lofgren, B.; Seppala, J. “Differential scanning calorimetry, small-angle X-ray scattering, and wide-angle X-ray scattering on homogeneous and heterogeneous ethylene-alpha-copolymers,” J. Polym. Sci. Pt. B-Polym. Phys. 2001, 39, 1860–1875. 19. Akpalu, Y.; Kielhorn, L.; Hsiao, B. S.; Stein, R. S.; Russell, T. P.; van Egmond, J.; Muthukumar, M. “Structure development during crystallization of homogeneous copolymers of ethene and 1-octene: Time-resolved synchrotron X-ray and SALS measurements,” Macromolecules 1999, 32, 765–770. 20. Akpalu, Y. A.; Amis, E. J. “Evolution of density fluctuations to lamellar crystals in linear polyethylene,” J. Chem. Phys. 1999, 111, 8686–8695. 21. Akpalu, Y. A.; Amis, E. J. “Effect of polydispersity on the evolution of density fluctuations to lamellar crystals in linear polyethylene,” J. Chem. Phys. 2000, 113, 392–403. 22. Owen, A.; Bergmann, A. “On the fractal character of polymer spherulites: An ultra-small-angle X-ray scattering study of poly[(R)-3-hydroxybutyrate],” Polymer International 2004, 53, 12–14.

12

Y. A. Akpalu

23. Haudin, J. M. “Optical Studies of Polymer Morphology” In Optical Properties of Polymer; Meeten, G. H., Ed.; Elsevier Applied Science Publishers: New York, 1986; pp 167–264. 24. Akpalu, Y. A.; Lin, Y. Y. “Multivariable structural characterization of semicrystalline polymer blends by small-angle light scattering,” J. Polym. Sci. Pt. B-Polym. Phys. 2002, 40, 2714–2727. 25. Li, Y.; Akpalu, Y. A. “Probing the melting behavior of a homogeneous ethylene/1-hexene copolymer by small-angle light scattering,” Macromolecules 2004, 37, 7265–7277. 26. Lindner, P.; Zemb, T. Neutron, X-ray and Light Scattering: Introduction to an Investigative Tool for Colloidal and Polymeric Systems; North-Holland, 1991. 27. Cebe, P. “Introduction to scattering from polymers” In Scattering from Polymers— Characterization by X-rays, Neutrons, and Light; Cebe, P.; Hsiao, B. S.; Lohse, D. J., Eds., 2000; Vol. 739; pp 2–22. 28. Kolb, R.; Wutz, C.; Stribeck, N.; von Krosigk, G.; Riekel, C. “Investigation of secondary crystallization of polymers by means of microbeam X-ray scattering,” Polymer 2001, 42, 5257–5266. 29. Wignall, G. D.; Melnichenko, Y. B. “Recent applications of small-angle neutron scattering in strongly interacting soft condensed matter,” Reports on Progress in Physics 2005, 68, 1761–1810. 30. Kanaya, T.; Kaji, K. “Dynamics in the glassy state and near the glass transition of amorphous polymers as studied by neutron scattering” In Polymer Physics and Engineering, 2001; Vol. 154; pp 87–141. 31. Ding, Y. F.; Novikov, V. N.; Sokolov, A. P.; Cailliaux, A.; Dalle-Ferrier, C.; Alba-Simionesco, C.; Frick, B. “Influence of molecular weight on fast dynamics and fragility of polymers,” Macromolecules 2004, 37, 9264–9272. 32. Egelstaff, P. A. An Introduction to the Liquid State, 2nd ed.; Oxford University Press: Oxford, New York, 1992. 33. Ilavsky, J.; Jemian, P. R. “Irena: tool suite for modeling and analysis of small-angle scattering,” Journal of Applied Crystallography 2009, 42, 347–353. 34. Schaefer, D. W.; Agamalian, M. M. “Ultra-small-angle neutron scattering: a new tool for materials research,” Curr. Opin. Solid State Mat. Sci. 2004, 8, 39–47. 35. Thompson, R. C.; Swan, S. H.; Moore, C. J.; vom Saal, F. S. “Our Plastic Age,” Philosophical Transactions of the Royal Society of London Series B 2009, 364, 1973–1976. 36. Song, J. H.; Murphy, R. J.; Narayan, R.; Davies, G. B. H. “Biodegradable and compostable alternatives to conventional plastics,” Philosophical Transactions of the Royal Society B-Biological Sciences 2009, 364, 2127–2139. 37. Noda, I.; Green, P. R.; Satkowski, M. M.; Schechtman, L. A. “Preparation and properties of a novel class of polyhydroxyalkanoate copolymers,” Biomacromolecules 2005, 6, 580–586. 38. Noda, I.; Lindsey Blake, S.; Caraway, D. “Nodax(tm) Class PHA Copolymers: Their Properties and Applications” In Plastics from Bacteria: Natural Functions and Applications; Chen, G.-Q., Ed.; Springer-Verlag: Berlin Hiedelberg, 2010; Vol. 14; pp 237–255. 39. Chen, G. Q. “A microbial polyhydroxyalkanoates (PHA) based bio- and materials industry,” Chem. Soc. Rev. 2009, 38, 2434–2446. 40. Maiti, P.; Batt, C. A.; Giannelis, E. P. “New biodegradable polyhydroxybutyrate/layered silicate nanocomposites,” Biomacromolecules 2007, 8, 3393–3400. 41. Xie, Y.; Kohls, D.; Noda, I.; Schaefer, D. W.; Akpalu, Y. A. “Poly(3-hydroxybutyrate-co-3hydroxyhexanoate) nanocomposites with optimal mechanical properties,“ Polymer 2009, 50, 4656–4670. 42. Pandey, J. K.; Kumar, A. P.; Misra, M.; Mohanty, A. K.; Drzal, L. T.; Singh, R. P. “Recent advances in biodegradable nanocomposites,” J. Nanosci. Nanotechnol. 2005, 5, 497–526. 43. Yu, L.; Dean, K.; Li, L. “Polymer blends and composites from renewable resources,” Prog. Polym. Sci. 2006, 31, 576–602. 44. Akiyama, M.; Tsuge, T.; Doi, Y. “Environmental life cycle comparison of polyhydroxyalkanoates produced from renewable carbon resources by bacterial fermentation,” Polymer Degradation and Stability 2003, 80, 183–194. 45. Harding, K. G.; Dennis, J. S.; von Blottnitz, H.; Harrison, S. T. L. “Environmental analysis of plastic production processes: Comparing petroleum-based polypropylene and polyethylene with

Scattering from Polymers

46.

47. 48. 49.

50. 51. 52.

53.

54. 55. 56.

57. 58. 59. 60.

13

biologically-based poly-beta-hydroxybutyric acid using life cycle analysis,” J. Biotechnol. 2007, 130, 57–66. Yu, J.; Chen, L. “The greenhouse gas emissions and fossil energy requirement of bioplastics from cradle to gate of a biomass refinery,” Environmental Science & Technology 2008, 42, 6961–6966. Noda, I. “Biodegradable copolymers and plastic articles comprising biodegradable copolymers.” 1996, US: 5,498,692 Yoshie, N.; Menju, H.; Sato, H.; Inoue, Y. “Complex composition distribution of poly(3hydroxybutyrate-co-3-hydroxyvalerate),” Macromolecules 1995, 28, 6516–6521. Satkowski, M. M.; Melik, D. H.; Autran, J.-P.; Green, P. R.; Noda, I.; Schechtman, L. A. “Physical and processing properties of polyhydroxyalkanoate copolymers.” In Biopolymers; Doi, Y. and Steinbuchel, A., Eds.; Wiley-VCH: Weinheim, 2001; Vol. 3b; pp 231–263. Noda, I.; Satkowski, M. M.; Dowrey, A. E.; Marcott, C. “Polymer alloys of Nodax copolymers and poly(lactic acid),” Macromol. Biosci. 2004, 4, 269–275. Poliakoff, M.; Noda, I. “Plastic bags, sugar cane and advanced vibrational spectroscopy: taking green chemistry to the Third World,” Green Chemistry 2004, 6, G37–G38. Noda, I.; Bond, E. B.; Green, P. R.; Melik, D. H.; Narasimhan, K.; Schechtman, L. A.; Satkowski, M. M. “Preparation, properties, and utilization of biobased biodegradable Nodax(tm) copolymers.” In Polymer Biocatalysis and Biomaterials; Cheng, H. N. and Gross, R. A., Eds.; American Chemical Society: Washington, DC, 2005; pp 280–291. Federle, T. W.; Barlaz, M. A.; Pettigrew, C. A.; Kerr, K. M.; Kemper, J. J.; Nuck, B. A.; Schechtman, L. A. “Anaerobic biodegradation of aliphatic polyesters: Poly(3-hydroxybutyrateco-3-hydroxyoctanoate) and poly(epsilon-caprolactone),” Biomacromolecules 2002, 3, 813–822. Peacock, A. J. Handbook of Polyethylene: Structures, Properties and Applications; Marcel Dekker: New York, 2000. Lenz, R. W.; Marchessault, R. H. “Bacterial polyesters: Biosynthesis, biodegradable plastics and biotechnology,” Biomacromolecules 2005, 6, 1–8. Qiu, Y. Z.; Ouyang, S. P.; Shen, Z. Y.; Wu, Q.; Chen, G. Q. “Metabolic engineering for the production of copolyesters consisting of 3-hydroxybutyrate and 3-hydroxyhexanoate by Aeromonas hydrophila,” Macromol. Biosci. 2004, 4, 255–261. Nakamura, K.; Goto, Y.; Yoshie, N.; Inoue, Y. “Biosynthesis of Poly(3-Hydroxyalkanoate) from Amino-Acids,” International Journal of Biological Macromolecules 1992, 14, 321–325. Madison, L. L.; Huisman, G. W. “Metabolic engineering of poly(3-hydroxyalkanoates): From DNA to plastic,” Microbiol. Mol. Biol. Rev. 1999, 63, 21. Gagnon, K. D.; Lenz, R. W.; Farris, R. J.; Fuller, R. C. “Chemical Modification of Bacterial Elastomers .1. Peroxide Cross-Linking,” Polymer 1994, 35, 4358–4367. Gagnon, K. D.; Lenz, R. W.; Farris, R. J.; Fuller, R. C. “Chemical Modification of Bacterial Elastomers .2. Sulfur Vulcanization,” Polymer 1994, 35, 4368–4375.

R Journal of Macromolecular Science , Part C: Polymer Reviews, 50:14–39, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583720903503460

Reviews SANS from Polymers—Review of the Recent Literature BOUALEM HAMMOUDA National Institute for Standards and Technology, Center for Neutron Research, Gaithersburg, MD This paper reviews the recently published literature on Small-Angle Neutron Scattering (SANS) from polymers. Papers published over the past three years and resulting from the use of the NIST Center for Neutron Research (NCNR) are included. Those with which this author is most familiar are summarized in a brief format. The intent of this review paper is to demonstrate the usefulness of the SANS technique and its impact on polymer research. SANS is a structural characterization method and a good probe for miscibility thermodynamics in polymer mixtures. SANS topics covered include polymer solutions, copolymers, polymer blends, branched or grafted polymers, polymer gels, polymer networks, polymer micelles, polymeric nanomaterials, and polymer membranes. Keywords small range neutron scattering, polymer structures, nanoscale, polymer solutions, copolymers, blends, gels, networks, branched polymers

1. The SANS Technique Small-angle neutron scattering (SANS) is an effective characterization method to investigate nanoscale structures. It is based at neutron scattering facilities and has experienced steady growth over the past thirty years. It probes structures with sizes from the near atomic to the near micrometer scale and has had impact in many research areas including polymers, complex fluids, biology, and materials science. The partial deuteration method (which consists of replacing hydrogen by deuterium atoms) gives the SANS technique unique advantage. Like other scattering methods, SANS yields measurements in the reciprocal (Fourier transform) space and depends therefore on data interpretation using models and not on direct space imaging like microscopy. The SANS instrument uses the following basic steps: 1. monochromation, 2. collimation, 3. scattering and 4. detection (Fig. 1). Monochromation consists of producing a monochromatic neutron beam from the Maxwellian neutron source spectrum and is performed using a velocity selector. Collimation is performed using a source aperture and a sample aperture in order to define an Received June 18, 2009; accepted October 27, 2009. Address correspondence to Dr Boualem Hammouda, Center for Neutron Research, 100 Bureau Drive, Gaithersburg 20899-6102, United States. E-mail: [email protected]

14

SANS from Polymers

15

Figure 1. Schematics of a SANS instrument. This Figure is not to scale; the total horizontal size is 30 m whereas the neutron detector height is 64 cm.

incident neutron beam with very small divergence. Scattering from samples of various forms (liquids, solids, gels, etc.) and in various environments (heating, pressure, shear, applied magnetic field, etc.) is measured in special cells. Detection of the scattered neutrons is performed using a 2D area sensitive detector. The pre-sample and the post-sample flight paths can be adjusted between 1 m and 15 m distance. The overall size of a SANS instrument is typically 30 m. The sample thickness is between 1 mm and 2 mm making the SANS technique a bulk probe. The NIST Center for Neutron Research (NCNR) facility operates two 30 m SANS instruments at the core of a thriving user program. Research on SANS from polymers constitutes the most active component.

2. SANS Data Analysis and Modeling Various aspects of the SANS technique including data analysis methods can be found in a recent book available online.1 SANS data analysis consists of one of three methods. 1. Rapid interpretation using standard (linear) plots such as the Guinier plot (to obtain a radius of gyration) or the Porod plot (to obtain a Porod exponent). Porod exponents vary between 1 (for 1D object such as a rod) and 4 (object with smooth surface). For example, the Porod exponent of a polymer coil in a good solvent is 5/3 while that for a polymer coil in poor solvent is 3. A Porod exponent of 2 characterizes either a polymer coil in theta solvent or a 2D structure (such as a lamella). This first data analysis method is used routinely. Extensive description of linear plots can be found in Chapter 22 of the SANS Toolbox.1 2. Nonlinear least-squares fitting to appropriate models. A large number of models are available for the analysis of SANS data from polymer systems. These are either for macromolecular scattering such as the Random Phase Approximation (RPA) or for particulate scattering such as the analytical solutions of the Ornstein-Zernike (OZ) equation. Polymer solutions and homogeneously mixed blends are well described by the RPA model while the various microphases in block copolymers (spherical, cylindrical, lamellar, etc.) are best described by solutions to the OZ equation. One such solution, the Percus-Yevick equation, offers a simple analytical form for hard-sphere interaction potential between spherical particles. The Mean Spherical Approximation is another analytical solution of the OZ equation for charged particulate systems. The zero average contrast method consists of using deuterated and non-deuterated polymer mixtures as well as deuterated and nondeuterated solvent

16

B. Hammouda mixtures in order to isolate the single-chain form factor even from concentrated polymer solutions. This method has been applied to polymer blend mixtures as well. Smearing of the models to account for instrumental resolution is performed prior to the fitting step. This second data analysis method is the most used. The RPA, the zero average contrast method, and the OZ models are described in detail in Chapters 31 and 32 of the SANS Toolbox respectively.1 3. Particle shape reconstruction and inverse Fourier transform methods are sophisticated approaches that perform in-depth analysis of SANS data using canned software packages. This data analysis method is rarely used owing to its high level of specificity.

The SANS signal is characterized by a constant (Q-independent) part due to incoherent scattering from (mostly) hydrogen in the sample as well as by a Q-dependent coherent scattering part which contains information about structure, morphology and phase transitions in the sample. Here Q is the scattering variable given in terms of the neutron wavelength λ and scattering angle θ as Q = (4π/λ) sin(θ/2). The coherent scattering cross section (units of cm−1) can be expressed as: d(Q) = φρ 2 VP P(Q)SI (Q). d

(1)

Here φ is the volume fraction and VP is the volume of the scattering “objects,” ρ 2 is the neutron contrast factor, P(Q) is the single object form factor, and SI (Q) is the inter-object structure factor. The scattering objects can be either polymer coils for macromolecular scattering or compact “particles” in the case of particulate scattering. The SANS Toolbox1 contains the form factors for various shape objects (Chapter 27) as well as for polymer chains with excluded volume (Chapter 28). The SANS technique is sensitive to composition fluctuations and is therefore a good probe for phase transition studies in polymer mixtures. The thermodynamics of miscibility are well described by the RPA model which predicts (for instance) the spinodal phase transition condition. Polymer mixtures (solutions or blends) either phase separate upon heating and are characterized by a lower critical solution temperature (LCST) or upon cooling and are characterized by an upper critical solution temperature (UCST). The mean field RPA approach uses the Flory-Huggins interaction parameters which can be measured by SANS.

3. SANS from Polymers A large number of SANS research from polymers has been covered at the tutorial level in the SANS Toolbox1 which contains a great deal of topics borrowed from this author’s research efforts. Chapter 37 describes the use of an empirical model to extract the correlation length (average distance between entanglements) in a semidilute polymer solution, applies the zero average contrast method to extract single-chain properties such as the radius of gyration and uses a simple extrapolation method to obtain an estimate of the spinodal (phase separation) temperature for an LCST system. Chapter 38 uses the Flory-Huggins Gibbs free energy to map out the phase diagram for a model polyolefin blend. Flory-Huggins interaction parameters were estimated and found to depend inversely on (absolute) temperature. The binodal and spinodal temperatures and the nucleation-and-growth region in-between were plotted. Chapter 39 describes SANS from copolymers. Here also, the RPA method is used to predict the order-to-disorder line for a diblock copolymer. SANS from copolymer spectra

SANS from Polymers

17

are characterized by a peak due to the correlation hole effect in the homogeneous phase or due to inter-domain spacing in the ordered phases. Ordered phases correspond to spherical, cylindrical, and lamellar morphologies (mostly). Chapter 40 summarizes SANS results from a ternary blend mixture in which two of the homopolymers are hydrogenated (i.e., non-deuterated). The Flory-Huggins interaction parameters were extracted for all three polymer pairs including the hydrogenated pair. Chapter 43 makes use of a sophisticated model to interpret SANS data from a crystalline polymer in solution. Thicknesses of the amorphous and lamellar regions as well as the number of lamellae per stack were obtained. A material balance approach was used along with nonlinear least squares fits to this model. Chapter 44 describes micelle formation in a triblock (Pluronic) copolymer solution. The unimers-to-spherical micelles conditions (temperature and concentration) were estimated. Cylindrical and lamellar micelles were also obtained upon further heating. The use of judicious sample environments has contributed greatly to the SANS from polymers research effort. Chapter 52 and 53 illustrate some of this effort using in-situ pressure and shear respectively. Use of the compressible RPA model along with an equationof-state yielded an estimation of the amount of free volume present in polymer blends. The use of the Clausius-Clapeyron equation helped predict the effect of pressure on the spinodal line. In some systems, pressure favored mixing while in other cases, it favored demixing. In-situ shear produces appealing 2D spectra with lots of spots and anisotropic features. Pluronic spherical micelles formed body-centered cubic structures that changed into facecentered cubic structures under Couette shear. Twinned structures were also observed. This review paper references some 76 papers on SANS from polymers resulting from use of the NCNR over the past three years. They are cataloged into broad categories that include polymer solutions,2–9 copolymers,10–19 polymer blends,20–23 branched or grafted polymers,24–32 polymer gels,33–40 polymer networks,41–48 polymer micelles,49–63 polymeric nanomaterials,64–70 and polymer membranes.71–77 Of these papers, about half are briefly summarized in order to represent the breadth of ongoing research. Those summarized are the ones with which this author is the most familiar with.

4. Polymer Solutions Chain conformations and demixing phase behaviors are common topics investigated using SANS from polymer solutions. Such topics include characteristic chain dimensions for various stiff or flexible polymers, polymer-solvent interactions, and phase transitions. Investigations of solution crystallization have been included in this section. Poly(cyclohexdiene) (PCHD) polymers contain six-member rings on the main chain. This characteristic gives them much desired mechanical properties and good thermal stability when compared to other vinyl polymers. For instance, PCHD polymers have the highest glass-rubber transition temperature (Tg around 231◦ C) of all hydrocarbon polymers. Solution properties of PCHD polymers in tetrahydrofuran and in chloroform solutions were investigated using conventional methods that included light scattering and SANS.2 These two techniques measured the radius of gyration (Rg ) and the second virial coefficient (A2 ) as a function of polymer concentration, temperature, and solvent quality. The Zimm plot method was used; it consists of an extrapolation to low scattering variables (Q) and low polymer fractions. A simple wormlike chain model reproduced the measured radii of gyration. It was found that the PCHD chain conformations were stiffer in chloroform than in tetrahydrofuran solutions.

18

B. Hammouda

Figure 2. SANS spectrum from 4% poly(ethylene oxide) mass fraction in d-water (left) showing the high-Q coherent signal (solvation intensity) and the constant incoherent background. The solvation intensity is plotted for 4% PEO in d-water/d-alcohol solvent mixtures (right). Non-ideal mixing is observed in all cases. The error bars correspond to one standard deviation.

The model water-soluble polymer poly(ethylene oxide) [–CH2 CH2 O–]n was used to investigate solvation properties in binary solvent mixtures consisting of water and other solvents (methanol, ethanol, ethylene glycol). The SANS technique was used to obtain a wide Q range.3 Deuterated solvents were used in order to enhance the neutron contrast. At low-Q, large length-scale features are characteristic of clustering (PEO in d-water) or crystallization (PEO in alcohols). The high-Q region probes polymer-solvent interactions (the solvation layer) as well as their mixing behavior (Fig. 2). The measurement temperature was kept above the crystal melting temperature when crystallization was present. A simple empirical model was used to fit SANS data and obtain solvation intensity, a correlation length, and a Porod exponent. The correlation length is an estimate of the average entanglement distance in the semidilute PEO solutions. Moreover, the random phase approximation model was used to back out Flory-Huggins interaction parameters for the ternary mixture PEO/d-water/d-methanol. It was found, for instance, that the solvation intensity for PEO in binary solvent mixtures was always lower than the ideal mixing prediction; non-ideal mixing seems to be the norm for PEO in mixed solvents. Mixed solvents seem to be better solvating agents for PEO than the individual solvents. The SANS technique was also used to investigate the solvation behavior of PEO in d-water in the dilute and semidilute regimes.4 The correlation length (obtained from the empirical model) was seen to decrease in dilute solutions but to increase in semidilute solutions. This behavior change yields an accurate method for measuring the overlap concentration used to delimit dilute from semilute solutions. The decrease in coil size in the dilute region is the precursor to the single-coil collapse transition that occurs in extremely dilute solutions of polymers with extremely high molecular weights. The temperature dependence of the correlation length shows that its inverse follows a linear behavior when plotted versus 1/T (where T is the absolute sample temperature). It remains to be seen whether this “universal” behavior observed for PEO would hold for other polymers in solution. Poly(ethylene oxide) assumes a coil conformation when dissolved in water, but it assumes a helical conformation when dissolved in isobutyric acid along with trace amounts

SANS from Polymers

19

Figure 3. Pair distribution function for a poly(ethylene oxide) solution in d-water (left) and in disobutyric acid (right). This distribution was obtained through the Fourier transform of the SANS data. The distribution on the right is characteristic of a mixture of helix and coil phases.

of water. The helix-to-coil transition can be reversibly effected by adding/removing trace amounts of water. SANS and polarimetry measurements were used to investigate the helical and coil structures in various solvent and temperature conditions.5 A number of data analysis methods and software packages were used to analyze the SANS data. These included standard plots (such as the Porod plot), fits to cylindrical structure models (to represent the helical structure), and inverse Fourier transform of the data to obtain a pair distribution function p(r). Porod exponents close to 1 were observed for rod-like (helical) structures at low-Q and close to 5/3 for swollen polymer coils. Note that the helical structures are characterized by high-Q Porod exponents close to 4 (smooth rod surfaces). The pair distribution functions showed pure coil phases for PEO/d-water and helical phases in PEO/d-isobutyric acid (Fig. 3). Mixtures of coil and helical phases were observed for high molecular weight PEO/d-isobutyric acid. The helical structure in solution was reminiscent of the crystalline structure of pure PEO (whereby 7 monomeric units form 2 helical turns). Similar investigations were performed on another (similar) water-soluble polymer, poly(ethylene imine), referred to as PEI [–CH2 CH2 NH–]n both in d-water and in d-isobutyric acid and similar conclusions were obtained. The helical structures were better developed with PEI than with PEO. The partitioning of PEO in water/isobutyric acid solvent mixtures was further investigated using SANS. Low molecular weight polymers (with Mw < 10 kg/mol) were seen to prefer dissolving in the isobutyric acid rich (top) phase and higher molecular weight ones end up mostly in the water (bottom) phase. Investigations of the phase boundaries for the PEO/water/ isobutyric acid ternary mixture were conducted.6 It was found that the addition of PEO tends to favor demixing in water/isobutyric acid solvent mixtures. Similar conclusions were obtained using star branched PEO instead of linear PEO. The early stages of crystallization of (low-molecular weight) polyethylene in d-xylene solutions were investigated by SANS.8 Crystallization was obtained upon cooling from the melt state (at 120◦ C) down to temperatures varying from 110◦ C to 85◦ C. Very early stages of crystal growth were investigated. The SANS technique was found to be sensitive to crystal volume fractions as small as 10−5. This sensitivity is much better than for SAXS where

20

B. Hammouda

low-Q background interferes with the useful signal. SAXS is the main diagnostic tool for investigations of polymer crystallization. Two competing modes of early-stage crystallization were presented with either spinodal decomposition or nucleation-and-growth as the driving force. The high sensitivity SANS measurements reported here favor the nucleation-andgrowth (crystal seeding-and-growth would be a better name) mechanism. Investigations of the late-stage crystal growth with well-formed lamellae were also discussed. Poly(ethylene oxide) forms a crystalline structure when dissolved in ethanol. SANS investigations were conducted on dilute and semidilute solutions of PEO/d-ethanol for temperatures above and below the crystallization temperature.9 Above the crystal melting temperature, fully-swollen polymer coils were observed while below the crystallization temperature, a sponge-like lamellar morphology was reported. DSC, WAXS, and confocal microscopy confirmed the SANS findings. The PEO/d-ethanol phase diagram showing both the crystallization and the extrapolated spinodal line has been mapped out. The extrapolated upper critical solution temperature (UCST) was found to be well below the crystallization temperature and therefore unreachable. The addition of a small amount of d-water to the PEO/d-ethanol mixture was found to destroy the crystalline morphology and yield regular polymer solution behavior. The phase diagram was seen to change to a lower critical solution temperature (LCST) when the water amount is increased.

5. Copolymers There are three categories of SANS investigations from copolymers; these used pure copolymers, copolymers in solution, and copolymers added to blends. These categories are represented here. The conformation of polymer chains in regular symmetric multiblock copolymers was investigated using SANS measurements from previously sheared samples.10 Symmetric copolymers tend to form lamellar structures which, when sheared, tend to align according to the “parallel” or A alignment (lamellae oriented in the shear/shear gradient plane) or the “perpendicular” C alignment (lamellae orientated in the shear/vorticity plane). The B alignment (lamellae oriented in the shear gradient/vorticity plane) is never observed under shear; it is observed only after shear cessation. The undecablock (containing 10 blocks) of poly(cyclohexylethylene) and poly(ethylene propylene) was used for these investigations. Mixtures of undecablocks and deuterated undecablocks (in equal fractions) were prepared in order to separate out scattering from the multiblock structure (characterized by an interblock Bragg peak) and scattering from partially deuterated polymer chain conformations (characterized by the radius of gyration for the entire copolymer). The Guinier plot at low scattering variable Q was used to measure this radius of gyration. These investigations showed that polymer chains tend to align in the B alignment plane when lamellae are aligned along the C alignment plane. Shear cessation relaxes the stretched copolymers into a 3D random walk spread out over many lamellar microdomains. The SANS technique was used to investigate vesicle formation when a poly(ethylene oxide)-poly(butylene oxide) diblock copolymer (EO6 BO11 ) is dissolved in water.12 The hydrophobic nature of the BO block drives the vesicle formation. At low diblock fraction and temperature, a wormlike micelle phase is observed. This is characterized by a 1/Q Porod behavior at low-Q (cylindrical structures). As the diblock fraction or temperature is increased, unilamellar, then multilamellar vesicles form. These are characterized by a 1/Q2 Porod behavior at low-Q (2D structures) and a Bragg peak at high-Q. At even higher diblock fraction or temperature, the vesicles form a lamellar phase. Indirect Fourier transform of the SANS data produced pair correlation functions which yielded estimates of vesicle sizes.

SANS from Polymers

21

Figure 4. Variation of the d-spacing (left) and of its ratio to the swelling asymmetry factor d/ξ for the PS-PDMS diblock copolymer in CO2 .

A morphology phase diagram (temperature versus diblock copolymer fraction in d-water) was mapped out. Pressurized CO2 plays the role of a selective solvent in a polystyrene-poly(dimethyl siloxane) diblock copolymer.16 This PS-PDMS copolymer is characterized by lamellar morphology with well-defined interdomain d-spacing. This d-spacing is inversely proportional to the scattering peak position. The SANS technique is useful for the characterization of this d-spacing as well as estimation of the various Flory-Huggins interaction parameters between each of the blocks (PS or PDMS) and the solvent (CO2 ) and between the two blocks (PS-PDMS). The swelling conditions are described by swollen block volume fractions fPS/CO2 and fPDMS/CO2 . These are determined from an equation-of-state for the swelling of the pure components (PS and PDMS) in CO2 along with the Flory-Huggins equation for polymer/solvent mixtures. Their ratio defines a swelling asymmetry factor ξ = fPS/CO2 /fPDMS/CO2 . SANS data were taken from the PS-PDMS diblock copolymer under CO2 pressure for three temperatures (40◦ C, 100◦ C, and 140◦ C). It was noted that whereas the interlamellar d-spacing (noted d) varies with the copolymer volume fraction φ diblock differently for each temperature, the ration d/ξ follows the same power law for all temperatures (Fig. 4). The microphase behavior for a series of poly(styrene sulfonate)-poly(methyl butylene) diblock copolymers was investigated using the SAXS and SANS techniques.17 High resolution of the SAXS technique allows the indexing of numerous Bragg reflections and therefore the possibility of resolving a wide range of ordered diblock copolymer microstructures. The observed copolymer morphologies include lamellae, gyroid, hexagonally perforated lamellae, and hexagonally packed cylinders. These morphologies were obtained by varying the copolymer molecular weight and sulfonation level as well as temperature. This range of morphologies was obtained with nearly symmetric diblock copolymers. TEM images confirmed some of the observed morphologies. SANS data were taken to estimate the Flory-Huggins interaction parameter which was used to predict the order-disorder transition conditions.

22

B. Hammouda

Figure 5. Phase diagram for the PEP/PBO/PEP-PBO ternary mixture for increasing copolymer fraction (left); i.e., along the isopleth line which is depicted within the triangle phase diagram (right).

The ternary phase diagram for a polymer mixture consisting of two homopolymers, poly(ethylene propylene) and poly(butylene oxide), and their diblock copolymer PEP-PBO has been investigated using a host of characterization methods that include SANS, smallangle x-ray scattering (SAXS), rheology, and optical microscopy.18 Deuterated PEP was used in order to enhance the SANS contrast. A ternary phase diagram similar to the one for a nonionic surfactant/water/oil micellar system was mapped out close to the isopleth line (line for equal homopolymer fractions but with increasing copolymer fraction). The upper critical binodal line for the PEP/PBO binary blend was also mapped out. Along the isopleth line, the phase separation boundaries between the mixed phase, the macrophase separated region, and the microphase separated region were delimited. Within the microphase separation region, the order-to-order phase transition lines for the lamellar and hexagonal microphases were obtained (note that no cubic phase was observed). SAXS was effective at differentiating the ordered phases. The macrophase separation region contains two-phase droplets (rich in PEP or PBO with PEP-PBO forming the boundary between them) and three-phase droplets (rich in PEP, in PBO, or in PEP-PBO). The SANS technique was useful for the characterization of a narrow bicontinuous microemulsion channel obtained for high copolymer fraction (80%) and high temperature (around 120◦ C). The Teubner-Strey model was used to fit the SANS data (Fig. 5). A-C diblock copolymers were mixed to weakly-segregated A/B homopolymer blends. Components A, B, and C used were polybutadiene (89% 1,2 addition), polyisobutylene and polybutadiene (63% 1,2 addition) respectively.19 The C block was characterized by attractive interactions with the B block but repulsive interactions with the A block. This SANS study showed that organized domains form with the addition of as little as 1% A-C diblock to a 50%/50% A/B blend. SANS data were taken from a series of samples for which the copolymer fraction as well as its molecular weight were varied. The random phase approximation (RPA) model was used to analyze SANS data in the homogeneous (mixed) phase region. The Teubner-Strey (TS) model and a self-consistent-field theory

SANS from Polymers

23

(SCFT) were used in the ordered microphase region. The RPA approach yielded FloryHuggins interaction parameters and mean-field phase boundary predictions. The TS model and the SCFT approach yielded predictions of characteristic microdomain d-spacings as observed by SANS. The combination of measurements and models produced reasonable agreement for most of the probed range except for the case with very small copolymer fractions where there is room for improvement.

6. Polymer Blends Polymer blends constitute another active area of SANS research. The SANS technique is sensitive to density and composition fluctuations and is therefore a good thermodynamic probe for investigation of phase transitions in blends. The same polymer blend A/B/A-C discussed before19 was used to investigate the effects of pressure on phase transitions.20 Block C is characterized by repulsive interactions with block A and attractive interactions with block B. At ambient pressure, the blend forms a lamellar microphase at low temperature, a bicontinuous microemulsion phase at intermediate temperature, and is macrophase separated at high temperature. The same blend, however, exhibits a mixed (homogeneous) phase when pressurized. This behavior was traced to intricate dependences of the Flory-Huggins interaction parameters on temperature and pressure; χ AC (positive) decreased with temperature but did not change with pressure, χ BC (negative) increased with temperature but decreased (became more negative) with pressure, and χ AB (positive) increased with temperature but decreased with pressure. The random phase approximation (RPA) and the self-consistent-field theory (SCFT) were used to analyze SANS data with in-situ pressure. The Teubner-Strey model was also used to fit data in the microemulsion phase region. A pressure-temperature phase diagram was mapped out showing boundaries between the microphase separation, the macrophase separation, and the homogeneous mixed-phase regions (Fig. 6). The observation that pressure induces the formation of a homogeneous phase (favoring mixing) in the A/B/A-C polymer blend is contrary to the observed effect of pressure in nonionic surfactant/water/oil ternary mixtures where pressure tends to favor demixing. The SANS technique can map out both the binodal line and the spinodal line. The binodal line is reached when the intercept of the Zimm plot I−1(0) becomes negative (here I(0) is the scattering intensity in the forward Q = 0 direction). The spinodal line is obtained by extrapolating the I−1(0) versus T−1 linear behavior to the limit I−1(0) = 0 (here T is the absolute sample temperature). This helps delimit the nucleation-and-growth region located between the binodal and spinodal lines. Growth kinetics were studied by SANS from a deuterated poly(methylbutylene)/poly(ethylbutylene) off-critical polyolefin blend sample following pressure jumps from the homogeneous (mixed) phase region into the nucleation-and-growth region.21 Pressure jumps are more rapid than temperature jumps and therefore more effective. Prior to jump experiments, the pressure-temperature phase diagram (showing both the binodal and spinodal lines) was mapped out for the same dPMB/PEB blend. Nucleation-and-growth kinetics measurements were performed for a number of pressure jumps and for many quench depths. Both single jumps and double jumps were performed. In the case of double jumps, the first jump was deep into the nucleation-and-growth region (to form the nucleation seeds) and the second jump was shallow (to follow the growth kinetics). Critical nucleus sizes and the times required to conclude the early stage of nucleation were measured. These were obtained from the time-dependent SANS intensity characterizing the growth kinetics.

24

B. Hammouda

Figure 6. SANS data from A/B/A-C polyolefin blend mixture of two homopolymers A/B and a diblock copolymer A-C under pressure (0.3 kbar) and for a sample temperature of 50 ◦ C (left) and pressure-temperature phase diagram (right). The Teubner-Strey model yields good model fit to the SANS data in the bicontinuous (microphase separated) phase region. The error bars correspond to one standard deviation.

SANS studies were performed to assess the possibility of swelling for high molecular weight “tracer” polymers (poly(ethylene oxide) and poly(methyl methacrylate) in low molecular weight polymer matrix.23 The same polymers (PEO and PMMA) were used for the matrix component. Three matrices were considered—pure PEO, pure PMMA, and 50%/50% (mass fractions) PEO/PMMA blend. Deuterated polymers were used for the tracer polymers and hydrogenated polymers were used for the matrix polymers. Low tracer polymer fractions were used. Measurement temperatures were chosen to be above the crystal meting temperature of PEO and above the glass-rubber transition temperature of PMMA. The SANS intensity was fit to a Debye function (form factor for unperturbed coils) in order to extract a radius of gyration in each case. The tracer polymers were found to follow unperturbed coil configurations in all cases; no chain swelling was observed.

7. Branched or Grafted Polymers Many investigations on SANS from branched or grafted polymers have been reported. These include single generation branching (stars and combs) as well as multi-generation branching (dendrimers and arborescent graft polymers). Polymer chain architecture plays a role in the mixing behavior of polymer blends. A systematic investigation has been undertaken24 using a series of branched polystyrene macromolecules (Fig. 7) with either varying number of branch points (and fixed number of chain ends) or varying number of chain ends (and fixed number of branch points). All polystyrene macromolecules were carefully synthesized and characterized. They all correspond to the same molecular weight. Blends were prepared using 50%/50% mixtures of linear deuterated polystyrene and branched (hydrogenated) polystyrenes. SANS data were taken from the two groups of blend samples at various temperatures. Random phase approximation (RPA) equations for blend mixtures of linear and specifically branched

SANS from Polymers

25

Figure 7. Polymer chain architectures with increasing number of branch points (left) or with increasing number of chain ends (right).

polymers were used to obtain an effective Flory-Huggins interaction parameter χ eff in each case. Partial form factors corresponding to the various regular branching architectures were used as inputs to the RPA fitting approach. The results showed that χ eff increases with increasing number of branch points (group I) and with increasing number of chain ends (group II). Branching seems to favor demixing in polymer blends. A Gaussian field theory was successful in predicting the overall χ eff trend for group II but not for group I. Arborescent graft polymers containing hyperbranched structures with a polystyrene comb-like backbone and poly(2-vinyl pyridine) chains grafted onto the “teeth” of the comb were investigated in either deuterated water or deuterated methanol dilute solutions.26 SANS and dynamic light scattering (DLS) were used to characterize the “fast” mode representing the local polyelectrolyte structure and the “slow” mode representing longrange clustering. SANS data showed a polyelectrolyte peak only when the pH was changed by adding hydrochloric acid. This peak is due to the so-called correlation-hole effect and is characteristic of an average distance between charged domains. The peak position scales like polymer fraction to the third power owing to the spherical (3D) symmetry of arborescent polyelectrolytes. Note that for linear polyelectrolytes, the peak position scales like the square root of the polymer fraction (2D symmetry). When enough acid is added to completely neutralize the charges on the P2VP blocks, the polyelectrolyte peak disappears again. Arborescent polymers with longer P2VP grafted blocks resulted in the formation of a gel for fractions greater than 1% mass fraction. The mean spherical approximation model for charged spheres was used to analyze the SANS data when long-range Coulomb interactions are present. Arborescent polymer sizes and degree of chain swelling were investigated using arborescent polystyrenes with two different size polystyrene side chains (with either 5K or 30K molecular weight) in dilute d-toluene solutions. Generations from G0 to G3 were measured using the SANS technique.27 A radius of gyration was obtained from the Guinier

26

B. Hammouda

Figure 8. Representation of an arborescent copolymer corresponding to two generations (plus backbone) of h-polystyrene onto which one generation of d-polystyrene was grafted. The h-polystyrene forms the “core” structure whereas the outer d-polystyrene forms the “shell” structure. When the solvent is hydrogenated, only the shell is visible (left) while when the solvent is deuterated, only the core is visible (right).

and the Kratky data analyses methods and compared to the Zimm-Stockmayer model for branched polymer systems with different architectures. This model gives reasonable agreement for the case with short (5 K) side branches but not for the case with long (30 K) side branches. The measured radius of gyration scales with molecular weight for the various generations like Rg ∼ Mw 0.26 for short side branches and Rg ∼ Mw 0.32 for long side branches. Arborescent polymers with long side branches act like compact particles with tight packing. The influence of polymer-solvent interactions on Rg was expressed in terms of the expansion factor due to excluded volume. Swelling effects were clearly observed for short side chains but not for long side chains. Inverse Fourier transforms of the SANS data yielded pair distance distribution functions p(r) which contain information about internal arborescent particle inhomogeneities within the core and within the shell regions. Furthermore, arborescent polymers were synthesized using hydrogenated polystyrene for the core and deuterated polystyrene for the side branches. This helped realize solvent contrast match conditions for either the hydrogenated core or the deuterated shell (Fig. 8) using solvents with more convenient scattering length densities (THF and cyclohexane and their deuterated versions). The higher generation arborescent polymers showed a better-defined core-shell structure than the lower generation ones. Comblike copolymers with polynorbornene (PNB) backbone and oligo ethylene glycol (OEG) side chains were measured in dilute d-water solutions.30 The diblock copolymer consists of a (hydrophobic) block with short OEG3 chains and the other (hydrophilic) block with long OEG6.6 chains. SANS measurements were performed over a temperature range between 25◦ C and 68◦ C. At 25◦ C, the copolymers were found to associate into micelles with the hydrophobic block forming an inner spherical core and the hydrophilic block forming cylindrical structures within the outer shell. As the temperature is increased, even long OE G6.6 chains become hydrophobic and phase separation occurs at the cloud point temperature of 60◦ C. Above this temperature, a transition to another phase characterized by sharp Bragg d-spacing of 349 Å is observed. A hybrid model for micelles with spherical core and cylindrical “spokes” radiating out to form the shell was used to analyze the SANS data for low temperatures. A scattering density profile for the micellar shell was obtained

SANS from Polymers

27

along with an aggregation number. The aggregation number was found to increase and the micellar size was found to decrease with increasing temperature. This is due to the increasing hydrophobicity of OEG chains upon heating. The hydrophilic block of the previously described diblock copolymer was used to investigate solvent effects in d-toluene and d-water dilute solutions.31 This corresponds to 50 monomers of norbornene backbone (NB) and oligo ethylene glycol side chains (OEG6.6 ). Four polymer fractions in the dilute solution range and four temperatures between 25◦ C and 74◦ C were measured by SANS. Chain dimensions (radius of gyration) and polymer-solvent interactions (second virial coefficient) were obtained from the familiar Zimm plot for dilute polymer solutions. Polymer chains were seen to follow random-coil conformations at low temperature and low polymer fraction when dissolved in d-toluene. Polymers were found to contract with increasing polymer fraction. Deuterated water is a selective solvent for the OEG blocks which tend to take a cylindrical shape forming the teeth of the comblike polymer. The theta temperature was estimated to be 45◦ C. At 74◦ C, hydrophobic interactions take over and most of the polymer precipitates out of solution (in d-water).

8. Polymer Gels Polymer gels form when crosslinking is introduced. The various microstructures formed are in the nanometer length scale making the SANS technique a useful characterization method. Gels were formed through (electron beam) radiation crosslinking of two diblock copolymers; poly(α-methylstyrene)-polyisoprene (PaMS-PI) and poly(vinylferrocenium triflate)-polyisoprene (PVFT-PI). Radiation crosslinks the PI blocks, induces chain scission in the PaMS blocks, and has no effect on the PVFT blocks. Swelling of the formed gels in partially deuterated solvents allowed the characterization of the crosslink density as function of the radiation dose by SANS, as well as by standard gel characterization methods.35 This allowed characterization of the copolymer microdomain morphology in the presence of crosslinks. It was found, for instance, that uncrosslinked PaMS blocks play an important role in the swelling behavior of PaMS-PI gels. Chain scission in PaMS seems to release stresses that may appear during gel formation. It was also found that chains in the PVFT blocks undergo stretching as the gel crosslinking becomes tighter. Information about solvent partitioning in the two swollen gels was obtained. Solvent-filled open channels were observed in the PaMS-PI gels whereas large length-scale clustering was observed in PVFT-PI gels. Poly(vinyl alcohol) hydrogels were formed by introducing physical crosslinks consisting of small ice crystals. These were created through freeze/thaw cycles. Samples that were stretched after the first cycle remained oriented. The PVA hydrogels are intended for potential use in biomedical applications. Their morphology and stress response were investigated using SANS and mechanical testing.36 Other characterization methods were also used (SAXS, TEM, C-13 NMR, etc.). The Debye-Bueche model was used to obtain a correlation length characterizing the average distance between crosslinks. The Teubner-Strey model was also used to obtain a correlation length along with a quasi-periodic d-pacing characterizing the hydrogel network at the local level. The dominant low-Q feature of the SANS signal was fitted to a power law behavior to represent long-range correlations. SANS results showed that the PVA hydrogel comprises a polymer-rich local phase formed

28

B. Hammouda

Figure 9. Representation of the flower-like spherical micelle (left) and lamellar micelle (right) formed of triblock copolymers where the central block is hydrophilic and the outer blocks are hydrophobic.

of nanopores 150 to 300 Å in size surrounded by a larger overall structure containing micron size morphology. This structure is held together through the small crystal crosslinks. PLA-PEO-PLA triblock copolymers are formed of hydrophyllic poly(ethylene oxide) blocks and hydrophobic poly(lactide) blocks. Since the hydrophilic block is the middle block, a flowerlike structure is obtained in water solution (Fig. 9). SANS measurements were made from PLA-PEO-PLA solutions in order to understand the morphology as function of PLA block length and stereospecifity.38 It was found that spherical micelles form when amorphous D/L-lactic acid blocks are used whereas lamellar micelles form when crystalline L-lactic acid blocks are used. Moreover, an increase in the triblock fraction (in d-water solutions) leads to gel formation. SANS data were analyzed using single micelle form factors and inter-micelle structure factors. In the case of spherical micelles, the familiar Percus-Yevick solution of the Ornsterin-Zernike equation was used while in the case of lamellar micelles, a lamellar stack model introduced to interpret data from lamellar stacks in crystalline polymers in solution was used. Micellar sizes and inter-distances were obtained. The association characteristics of the micelles were found to be controlled by the length and crystallinity of the PLA blocks. Double-network hydrogels formed of a charged crosslinked polymer network (PAMPS) and neutral linear polyacrylamide (PAA) polymers were investigated using the SANS and Ultra-SANS techniques.40 The USANS technique can probe size scales up to 20 µm. Deuterated PAA and d-water were used in order to enhance the neutron contrast. Investigation of PAMPS solutions in d-water and of PAMPS/d-PAA solutions in mixtures of d-water and h-water were conducted in order to measure the various Flory-Huggins interaction parameters for the various components. It was found that χ PAMPS/PAA χ PAMS/water < χ PAA/water . Measurements from PAMPS/d-PAA/d-water were also taken from the fully formed double-network hydrogels. A random phase approximation model that incorporates charge interactions (through a Debye-Huckel factor) and a crosslinked network (though a characteristic mesh size) was used to fit the SANS data. This model reproduces the high-Q SANS data well and yields some understanding of double-network hydrogel structures at the molecular level.

SANS from Polymers

29

Figure 10. Inhomogeneous structure of PAA (left) and PAMPS (right) polymerized in water as inferred from SANS data.

9. Polymer Networks The boundary between gels and networks is fuzzy. These appellations could be used interchangeably. Gels usually form beyond a sol-gel transition which is reversible while networks are usually not reversible; they involve covalent crosslinking. The PAMPS/PAA double-network hydrogel described earlier40 was used to understand the molecular origin of some unusual mechanical properties (Fig. 10). In-situ SANS investigations were undertaken using samples deformed in a compression device.41 Molecular conformations of the deuterated PAA chains were monitored. Possible molecular origin of the correlation between enhancement in solution viscosity and fracture toughness of crosslinked gels was discussed. It was argued that molecular association between the PAMPS and PAA components of the double-network could be at the origin of the unusual mechanical performance. The attractive interactions between these two polymers could explain the increased toughness of these polymeric materials. Polymer blends consisting of stiff liquid crystalline polyurethane and flexible polystyrene-poly(vinyl phenol) copolymers form hydrogen bonded networks characterized by a wide miscibility window.45 The polystyrene blocks were deuterated to enhance the neutron contrast. SANS measurements from the pure copolymer and with increasing amount of polyurethane allowed the monitoring of crystalline polyurethane chain conformations within an amorphous flexible polymer matrix. FTIR studies showed clear evidence of hydrogen-bonding between the two network components. Increasing the polyurethane fraction succeeded in breaking down the network of hydrogen bonds and led to the formation of new large-scale structures. The semiflexible polyurethane chains assume anisotropic conformations as observed by the SANS high-Q data and fits to the Kratky-Porod wormlike chain model. Interpenetrating polymer networks (IPNs) are formed through the in-situ synthesis of a network within the matrix of another. The state of miscibility of the mixed components during synthesis dictates the resulting network morphology. SANS and dynamic mechanical thermal analysis (DMTA) were performed on a series of methacrylate/epoxy interpenetrating polymer networks in order to assess the extent of molecular miscibility.46 SANS is sensitive to composition fluctuations and DMTA can measure the glass-rubber transition temperatures (Tg s) for polymeric materials precisely. A single Tg is a signature of a homogeneously mixed phase while two Tg s point to a demixed two-phase system. The

30

B. Hammouda

Debye-Bueche model was used to interpret SANS data from the interpenetrating polymer networks and yielded a phase separation size scale of order 180 Å. DMTA of the rubbery region of some of the IPNs revealed bicontinuous structures and the extent of phase separation. A number of amphiphilic co-networks of methacrylic acid (MAA) and 2-butyl-1-octyl methacrylate (BOMA) were synthesized and characterized using a host of methods.48 Some were model co-networks containing A-B-A copolymer chains between cross-links of precise length and composition. Here the A block contains multiple MAA monomers and the B block contains multiple BOMA monomers. The linear co-network precursors used in the group transfer polymerization were characterized by GPC and NMR for their molecular weight and composition. The degree of swelling of these amphiphilic polymer co-networks was investigated in water and in THF over a broad ionization range of the MAA monomers. It was found that the degree of swelling in water increased with the degree of ionization and the size of the MAA blocks. The degree of swelling in THF also increased with the length of the copolymers between cross-link points. The SANS technique and atomic force microscopy were used to characterize the co-network morphology.

10. Polymer Micelles Water-soluble polymers can form nonionic micelles owing to their hydrophobic/ hydrophilic characteristics. Here also, the SANS technique has been an effective tool for the characterization of structure and miscibility. Wormlike micelles are formed using trimethylammonium cations and 4-vinylbenzoate counterions in aqueous solution. The resulting polymer-surfactant aggregates were polymerized to obtain rodlike ionic micelles.49 The micelle radius was controlled by varying the hydrocarbon length on the trimethylammonium and its length was controlled by varying the initiator decomposition half-life. This was done by varying temperature or using different initiators. The micelle radius was varied between 17 Å and 24 Å and its length was varied between 800 Å and 5000 Å as characterized by SANS measurements. This approach yielded stable polymerized rodlike micelles with controllable sizes that are independent of surfactant concentration (provided that it is higher than the critical micelle concentration). Long polymerized micelles were obtained using low initiator content and low temperature. Free radical polymerization of the mixture of cetyltrimethylammonium (CTVB) and sodium 4-styrenesulfonate (NaSS) in aqueous solution produced stable rodlike particles (Fig. 11) with controlled surface charge density.51 Rodlike particle dimensions were

Figure 11. Chemical representation of CTVB and NaSS (left) and schematic representation of the polymerized rodlike micelle (right).

SANS from Polymers

31

characterized by SANS; the diameter was found to be constant (equal to 40 Å) and the length varied between 240 Å and 850 Å depending on NaSS concentration. When the NaSS concentration was increased, the rodlike particle length increased then decreased. The longest particles were obtained at the charge neutralization condition. Zeta potential measurements were also performed to characterize NaSS charge effects. Polymer micelles were formed using an A-C diblock copolymer (acting as surfactant) mixed with A and B homopolymers. The C block was characterized by repulsive interactions with the A block but attractive interactions with the B block.56 This project was also included in earlier sections (SANS from copolymers and blends). The SANS technique was effective at mapping out a microphase separation region, a macrophase separation (twophase) region, and a homogeneously mixed (one-phase) region in-between. The random phase approximation (RPA), the self-consistent-field theory (SCFT) and Teubner-Strey (TS) model were used to analyze the SANS data. Reasonable agreements were found. Small molecule surfactants cause the melting of polymeric micelles formed of amphiphilic diblock copolymers. The poly(butylacrylate)-poly(acrylic acid) (PBA-PAA) diblock forms micelles in aqueous solution. Neutral or ionic surfactants (such as C12E6 for example) break the polymer micelles as documented by light scattering (SLS and DLS), SAXS, SANS, cryo-TEM, and capillary electrophoresis;58 they reduce the interfacial tension and gradually produce two populations—one rich in large polymer micelles and one rich in small surfactant micelles. Before adding small surfactants, fits of a core-shell model to SANS data yielded an estimate of the core radius for the polymeric micelles around 80 Å (polydispersity of 0.2). After adding 1.5% mass fraction of C12E6 small molecule surfactant, the mean micelle radius decreased to 34 Å (polydispersity of 0.14). This size (34 Å) is very close to the size of pure-surfactant micelles. Simple interfacial tension arguments with and without surfactant permits the interpretation of the observed trends. A model hydrophobically modified polymer was used to crosslink wormlike micelles59 in water. Water-soluble poly(ethylene oxide) containing hydrocarbon tails (C14 to C22 ) at their ends was used to form bridges across wormlike micelles (1% mass fraction CTAT in water). SANS and rheology were used to characterize crosslinked network. The hydrophobic end groups stick to the wormlike micelles while PEO chains remain dissolved. Three types of PEO chains were used— 1. PEO chains with a sticker at one end, 2. PEO chains with a sticker at each end, and 3. 3-arm PEO stars with a sticker at each end. The wormlike micelles are likely breaking locally and reforming to shield the hydrophobic stickers from contact with water. The PEO chains form bridges between the wormlike micelles (Fig. 12). Pluronic triblock copolymers (PEO-PPO-PEO) form micellar structures at ambient temperatures. Spherical micelles formed when Pluronic F127 solutions in d-water (20% mass fraction) were sheared in a Couette shear cell. In-situ SANS investigations were performed using the radial and tangential neutron beam configurations.60 The spherical micelles were seen to form layered macrolattice structures characterized by single-crystal type scattering (with bright diffraction spots). Transition from face-centered cubic structure at low shear rates to random layer stacking at high shear rates was observed. This was the cause of the observed shear thinning behavior. A model comprising intra-layer sphere arrangement as well as inter-layer structure was used to interpret the SANS data under shear. This model incorporated oriented stacks of micellar layers and allowed for disorder

32

B. Hammouda

Figure 12. Two possible clustering configurations. There is evidence that the configuration on the left is more likely than the one on the right.

effects; it reproduced the observed cubic structures and a transition to random layer stacking leading to a hexagonal phase. The SANS technique was used to investigate the structure of polystyrene-poly(acrylic acid, sodium salt) copolymers in water solution under Couette and plate-plate shear.63 This investigation reported a wide range of copolymer fractions in water solution using PSPAA of Mn = 4,700–11,300 g/mol. Since the polystyrene block is hydrophobic, wormlike micelles form. These were found to form hexagonally close-packed cylinders under shear with the cylinder axis parallel to the shear gradient direction; i.e., perpendicular to the shear flow direction. SAXS and electron microcopy confirmed the oriented cylinders structure. Evidence for bridging between cylinders was reported. This may be due to the long PAA block length and some clustering driving force which depends on the cylindrical micelle fraction.

11. Polymeric Nanomaterials This is a category gathering SANS investigations from various materials including those described here. SANS from mixtures of deuterated and non-deuterated isotopic blends is a good monitor of chain orientation in (for instance) injection-molded samples. A series of isotopic polystyrene blends has been injection molded, while varying a number of experimental conditions, including injection molding speed, mold thickness, and mold temperature.64 Elliptical averaging of the anisotropic 2D data yielded an eccentricity factor which is a measure of the degree of chain orientation. This eccentricity factor was found to decrease with injection speed, mold thickness, and injection molding temperature. It was also found to decrease with the length scale probed showing that nano-stresses acting upon polymer chains relax at the local chain segment level. SANS and x-ray reflectometry have been used to characterize the porosity of a poly(phenylene) low-k dielectric thin film material.65 This material contains porogen which degrades upon baking at elevated temperatures thereby forming the pores. Three baking temperatures were used (150◦ C, 400◦ C, and 430◦ C). Since the thickness of each film was around 1 micron, many wafers were stacked in order to enhance the SANS signal. The

SANS from Polymers

33

Figure 13. Porosity representation of the low-k dielectric material containing porogen after baking at 400◦ C without (left) and with solvent contrast match (right). The partially deuterated solvent vapor matches the scattering length density of the matrix and fills the pores making them invisible.

samples were placed in a custom-built flow-through cell in order to control the vapor pressure. Mixtures of deuterated and non-deuterated methanol or toluene solvent vapors were used to adjust the neutron contrast by filling the pores (Fig. 13). The pore size distribution and pore fraction were determined for each baking strategy. The fraction of porogen that remains in the low-k films was estimated and found to decrease with baking temperature. Body armors are made out of high strength material containing poly(p-phenylene-2,6benzobisoxazole) fibers. An extensive study of such material aged at elevated temperatures for extended periods of time (around half a year) was conducted using a battery of characterization methods including mechanical testing (tensile strength measurements), SANS, FTIR, atomic force microscopy, and confocal microscopy.66 A 30% decrease in yarn tensile strength upon aging in humid environment was correlated with the hydrolysis of specific chemical groups as observed by FTIR. When aging was performed in an inert (argon) environment, this decrease went down to 4%. This demonstrates that moisture is a key factor in fiber degradation. SANS studies were conducted on composites formed of isotopic polystyrene blends mixed with silica nanoparticles under contrast match condition.69 It was found that polymer chain conformations follow unperturbed Gaussian chain statistics regardless of polymer molecular weight and nanoparticle loading. The polymer reference interaction site model (RISM) was used to model polymer/nanoparticle interactions.

12. Polymer Membranes The hydrogen fuel cell technology has become an active SANS area of research. Fuel cells use hydrogen to produce electricity. Hydrogen fuel is channeled to the anode on one side while oxygen is channeled to the cathode on the other side. At the anode, a catalyst causes hydrogen to ionize into protons and electrons. The polymer membrane allows only the protons to pass through to the cathode. The electrons, on the other hand, travel along an external circuit thereby generating an electrical current (Fig. 14). SANS has been used to characterize the structure of some polymer membranes.

34

B. Hammouda

Figure 14. Schematics of a hydrogen fuel cell (left) and bicontinuous morphology of a polymer membrane (right).

Poly(perfluorosulfonic acid) also referred to as Nafion is a fuel cell membrane material. SANS investigations have been conducted on Nafion membranes in an in-situ vapor sorption cell used to control the relative humidity.71 Different membrane processing conditions (melt extrusion or solvent casting), thermal pretreatment histories, and submicron membrane thicknesses were studied. The SANS data showed features characterizing semicrystalline copolymers and porous media with ion channels in the nanometer size range. A strong correlation was found between the interionic domain distance and the relative humidity. Diffusion coefficients of water vapor were estimated based on the observed structural evolution. Diblock copolymer films composed of a fluorocarbon block and a sulfonated polystyrene block were investigated by SANS and TEM.73 These are the potential proton exchange membranes for low-temperature fuel cells. Two hierarchical structure levels were observed—one due to the block copolymer microstructure and one due to the charged domains structure. The copolymer microstructure shows clearly fluorous domains and sulfonated polystyrene (darker) domains by TEM. Longer and partially sulfonated polystyrene blocks yield well-ordered microdomains, whereas shorter and fully sulfonated polystyrene blocks yield more disordered structures. Polystyrene sulfonate-b-poly(methyl butytlene) block copolymers have also been investigated as potential membrane material for hydrogen fuel cells.75 The microdomain morphology was controlled by varying the molecular weight of the polystyrene sulfonate (PSS) block and therefore the size of the hydrophilic proton channels. A drastic trend reversal was observed for channel sizes around 50 Å. The proton conductivity was found to decrease with increasing temperature for channel sizes higher than 50 Å (i.e., when high molecular weight PSS blocks are used) and to increase with increasing temperature for channel sizes smaller than 50 Å (i.e., when low molecular weight PSS blocks are used).

SANS from Polymers

35

These findings were supported from evidence based on TEM and SANS studies. Some of these studies were conducted under systematic moisture control. The correlation between the moisture content and proton conductivity in these membranes was clearly demonstrated. The phase behavior of one of the above-described block copolymer PSS-PMB used as ion-containing membrane has been investigated as function of relative humidity (RH). The copolymer ion content (sulfonation level) represented by the ion-exchange capacity (IEC) parameter was varied as well as the membrane temperature.76 SAXS, TEM, SANS, DSC and water content measurements were used to probe the rich phase behavior. Indexing of the various SAXS Bragg peaks helps in structure determination. The disordered phase was observed for low IEC and low RH values, the gyroid phase was observed for high IEC and low RH values, whereas the lamellar phase was observed for high RH values.

13. Summary and Future Prospect Use of the SANS technique has been ever-growing in the area of polymer research. Improvements of SANS instrument capabilities, sophistication of data analysis methods, and the advent of judicious sample environments have brought about renewed interest over the past thirty years. One of the leading SANS research centers, the NCNR, operates two 30 m SANS instruments in the user mode. This attracts over 200 SANS users per year resulting in over 80 publications in refereed journals. Polymer research constitute about one-third of this effort. Increased demand for additional SANS beamtime is the driving force behind a major upgrade of the NCNR facility. This upgrade includes the construction of a higher resolution SANS instrument. The new 40 m VSANS instrument (V is for “very”) will lower the measurement range (Qmin ) by an order of magnitude without too much loss in neutron flux on sample. The new (lower) Qmin will be around 0.0002 Å−1 which will be achieved by using multiple-hole converging collimation. The measurement window of the VSANS instrument will overlap nicely with the Bonse-Hart USANS instrument. The new capability VSANS/USANS combination will cover 5 orders of magnitude in scattering variable. This will allow the probing of polymer structures from the near atomic (nanometer) scale to well into the optical (20 micrometers) size scale and will open up exciting new prospects for polymer research. Other neutron scattering facilities in the US, in Europe, as well as elsewhere are also thriving. The SANS technique is the major driving force fueling the success of these facilities.

Disclaimer/Acknowledegments Certain commercial equipment, instruments, or materials are identified in this paper to foster understanding. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipments identified are necessarily the best available for the purpose. This work is based upon activities supported in part by the National Science Foundation under Agreement No. DMR-0454672.

References 1. Hammouda, B. Probing nanoscale structures—The SANS toolbox, 2009, available online at http:www.ncnr.nist.gov/staff/hammouda/the sans toolbox.pdf.

36

B. Hammouda

2. Yun, S. I.; Terao, K.; Hong, K.; Melnichenko, Y. B.; Wignall, G. D.; Britt, P. F.; Mays, J. W. “Solution properties of 1,3-cyclohexadiene polymers by laser light scattering and small-angle neutron scattering,” Macromolecules, 2006, 39, 897–899. 3. Hammouda, B. “Solvation characteristics of a model water-soluble polymer,” Journal of Polymer Science: Part B: Polymer Physics, 2006, 44, 3195–3199. 4. Hammouda, B., Ho, D. L. “Insight into chain dimensions in PEO/water solutions,” Journal of Polymer Science: Part B: Polymer Physics, 2007, 45, 2196–2200. 5. Norman, A. I.; Fei, Y.; Ho, D. L.; Greer, S. C. “Folding and unfolding of polymer helices in solution,” Macromolecules, 2007, 40, 2559–2567. 6. Norman, A. I.; Ho, D. L.; Greer, S. C. “Partitioning, fractionation, and conformations of star poly(ethylene glycol) in isobutyric acid and water,” Macromolecules, 2007, 40, 9628–9639. 7. Rahman, M. H.; Chen, C. Y.; Liao, S. C.; Chen, H. L.; Tsao, C. S.; Chen, J. H.; Liao, J. L.; Ivanov, V. A.; Chen, S. A. “Segmental alignment in the aggregate domains of poly(9,9-dioctylfluorene) in semidilute solution,” Macromolecules, 2007, 40, 6572–6578. 8. Wang, H. “SANS study of the early stages of crystallization in polyethylene solutions,” Polymer, 2006, 47, 4897–4900. 9. Ho, D. L.; Hammouda, B.; Kline, S. R.; Chen, W. R. “Unusual phase behavior in mixtures of poly(ethylene oxide) and ethyl Alcohol,” Journal of Polymer Science: Part B: Polymer Physics, 2006, 44, 557–564. 10. Wu, L.; Lodge, T. P.; Bates, F. S. “SANS determination of chain conformation in perpendicularaligned undecablock copolymer lamellae,” Macromolecules, 2006, 39, 294–299. 11. Foster, L. J. R.; Knott, R.; Sanguanchaipaiwong, V.; Holden, P. J. “Polyhydroxyalkanoate-based natural–synthetic hybrid copolymer films: A small-angle neutron scattering study,” Physica B, 2006, 385–386, 770–772. 12. Norman, A. I.; Ho, D. L.; Lee, J. H.; Karim, A. “Spontaneous formation of vesicles of diblock copolymer EO6BO11 in water: A SANS study,” J. Phys. Chem. B, 2006, 110, 62–67. 13. Triftaridou, A. I.; Vamvakaki, M.; Patrickios, C. S. “Cationic amphiphilic model networks based on symmetrical ABCBA pentablock terpolymers: Synthesis, characterization, and modeling,” Biomacromolecules, 2007, 8, 1615–1623. 14. Jacquin, M.; Muller, P.; Talingting-Pabalan, R.; Cottet, H.; Berret, J. F.; Futterer, T.; Th´eodoly, O. “Chemical analysis and aqueous solution properties of charged amphiphilic block copolymers PBA-b-PAA synthesized by MADIX,” Journal of Colloid and Interface Science, 2007, 316, 897–911. 15. Agrawal, S. K.; Sanabria-DeLong, N.; Jemian, P. R.; Tew, G. N.; Bhatia, S. R. “Microto nanoscale structure of biocompatible PLA-PEO-PLA hydrogels,” Langmuir, 2007, 23, 5039–5044. 16. Francis, T. J.; Vogt, B. D.; Wang, M. X.; Watkins, J. J. “Scaling of interdomain spacing of diblock copolymers in a selective diluent,” Macromolecules, 2007, 40, 2515–2519. 17. Park, M. J.; Balsara, N. P. “Phase behavior of symmetric sulfonated block copolymers,” Macromolecules, 2008, 41, 3678–3687. 18. Zhou, N.; Lodge, T. P.; Bates, F. S. “Influence of conformational asymmetry on the phase behavior of ternary homopolymer/block copolymer blends around the bicontinuous microemulsion channel,” J. Phys. Chem. B, 2006, 110, 3979–3989. 19. Ruegg, M. L.; Reynolds, B. J.; Lin, M. Y.; Lohse, D. J.; Balsara, N. P. “Minimizing the concentration of diblock copolymer needed to organize blends of weakly segregated polymers by tuning attractive and repulsive interactions,” Macromolecules, 2007, 40, 1207–1217. 20. Ruegg, M. L.; Reynolds, B. J.; Lin, M. Y.; Lohse, D. J.; Krishnamoorti, R., Balsara, N. P. “Effect of pressure on a multicomponent A/B/A-C polymer blend with attractive and repulsive interactions,” Macromolecules, 2007, 40, 355–365. 21. Patel, A. J.; Balsara, N. P. “Observing nucleation close to the binodal by perturbing metastable polymer blends,” Macromolecules, 2007, 40, 1675–1683. 22. Wanakule, N. S.; Nedoma, A. J.; Robertson, M. L.; Fang, X.; Jackson, A.; Garetz, B. A.; Balsara, N. P. “Characterization of micron-sized periodic structures in multicomponent polymer blends

SANS from Polymers

23. 24.

25.

26. 27.

28.

29.

30.

31.

32.

33. 34. 35.

36. 37.

38.

39. 40. 41.

37

by ultra-small-angle neutron scattering and optical microscopy,” Macromolecules, 2008, 41, 471–477. Zeroni, I.; Lodge, T. P. “Chain dimensions in poly(ethylene oxide)/poly(methyl methacrylate) blends,” Macromolecules, 2008, 41, 1050–1052. Lee, J. S.; Foster, M. D.; Wu, D. T. “Effects of branch points and chain ends on the thermodynamic interaction parameter in binary blends of regularly branched and linear polymers,” Macromolecules, 2006, 39, 5113–5121. Rathgeber, S.; Monkenbusch, M.; Hedrick, J. L.; Trollsas, M.; Gast, A. P. “Starlike dendrimers in solutions: Structural properties and internal dynamics,” The Journal of Chemical Physics, 2006, 125, 204908-1 to 204908-11. Yun, S. I.; Briber, R. M.; Kee, R. A.; Gauthier, M. “Dilute-solution structure of charged arborescent graft polymer,” Polymer, 2006, 47, 2750–2759. Yun, S. I.; Lai, K. C.; Briber, R. M.; Teertstra, S. J.; Gauthier, M.; Bauer, B. J. “Conformation of arborescent polymers in solution by small-angle neutron scattering: Segment density and core-shell morphology,” Macromolecules, 2008, 41, 175–183. Chen, W. R.; Porcar, L.; Liu, Y.; Butler, P. D.; Magid, L. J. “Small angle neutron scattering studies of the counterion effects on the molecular conformation and structure of charged G4 PAMAM dendrimers in aqueous solutions,” Macromolecules, 2007, 40, 5887–5898. Cheng, G.; Melnichenko, Y. B.; Wignall, G. D.; Hua, F.; Hong, K.; Mays, J. W. “Conformation of oligo(ethylene glycol) grafted polystyrene in dilute aqueous solutions,” Polymer, 2007, 48, 4108–4113. Cheng, G.; Hua, F.; Melnichenko, Y. B.; Hong, K.; Mays, J. W.; Hammouda, B.; Wignall, G. D. “Association and structure of thermosensitive comblike block copolymers in aqueous solutions,” Macromolecules, 2008, 41, 4824–4827. Cheng, G.; Hua, F.; Melnichenko, Y. B.; Hong, K.; Mays, J. W.; Hammouda, B.; Wignall, G. D. “Conformation of oligo(ethylene glycol) grafted poly(norbornene) in solutions: A small angle neutron scattering study” European Polymer Journal, 2008, 44, 2859–2864. Mortensen, K.; Gasser, U.; Gursel, S. A.; Scherer, G. G. “Structural characterization of radiationgrafted block copolymer films using SANS technique,” Journal of Polymer Science: Part B: Polymer Physics, 2008, 46, 1660–1668. Loizou, E.; Butler, P.; Porcar, L.; Schmidt, G. “Dynamic responses in nanocomposite hydrogels,” Macromolecules, 2006, 39, 1614–1619. Matos, M. A.; White, L. R.; Tilton, R. D. “Electroosmotically enhanced mass transfer through polyacrylamide gels,” Journal of Colloid and Interface Science, 2006, 300, 429–436. Durkee, D. A.; Gomez, E. D.; Ellsworth, M. W.; Bell, A. T.; Balsara, N. P. “Microstructure and solvent distribution in cross-linked diblock copolymer gels,” Macromolecules, 2007, 40, 5103–5110. Millon, L. E.; Nieh, M. P.; Hutter, J. L.; Wan, W. “SANS characterization of an anisotropic poly(vinyl alcohol) hydrogel with vascular applications,” Macromolecules, 2007, 40, 3655–3662. Jiang, J.; Malal, R.; Li, C.; Lin, M. Y.; Colby, R. H.; Gersappe, D.; Rafailovich, M. H.; Sokolov, J. C.; Cohn, D. “Rheology of thermoreversible hydrogels from multiblock associating copolymers,” Macromolecules, 2008, 41, 3646–3652. Agrawal, S. K.; Sanabria-DeLong, N.; Tew, G. N.; Bhatia, S. R. “Structural characterization of PLA-PEO-PLA solutions and hydrogels: Crystalline vs amorphous PLA domains,” Macromolecules, 2008, 41, 1774–1784. Tirumala, V. R.; Tominaga, T.; Lee, S.; Butler, P. D.; Lin, E. K.; Gong, J. P.; Wu, W. L. “Molecular model for toughening in double-network hydrogels,” J. Phys. Chem. B, 2008, 112, 8024–8031. Tominaga, T.; Tirumala, V. R.; Lee, S.; Lin, E. K.; Gong, J. P.; Wu, W. L. “Thermodynamic interactions in double-network hydrogels,” J. Phys. Chem. B, 2008, 112, 3903–3909. Tominaga, T.; Tirumala, V. R.; Lin, E. K.; Gong, J. P.; Furukawa, H.; Osada, Y.; Wu, W. L. “The molecular origin of enhanced toughness in doublenetwork hydrogels: A neutron scattering study,” Polymer, 2007, 48, 7449–7454.

38

B. Hammouda

42. Sun, Y. S.; Jeng, U. S.; Huang, Y. S.; Liang, K. S.; Lin, T. L.; Tsao, C. S. “Complementary SAXS and SANS for structural characteristics of a polyurethethane elastomer of low hard-segment content,” Physica B, 2006, 385–386, 650–652. 43. Li, Y. C.; Chen, K. B.; Chen, H. L.; Hsu, C. S.; Tsao, C. S.; Chen, J. H.; Chen, S. A. “Fractal aggregates of conjugated polymer in solution state,” Langmuir, 2006, 22, 11009–11015. 44. Burford, R. P.; Markotsis, M. G.; Knott, R. B. “Real-time SANS study of interpenetrating polymer network (IPN) formation,” Physica B, 2006, 385–386, 766–769. 45. Mehta, R.; Dadmun, M. D. “Small angle neutron scattering studies on miscible blends of poly(styrene-ran-vinyl phenol) with liquid crystalline polyurethane,” Macromolecules, 2006, 39, 8799–8807. 46. Dean, K. M.; Cook, W. D.; Lin, M. Y. “Small angle neutron scattering and dynamic mechanical thermal analysis of dimethacrylate/epoxy IPNs,” European Polymer Journal, 2006, 42, 2872–2887. 47. Kali, G.; Georgiou, T. K.; Ivan, B.; Patrickios, C. S.; Loizou, E.; Thomann, Y.; Tiller, J. C. “Synthesis and characterization of anionic amphiphilic model co-networks based on methacrylic acid and methyl methacrylate: Effects of composition and architecture,” Macromolecules, 2007, 40, 2192–2200. 48. Kali, G.; Georgiou, T. K.; Iv´an, B.; Patrickios, C. S.; Loizou, E.; Thomann, Y.; Tiller, J. C. “Synthesis and characterization of anionic amphiphilic model co-networks of 2-Butyl-1-octylmethacrylate and methacrylic acid: Effects of polymer composition and architecture,” Langmuir, 2007, 23, 10746–10755. 49. Gerber, M. J.; Walker, L. M. “Controlling dimensions of polymerized micelles: Micelle template versus reaction conditions,” Langmuir, 2006, 22, 941–948. 50. Kuntz, D. M.; Walker, L. M. “Solution behavior of rod-like polyelectrolyte-surfactant aggregates polymerized from wormlike micelles,” J. Phys. Chem. B, 2007, 111, 6417–6424. 51. Kim, T. H.; Choi, S. M.; Kline, S. R. “Polymerized rodlike nanoparticles with controlled surface charge density,” Langmuir, 2006, 22, 2844–2850. 52. Kim, T. H.; Doe, C.; Kline, S. R.; Choi, S. M. “Water redispersible isolated single-walled carbon nanotubes fabricated by in-situ polymerization of micelles,” Adv. Mater. 2007, 19, 929– 933. 53. Pozzo, D. C.; Walker, L. M. “Shear orientation of nanoparticle arrays templated in a thermoreversible block copolymer micellar crystal,” Macromolecules, 2007, 40, 5801–5811. 54. Pozzo, D. C.; Walker, L. M. “Small-angle neutron scattering of silica nanoparticles templated in PEO–PPO–PEO cubic crystals” Colloids and Surfaces A: Physicochem. Eng. Aspects, 2007, 294, 117–129. 55. Pozzo, D. C.; Walker, L. M. “Macroscopic alignment of nanoparticle arrays in soft crystals of cubic and cylindrical polymer micelles,” Eur. Phys. J. E, 2008, 26, 183–189. 56. Ruegg, M .L.; Reynolds, B. J.; Lin, M. Y.; Lohse, D. J.; Balsara, N. P. “Microphase and macrophase separation in multicomponent A/B/A-C polymer blends with attractive and repulsive interactions,” Macromolecules, 2006, 39, 1125–1134. 57. Gomez, E. D.; Ruegg, M. L.; Minor, A. M.; Kisielowski, C.; Downing, K. H.; Glaeser, R. M.; Balsara, N. P. “Interfacial concentration profiles of rubbery polyolefin lamellae determined by quantitative electron microscopy,” Macromolecules, 2008, 41, 156–162. 58. Jacquin, M.; Muller, P.; Cottet, H.; Crooks, R.; Th´e odoly, O. “Controlling the melting of kinetically frozen poly(butylacrylate-b-acrylic acid) micelles via addition of surfactant,” Langmuir, 2007, 23, 9939–9948. 59. Lodge, T. P.; Taribagil, R.; Yoshida, T.; Hillmyer, M. A. “SANS evidence for the cross-linking of wormlike micelles by a model hydrophobically modified polymer,” Macromolecules, 2007, 40, 4728–4731. 60. Jiang, J.; Burger, C.; Li, C.; Li, J.; Lin, M. Y.; Colby, R. H.; Rafailovich, M. H.; Sokolov, J. C. “Shear-induced layered structure of polymeric micelles by SANS,” Macromolecules, 2007, 40, 4016–4022.

SANS from Polymers

39

61. Jiang, J.; Li, C.; Lombardi, J.; Colby, R. H.; Rigas, B.; Rafailovich, M. H.; Sokolov, J. C. “The effect of physiologically relevant additives on the rheological properties of concentrated Pluronic copolymer gels,” Polymer, 2008, 49, 3561–3567. 62. O’Driscoll, B. M. D.; Hawley, A. M.; Edler, K. J. “Incorporation of sparingly soluble species in mesostructured surfactant–polymer films,” Journal of Colloid and Interface Science, 2008, 317, 585–592. 63. Grandjean, J.; Mourchid, A. “Restricted swelling and its orientation effect on copolymer micellar solutions of hexagonal-packed cylinders under steady shear flow,” Langmuir, 2008, 24, 2318–2325. 64. Healy, J.; Edward, G. H.; Knott, R. B. “Residual orientation in injection micro-molded samples,” Physica B, 2006, 385–386, 620–622. 65. Silverstein, M. S.; Bauer, B. J.; Hedden, R. C.; Lee, H. J.; Landes, B. G. “SANS and XRR porosimetry of a polyphenylene low-k dielectric,” Macromolecules, 2006, 39, 2998–3006. 66. Chin, J.; Forster, A.; Clerici, C.; Sung, L.; Oudina, M.; Rice, K. “Temperature and humidity aging of poly(p-phenylene-2,6- benzobisoxazole) fibers: Chemical and physical characterization,” Polymer Degradation and Stability, 2007, 92, 1234–1246. 67. Stefanescu, E. A.; Dundigalla, A.; Ferreiro, V.; Loizou, E.; Porcar, L.; Negulescu, I.; Garno, J.; Schmidt, G. “Supramolecular structures in nanocomposite multilayered films,” Phys. Chem. Chem. Phys., 2006, 8, 1739–1746. 68. Li, J.; Jiang, J.; Li, C.; Lin, M. Y.; Schwarz, S. A.; Rafailovich, M. H.; Sokolov, J. “Effect of temperature on shear-induced anisotropic structure in polymer clay hydrogels,” Macromol. Rapid Commun. 2006, 27, 1787–1791. 69. Sen, S.; Xie, Y.; Kumar, S. K.; Yang, H.; Bansal, A.; Ho, D. L.; Hall, L.; Hooper, J. B.; Schweizer, K. S. “Chain conformations and bound-layer correlations in polymer nanocomposites,” Phys. Rev. Lett., 2007, 98, 128302-1 to 128302-4. 70. Chatterjee, T.; Krishnamoorti, R. “Dynamic consequences of the fractal network of nanotubepoly(ethylene oxide) nanocomposites,” Physical Review E, 2007, 75, 050403-1 to 050403-4. 71. Kim, M. H.; Glinka, C. J.; Grot, S. A.; Grot, W. G. “SANS study of the effects of water vapor sorption on the nanoscale structure of perfluorinated sulfonic acid (Nafion) membranes,” Macromolecules, 2006, 39, 4775–4787. 72. Gao, J.; Yang, Y.; Lee, D.; Holdcroft, S.; Frisken, B. J. “Self assembly of latex particles into proton-conductive membranes,” Macromolecules, 2006, 39, 8060–8066. 73. Rubatat, L.; Shi, Z.; Diat, O.; Holdcroft, S.; Frisken, B. J. “Structural study of proton-conducting fluorous block copolymer membranes,” Macromolecules, 2006, 39, 720–730. 74. Shin, K.; Obukhov, S.; Chen, J. T.; Huh, J.; Hwang, Y.; Mok, S.; Dobriyal, P.; Thiyagarajan, P.; Russell, T. P. “Enhanced mobility of confined polymers,” Nature Materials, 2007, 6, 961–965. 75. Park, M. J.; Downing, K. H.; Jackson, A.; Gomez, E. D.; Minor, A. M.; Cookson, D.; Weber, A. Z.; Balsara, N. P. “Increased water retention in polymer electrolyte membranes at elevated temperatures assisted by capillary condensation,” Nano Letters, 2007, 7, 3547–3552. 76. Park, M. J.; Nedoma, A. J.; Geissler, P. L.; Balsara, N. P.; Jackson, A.; Cookson, D. “Humidityinduced phase transitions in ion-containing block copolymer membranes,” Macromolecules, 2008, 41, 2271–2277. 77. Nieh, M. P.; Guiver, M. D.; Kim, D. S.; Ding, J.; Norsten, T. “Morphology of comb-shaped proton exchange membrane copolymers based on a neutron scattering study,” Macromolecules, 2008, 41, 6176–6182.

R Journal of Macromolecular Science , Part C: Polymer Reviews, 50:40–58, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583720903503478

X-ray Scattering for the Monitoring of Processes in Polymer Materials with Fiber Symmetry NORBERT STRIBECK Department of Chemistry, University of Hamburg, Hamburg, Germany Synchrotron facilities with improved equipment will grant the execution of very advanced scattering experiments to every interested polymer scientist. The patterns recorded are two-dimensional, high-resolution, and low-noise images. They permit to monitor structure evolution during the processing of polymer materials with repeat rates of several Hertz. In a different class of novel experiments the structure gradient inside of graded polymer materials will be determined with a spatial resolution of less than 1 µm. Scattering patterns must be complete. In order both to record complete patterns, and to evaluate them within tolerable time it appears reasonable to study parts with fiber symmetry. In the present paper a review of corresponding methodical work and the results of test experiments is presented that has recently been compiled in the group of the author. Keywords

polymers, fibers, WAXD, SAXS, tomography

1. Introduction Presently we are experiencing that intense, reliable X-ray sources and fast detectors become generally available. At DESY in Hamburg PETRA III will provide one of the most brilliant X-ray sources worldwide. There less than a second will be sufficient to expose a lownoise two-dimensional (2D) scattering pattern of a polymer sample. Moreover, in detector development a similar breakthrough has been achieved. Cycle times of less than a second can easily be realized with the novel PILATUS1, 2 detectors. It may be objected that for a long time low-noise scattering curves can be recorded with repeat frequencies of several Hertz. Nevertheless, the value of those older data is limited. Either the material is exhibiting isotropic scattering—in this case a comprehensive analysis requires assumptions on the structure (e.g. the assumption of an ideal lamellar stack). Or the material is anisotropic—then a measured curve does not describe the complete scattering pattern or the old 2D pattern is noisy. On the other hand, by application of the new technology it will become possible for many polymer scientists to monitor the structure evolution in oriented polymer materials in real time. The results of such experiments offer the potential to strengthen the understanding of structure evolution mechanisms in polymers. Moreover, the new instruments provide very narrow X-ray beams with diameters of less than 1 µm (“microbeams”) that permit to study the variation of structure inside the material with respective spatial resolution. A complete sweep of a polymer fiber by a microbeam Received September 21, 2009; accepted November 5, 2009. Address correspondence to Norbert Stribeck, Department of Chemistry, University of Hamburg, D-20146 Hamburg, Germany. E-mail: [email protected]

40

Monitoring of Processes in Polymer Materials

41

will take ca. 30 s. Thus, spatial resolution and time resolution may be combined to study the response of different annular zones in a polymer fiber to, e.g. mechanical or thermal load. Parts with fiber symmetry, like fibers, rods, or tubes, are not only of practical interest because they are ubiquitous in everyday life. Moreover, they appear particularly suited for investigation by means of scattering methods. As the part is rotated about its axis, the 2D scattering pattern does not change. Thus, there is no need to take patterns at different rotation angles, as it would be necessary for materials with less symmetry. The recorded 2D scattering pattern of a fiber is oriented and contains all the accessible information on the structure of the sample. Nevertheless, it will not be sufficient to simply engage the novel instruments—the anticipated data must be adequately evaluated in due time, as well. In this connection two big problems are arising. On the one hand, the data flood is increasing to such an extent that automated evaluation methods must be developed. On the other hand, the data structure has changed, in principal. Instead of scattering curves scattering images must be evaluated. In fact, reasonable scattering curves can be extracted from scattering images as well,3 if loss of information is accepted. Nevertheless, it appears more reasonable to develop evaluation methods adapted to the processing of images as a whole. This goal can be achieved by combining digital image processing and scattering theory.4 In this review several methods are presented that have recently been developed in our group in order to quantitatively evaluate extensive sets of 2D scattering patterns from polymer materials with fiber symmetry. Two previously published reviews deal with other aspects of the scattering experiments from polymer materials, namely a presentation of scattering theory5 from the point of view of the materials scientists and a presentation of instrumental development.6

2. Practice of Experiment and Data Analysis It is not seldom that users who have carried out experiments at a synchrotron source return with incomplete data. In the worst case only images (the TIFF files from the detector) have been collected and relevant environmental data are missing. Such data are the primary beam intensities before and after the sample, exposure time, exposure mode, time stamp, etc. More frequently the user has forgotten to record a machine background or parameters of the setup are incomplete. Such errors could be reduced if the operator of the synchrotron source would offer a data pre-evaluation service. The required data and the steps of data preevaluation are described in a text book4 of the author. With respect to the description given in the book the procedure of machine-background elimination has recently been changed so that it becomes consistent with tomographic experiments, as well. Now the intensity in the scattering pattern is divided by the linear absorption factor of the sample and from this image the measured machine background is subtracted. This method compensates intensity loss by absorption inside the sample. For the professional analysis of anisotropic 2D scattering patterns of polymer materials there are no user-friendly standard computer programs, because materials scientists do not carry out standard experiments (like, e.g. protein crystallographers). Therefore some program modules must be adapted to the experiment in order to reflect the actual setup or the actual mode of operation. After adaption and proper combination of modules the data can be evaluated automatically. The other option is cumbersome frame-by-frame manual evaluation that is prone to inconsistency or incompleteness.

42

N. Stribeck

The sequence of evaluation steps is a function of several parameters like the setup, the individual scattering power of the studied material, and the features of the anisotropic scattering pattern. New program modules must be constructed if the experimental procedure is fundamentally changed. Complex correction modules may be required if experimental shortcomings are detected during data evaluation. A typical shortcoming is insufficient synchronization among the different recorders of the experiment. Thus, the acquisition of programming skills is recommended. Moreover, it is helpful to choose a computing language that is optimized for the processing of multidimensional data. Suitable commercial platforms are, e.g. PV-WAVE, IDL or MatLab. If free programming tools shall be used, ImageJ offers a good starting point for arduous programmers. Actual references are readily found by search engines on the worldwide web. Our group uses PV-WAVE.7 The sources of the modules are free and available.8

3. WAXD Fiber Mapping 3.1 Motivation and Method Design In wide-angle X-ray diffraction (WAXD) experiments the diffraction patterns must be mapped into reciprocal space before they can be analyzed quantitatively. For this purpose interactive computer programs9, 10 are utilized that rest upon unnecessary11 approximations. Such a design is no disadvantage in crystallography, because sophisticated interactive refinement methods are required anyway for the exact determination of crystal structure parameters in manageable series. In contrast, in materials science frequently time-resolved experiments are carried out, and voluminous series of diffraction patterns must be processed. The materials scientist already knows the unit cell parameters. Thus, minor inaccuracy of the mapping can be tolerated, if in the experiment variation of peak intensity or shape shall be monitored. Here it is important to carry out the mapping fast and automatically. Because the fiber tilt may change during the experiment, the algorithm must be able to track and to compensate such variation. By revisiting the theoretical treatment of the fiber mapping it has been demonstrated11 that there is no principal reason to refine an approximate center of the fiber pattern iteratively. Moreover, instead of an approximation12, 13 of the tilt angle β of the fiber an exact equation11 can be employed. In the methodical paper14 an algorithm is presented by which the mapping can be performed automatically. Its design rests on the application of the mentioned findings. Intricate parametrization is simplified, and slow trigonometric functions are avoided to a large extent. The method is unsuitable for diffuse scattering patterns. If inaccuracies of 2 pixels can be tolerated, a pattern is automatically mapped into reciprocal space in real time.

3.2 Actions Required by The User For each series of diffraction patterns from a time-resolved experiment, some mapping parameters must once be determined interactively. Our procedure wf premap assumes that the studied material exhibits a sharp reflection that is located neither on the equator nor on the meridian. In the example we investigate polypropylene and select the (131) reflection of the crystallographic α2 -modification as the internal standard. With the crystallographic c-axis parallel to the meridian, the reflection is characterized by the parameters15 dhk = d131 = 0.406 nm and by c/ = 0.6504 nm that defines the position of the reflection ring

Monitoring of Processes in Polymer Materials

43

Figure 1. (a) Interactive mode of wf premap: Draw the reflection circle through the centers of the reflection spots of an internal standard reflection (here: polypropylene (131) reflection), and widen the circle into a belt that contains the maxima of the reflection spots. Finally input 4 circles (one indicated at bottom right). (b) The procedure wf map has mapped the fiber diffraction pattern into reciprocal space.

on the Polanyi sphere.11, 16, 17 The wavelength of X-radiation (here: λ = 0.15 nm) must be known, too. In this case the first pre-mapping run of the series is started by wave> ab = wf premap(ss,a,0.15,0.406,0.6504)

The procedure requires a diffraction pattern as input (a). It generates output both in a “saveset” (ss), and in a background-corrected diffraction pattern (ab). If the provided save-set has never been used before, the procedure enters interactive mode and the user is presented the pattern as shown in Fig. 1. Obviously, in Fig. 1a the meridian is not vertical and the fiber is tilted. By means of the pointing device the user draws the reflection circle through the maxima of the reflection spots of the (131)-reflection. Thereafter he transforms the circle into a belt that is wide enough to contain the maxima of the spots (Fig. 1a). Now the procedure cuts out this belt and presents it to the user, who finally specifies 4 disjoint clips (regions of interest on the reflection circle). This is done by punching out circular regions from the belt (the bottom right one is shown in Fig. 1a. The clips are the intersections of the belt (white double-band) and the smaller circles. The program determines the positions of the maxima in the clips. From these 4 maxima-positions the best reflection circle is determined by regression. The error of determination is computed. In general, it is below 1 pixel. After that the orientation of the meridian is computed both from the upper and from the lower pair of spots. The difference is, in general, approximately 0.1o . Based on this information the diffraction pattern is centered and aligned. Finally, the program computes the tilt angle β of the fiber from the orientation angles11 δ and δ of the spots using the exact equation11 4 − λ2 sr2 tan β = (cos δ − cos δ). 2λsr Here sr = 1/d131 is the radius of the Polanyi sphere of the reference (131)-reflection. Finally, the save-set ss is filled with several data: the tilt angle β; the true radius pr

44

N. Stribeck

of the reference-reflection circle on the detector; the crystallographic data of the reference reflection; the wavelength λ; and the geometrical data of the 4 clips. This information is sufficient for an automatic processing of the complete series of diffraction patterns recorded in a time-resolved X-ray diffraction experiment of materials with fiber symmetry. 3.3 Automated Mapping If the procedure wf premap is called with a filled save-set, the interactive part is skipped, and the geometry of the 4 clips is taken from the save-set. For each clip the position of the spot maximum is computed. From these positions the fiber tilt angle β is computed, and the image is centered and aligned. Finally, the mapping into reciprocal space is accomplished by the routine wf map. The function call wave> rec=wf map(ss,ab)

maps the diffraction pattern ab into reciprocal space using the save-set ss. The result is called rec. A more detailed description is in the original paper.14 Figure 1b shows a typical result obtained by automatic direct mapping. In the example the computed tilt angle is β = 5.85◦ . Because the result pattern is in reciprocal space, it should exhibit symmetry in 4 quadrants. Thus, the quality of the mapping can be assessed by comparison to a 4-quadrant average of the pattern. 3.4 Application In a study18 that applies the method, uniaxially oriented polypropylene (PP) is molten and crystallized isothermally from the oriented, quiescent melt. The results show that nucleation and growth of differently oriented sets of crystallites (c-set and a ∗ -set) are decoupled. After shallow quench crystallization is preceded by (spinodal) decomposition. Peak integrals (crystallinity) and minimum crystallite size are tracked. In the commercial starting material a ∗ -set crystallites melt at 158◦ C. The c-set melts at 170◦ C furnace temperature. After recrystallization both sets melt at 170◦ C. Isothermal crystallization is divided in two distinct phases. During nucleation the crystallinity stays low. The second phase is dominated by crystallinity growth. At 150◦ C the c-set is seeded first. At 145◦ C and 140◦ C a ∗ -oriented crystallites are the first. The first-seeded set starts to grow first, as well. c-set crystallinity is always growing faster than a ∗ -set crystallinity. The evolution of the corresponding SAXS19 cross-diagram in the growth phase can both be explained by lamellae growing at right angles, and by block merging. Figure 2 shows β (t, T ) of one of the experiments. The tracking curve appears smooth and demonstrates the reliability of the tilt-angle determination. Tilt-angle variation is an issue, because the oriented PP film is heated until it becomes a viscous melt. Therefore the material shrinks and bends in the synchrotron X-ray beam. After the mapping the intensity distribution is known almost everywhere in reciprocal space except for a wedge region near the meridian (cf. Fig. 1b). Thus, reflection intensities can readily be integrated in reciprocal space. In reciprocal space the relation between scattering and structure is clear from scattering theory,4, 18 and structural data can be computed,18 e.g., from total intensities (crystallinity) and integral breadths (crystallite sizes). Figure 3 shows the evolution of the weight crystallinities of the two sets of crystallites at 3 different crystallization temperatures. As shown by Ruland,20 such reflection integrals that

Monitoring of Processes in Polymer Materials

45 200

20

10

T(t)

T [°C]

β(t) [°]

150 0 100 -10 50 -20 -20

-10

0

10

20

30

40

t [min] Figure 2. Tilt-angle tracking curve β (t, T ) from the automatic fiber-mapping procedure in an experiment in which β is changing considerably (hard-elastic polypropylene; melt-annealing at 171◦ C 18 c with permission of the ACS). and recrystallization at 150◦ C) (Reproduced

are complete in reciprocal space are proportional to the weight crystallinity of the perfect crystallites that produce the reflections. Latency periods between the quench and the start of the crystallization, as well as crystallization velocities of the two kind of crystallites can be extracted from these data. Finally, conclusions concerning the crystallization mechanisms can be drawn.18

Figure 3. Evolution of relative weight crystallinities S, and S ∗ of c-oriented crystallites (dashed regression lines) and a ∗ -oriented crystals (solid regression lines), resp. during isothermal, oriented crystallization of hard-elastic PP from a quiescent melt as a function of crystallization temperature. Double-head arrows pointing at the t-axis indicate the first sighting of the a ∗ -set (full arrow head), and of the c-set (open arrow head), respectively.

46

N. Stribeck

4. X-Ray Scattering Fiber Tomography 4.1 Motivation The structure inside a polymer part is not necessarily homogeneous. With fibers, extrudates, and latex particles21 frequently a core-shell structure is reported. Moreover, engineers deliberately generate structure gradients in polymer parts for special functionality.22, 23 If such a part is irradiated by an X-ray beam, the recorded pattern is an integral superposition of all the SAXS patterns emerging from the sequence of volume elements (voxels) along the beam path. From the mathematical point of view such a superposition is a projection, and a single projection is of little use for the study of graded materials. A first step towards a study of structure gradients has been the development of the X-ray microbeam technique.24–26 Here only the diameter of the beam is limiting the lateral spatial resolution. Nevertheless, the longitudinal spatial resolution is simply the thickness of the sample. Microbeam scanning experiments, in particular of fibers, have been performed for many years and the raw data have been discussed, although the corresponding shortcoming has been known (Paris et al.:27 “The long-term goal is to proceed from microbeam scanning experiments to a real imaging technique”). The solution of the problem is tomographic reconstruction. Problems arise from the fact that scattering patterns are multidimensional but not simply a number (like the absorption in classical tomography). Thus, approximate tomographic reconstruction of scattering data with a manageable amount of artifacts is difficult. An exception is the case of part with uniaxial symmetry. In this case a reconstruction of the scattering patterns is possible that would have emerged from individual voxels in a plane perpendicular to the “fiber” axis.28 Nevertheless, this method is only of academic value. The exposure time for the recording of scattering data of one cross-section of the part is in the order of days. The computing time for the tomographic reconstruction of the scattering patterns is at least a week.28

4.2 Introduction of the Method A more practical tomographic method can be applied, if the part to be studied both exhibits macroscopic fiber symmetry, and the structure only varies as a function of the distance from its central axis. By means of this method fibers, pipes, and extruded strands can be investigated. Thus, we call it “X-ray scattering fiber computer tomography” (XSF-CT). A complete set of projected scattering patterns is collected in a single microbeam scan across the fiber, because the set of projections does not change as the sample is rotated about its axis. Such an experiment is completed in about 30 min. Moreover, compared to the general tomography the mathematics of image reconstruction is simplified considerably and the computational effort decreases by 5 orders of magnitude. A set of 40 measured scattering patterns is reconstructed in typically less than 10 min. From medicine and other fields of science the considerable potential of information increase after tomographic imaging is well-known. In a general tomographic X-ray experiment,29 a voluminous sample is scanned by a thin X-ray beam. As a function both of the position x of the scanning beam on the sample, and of the sample rotation angle φ, projections (notation: { }) of the absorption {A} (x, φ) or even of complete scattering patterns {I } (s, x, φ) are measured, in order to analyze the structure variation in the plane of the sample that is scanned by the X-ray beam.28, 30, 31 Here s is the scattering vector with |s| = s = (2/λ) sin θ , the X-ray wavelength λ and the scattering angle 2θ . In the examples mentioned, a tomographic image reconstruction29, 32, 33 returns

Monitoring of Processes in Polymer Materials

47

Figure 4. (a) cross-section irradiated by an X-ray beam at an offset x from its center. The Fiber structure ρ ρf shows fiber symmetry. From all structures along the beam path a superposition is probed. y is the variable of the integration. (b) One-dimensional tomographic reconstruction turns the measured series of projected scattering patterns that carry the accumulated the structure information passed by the beam (vertical bars) into the image patterns from voxels (quadratic boxes) residing on the fiber radius.

either the spatial variation of the absorption in the plane, or of the scattering emanating from the resolved voxels in the plane. The smearing caused from projection is eliminated by application of the Fourier transform theory, and a clear image of the inner structure is obtained. If the studied material shows cylindrical symmetry, the results of the measurement are no function of φ any more, and the complete image information is in a single microbeam scan. The fundamental geometry is sketched in Fig. 4a. The information in the measured signal {A} (x) or {I } (s, x), respectively, does not represent the sought information A (x) or I (s, x) originating from the small square (voxel) around the position x. Instead, to a first approximation it is represented by the projection integral ˆ ∞ {I } (s, x) = 2 I (s, x 2 + y 2 ) dy (1) ˆ

x ∞

=2 x

I (s, ρf ) ρf dρf . ρf2 − x 2

(2)

In the equation only the accumulated attenuation of the primary beam by X-ray absorption is not accounted for. This is no problem for very thin polymer parts with low absorption. For thicker samples the absorption correction from Section 2 is sufficient to account for it. The sought information in image space (I (ρf )) along the radius ρf of the fiber has to be reconstructed from the information in projection space ({I } (x)). Equation (1) is the definition of the Abel transform.33 In X-ray scattering Eq. (1) is established textbook knowledge.4, 34–39 There it describes the slit smearing. Even the inverse Abel transform ˆ 1 ∞ d{I }(s, ρf ) dy I (s, x) = − (3) π 0 dρf ρf ˆ 1 ∞ d{I }(s, ρf ) dρf , (4) =− π x dρf ρf2 − x 2

48

N. Stribeck

Figure 5. Sketch of local fiber symmetry inside the cross-section of the fiber. Each voxel has its own fiber axis. Inside the voxel there is only axial grain.

which first has been derived by Niels Abel40 is found in scattering textbooks since Guinier41 and DuMond.42 Similar to the filtered backprojection algorithm of the general tomography, low-noise reconstruction algorithms43–45 for the tomography of materials with cylindrical symmetry are readily available in the field of “one-dimensional tomography.” The principle of one-dimensional tomographic reconstruction is sketched in Fig. 4b. It must be mentioned that here we implicitly assume that the scattering from every irradiated voxel in the fiber shows fiber symmetry itself (local fiber symmetry). Otherwise characteristic reconstruction aberrations are expected.28, 46 The meaning of local fiber symmetry is sketched in Fig. 5. Deviations from local fiber symmetry (i.e. tangential or radial grain, resp.46 ) cause restricted or shifted visibility of scattering features along the fiber radius. These aberrations can be detected and result in additional information on the structure inside the fiber.46 4.3 Application In an application-oriented feasibility study46 precursors of polymer microfibrillarreinforced composites (MFC) containing poly(ether)-block-amide (PEBA) and poly (ethylene terephthalate) (PET) with varying cold-draw ratio are studied. The studied strands are relatively thick, because presently the achievable “microbeam” at HASYLAB in Hamburg is relatively wide. The results from a direct analysis of the smeared measured patterns are compared to results obtained after tomographic reconstruction. Ideas for advanced practical applications of the XSF-CT method are discussed. Data are presented from a cold-drawn (draw ratio λd ≈ 3) co-extrudate of 70 wt.-% PEBA and 30 wt.-% PET (abbreviated: MFC73). In the scanning-microbeam experiment

Monitoring of Processes in Polymer Materials

49

Figure 6. Cold-drawn (λd ≈ 3) MFC73 in a scanning microbeam experiment. Measured scattering intensity {I }(s12 , s3 , ρf ) (top row) and reconstructed scattering I (s12 , s3 , ρf ) (bottom row) for short distances ρf from the fiber axis. The patterns display the range −0.1 nm−1 ≤ s12 , s3 ≤ 0.1 nm−1 in uniform logarithmic scale (s = (2/λ) sin θ).

the strand shows an isotropic long period and an equatorial streak at almost every beam position (Fig. 6, top row). Only the tomographically reconstructed patterns (bottom row) exhibit that the long-period ring-reflection is not present in the core of the fiber. As the feature becomes visible, it first shows up at the equator. With increasing distance from the fiber axis, reflection arcs are growing towards the meridian. Above ρf = 300 µm the arcs join into a closed circle. Because such behavior has been observed with neat PEBA as well, this orientation phenomenon is not indicating some interaction between the PEBA and the PET microfibrils. The reconstructed central voxel (ρf = 0) is dominated by one of the reconstruction aberration effects (from voxels with tangential grain).46 Examination of the equatorial streak exhibits only in the reconstruction that it grows broader towards the center of the fiber. The streak is allocated to needle-shaped domains, and is only observed with strands from co-extruded blends. If these needles are thin PET microfibrils, the tomography shows that in the center of the fiber these microfibrils are shorter than in an intermediate region. The structure gradient in the outer region of the strand is discussed in the original paper.46

5. SAXS Monitoring of Mechanical Tests 5.1 Motivation and Method Development Advanced polymer materials are urgently sought after e.g. in the automotive industry in order to accomplish the goals of climate protection by reduction of weight. Such newly engineered materials have to prove their serviceability in mechanical tests. Classical tensile tests are performed to determine the modulus and ultimate properties. Load cycling experiments are carried out to determine the fatigue behavior. In order to reveal the mechanisms of, e.g., failure or fatigue, it is desirable to monitor mechanical tests by SAXS. In this way the response of the nanostructure to mechanical load is revealed. Since about 2005 the advance of instrumentation at synchrotron beamlines facilitates considerable reduction of the exposure required for the recording of low-noise SAXS patterns. Such quality is

50

N. Stribeck

required for a quantitative analysis by, e.g., the CDF method.47, 48 Concerning the monitoring of tensile tests by SAXS this advance has meant that change-over from the stretch-hold technique49 to the more practical continuous straining has become possible,50 although the achievable strain rate is frequently still by a factor of 100 lower than that relevant in industry and service. With the advent of very high-brilliance synchrotron sources even this limitation is presently being abolished. In order to gain full control on tensile and load-cycling tests a tensile tester has been R built for operation at synchrotron beamlines.50 The control program written in LabVIEW is continuously adapted to more complex experiments. Presently it can handle different kinds of continuous load-cycling programs that now are strictly synchronized with both the SAXS detector and a video grabber module (recording sequences of pictures of the sample with fiducial marks). The latest methodical development51 is an automatic evaluation method of the video frames, by which the true macroscopic elongation of the sample can be determined with an accuracy of 3 decimals. Such high accuracy is required for thermoplastic materials that fail at low elongations (ca. 0.1). The method extracts the grating of the fiducial marks on the sample from the actual video image, computes the 1D correlation function,52, 53 and evaluates its “long period.”

5.2 Results In a study50 of hard-elastic polypropylene (PP), scattering patterns recorded during continuous straining differ considerably from those recorded in the step-hold technique. Even though during exposure the elongation is no longer constant when applying the dynamic technique, the images collected in stretch-hold technique appear much more blurred (Fig. 7). This result indicates relaxation of nanostructure while the extensometer stands still. Quantitative analysis shows that during relaxation the extension of crystalline lamellae

Figure 7. Oriented PP films in tensile tests. Comparison between SAXS patterns I (s12 , s3 ) recorded during continuous straining (top row, ε˙ ≈ 10−3 s−1 ) and patterns from an experiment in stretch-hold technique (bottom row). Straining direction and meridian is vertical (s3 ). Equator (s12 ) is horizontal 50 c with permission of Wiley Interscience). (Reproduced

Monitoring of Processes in Polymer Materials

51

Figure 8. Oriented PP. Load cycling monitored by SAXS. Circular dots show where SAXS patterns have been recorded. Numerical labels indicate their sequence. The highlighted part of the curve near label 11 indicates the part traversed during the recording of pattern 11. The drawing direction with 48 c with permission of respect to the patterns, s3 , is indicated by a double-head arrow (Reproduced Wiley-VCH).

is increasing. Lamellae thicknesses are becoming non-uniform. The range of order is shortening. Cross-hatched lamellae are formed. In another study48 slow continuous mechanical tests of oriented PP are monitored by SAXS and quantitatively evaluated by the CDF method.47 A continuous-strain test exhibits fracture and release of weak lamellae (2 – 10% strain). Beyond that conversion of lamellae into needles is observed. As all layers are consumed, the material breaks. Fatigue is studied in a load-reversal experiment (between 10% and 35% strain, Fig. 8). In each cycle crystallization, layer break, and relaxation melting are observed. Figure 9 presents the result. The top chart shows the true elongation, ε (t), imposed to the material and measured at the point of X-ray irradiation. The four load-reversal cycles are easily identified. The chart in the middle reports the extracted nanostructure parameters, L (t), ecac (t), and S (t). The bottom diagram presents the macroscopic resistance, σ (t), which the material is opposing to strain. The figure shows that crystallization, rupture of lamellae, and melting of fragments are continuously reshaping the domains of the PP material in the cycles of the fatigue test. The long-period cycle exhibits a phase shift with respect to the imposed strain cycling. Moreover, there is an indication of amplitude attenuation. Fatigue is demonstrated in the curve σ (t) by decreasing stress peaks. An in-depth discussion of the zones indicated in Fig. 9 is in the original paper.48 In particular, the combination of SAXS and fatigue test shows the transition from stress-induced crystallization to crystallite rupture σ (t) ≈ 20 MPa. In a recent study51 of microfibrillar reinforced blends based on polyethylene we have switched from hard-elastic materials to materials that fail at the low elongations typical for plastic polymers. Because of the much lower elongation at break (εb ≈ 0.1), now the macroscopic elongation εm and nanoscopic elongation εn must be determined with high precision. Respective methods are presented. The results show that the hardest materials exhibit a very inhomogeneous nanodomain structure. During straining their domains appear to be wedged together and inhibit transverse contraction on the nanometer scale.

52

N. Stribeck

Figure 9. Dynamic load-reversal mechanical test of hard-elastic PP film. As a function of the elapsed time t the macroscopic parameters elongation, ε (top graph), and tensile stress, σ (bottom graph) are displayed together with topological nanostructure parameters (middle). In the middle diagram the solid line shows the long period, L. The broken line displays the lateral extension, ecac , of a sandwich made from two crystalline lamellae. The line with circular dots exhibits the variation of the strength, S, of the CDF. Vertical bars indicate zones of strain-induced crystallization (dark bars) c Reproduced48 with permission of and relaxation-induced melting (light gray bars), respectively ( Wiley-VCH).

Further components are polyamides (PA6, PA12) (20–30%) and as compatibilizer R 8102 (YP) (0–10%). Some HDPE/PA6 blends are additionally loaded with nanoYparex R R clays (Nanomer or Cloisite ). Blending of HDPE with PA12 causes no synergistic effect. In the absence of nanoclay, PA6 and HDPE form a heterogeneous nanostructure with high Young’s modulus. After addition of YP a more homogeneous scaffold structure is observed in which some of the PA6 microfibrils and HDPE crystallites appear to be rigidly connected, but the modulus has decreased. Both kinds of nanoclay induce a transition from a structure without transverse correlation among the microfibrils into a macrolattice with 3D correlations among HDPE domains from neighboring microfibrils. For extensions between 0.7% and 3.5% the scattering entities with 3D correlation exhibit transverse elongation instead of transverse contraction. The process is interpreted as overcoming a correlation barrier executed by the crystallites in an evasion-upon-approaching mechanism. During continued straining the 3D correlation is reduced or removed. The true macroscopic elongation εm is determined from video frames taken of the sample that carries fiducial marks (see Fig. 10a). Once for an experiment the user has to provide some input. It is based on the first image (Fig. 10a) of the series. The center of the X-ray beam on the sample is marked by a cross in the image. Close to this center the user defines a rectangular region of interest (ROI), ρm (x, y). In Fig. 10a this region is bordered by a dashed line. x and y are pixel coordinates in the direction of strain and perpendicular to it, respectively. The same ROI is applied to all video frames of the experiment. The ROI is structured by the fiducial marks running perpendicular to the straining direction. As is known from scattering theory,4 from such a structure function ρm (x, y) the 2D correlation

Monitoring of Processes in Polymer Materials

53

Figure 10. Elongation from recorded video frames. Inset a: In the first video frame a region of interest (ROI) with fiducial marks is defined. Inset b: From the ROI the 2D correlation function γ2 (x, y) is computed. Main drawing: The center of the long-period peak in γ1 (x) = γ2 (x, 0) is fitted c Reproduced51 by a parabola (dashed line) to compute the distance between the fiducial marks ( with permission of Wiley Interscience).

function γ2 (x, y) =

ρm 2 (x, y) ρm 2 (0, 0)

can be computed, with ρm (x, y) = ρm (x, y) − ρ¯ m representing the fluctuation of ρm (x, y) about its average ρ¯ m , and the autocorrelation being defined by the integral ¨ ∞ f 2 (x, y) = f (u, v) f (u + x, v + y) dudv. −∞

In Fig. 10b the colored caps demonstrate, where γ2 (x, y) is positive. The macroscopic elongation εm in straining direction is the section γ1 (x) = γ2 (x, y) 1 (x) of γ2 in straining direction. Figure 10 presents this curve and its analysis. Its first positive peak is the longperiod peak that is related to the actual average distance of the fiducial marks, . A parabola (dashed line) is fitted to the long-period maximum, and the position of its vertex is determined (arrow). Thus, can be determined with an accuracy of 0.01 pixels. Let 0 the initial distance between the marks, then the macroscopic elongation is εm = /0 − 1. Concerning the scattering patterns, similar analysis is possible. Nevertheless, the peaks observed in the scattering pattern or in the CDFs47 computed from the pattern are no longer one-dimensional but two-dimensional – positioned in the plane (s12 , s3 ) or (r12 , r3 ), respectively. Again, the user defines a ROI in which the analysis procedure searches for the peak maximum, extracts the peak cap, and fits now a bivariate54 polynomial of 2nd order to it. From the coefficients of the polynomial several parameters are readily determined. Most important are the peak position and the widths of the peak in meridional and equatorial

54

N. Stribeck

Figure 11. MFC precursor blends from HDPE, two different polyamides (PA12, PA6) and a compatibilizer (YP) in tensile tests. Evolution of macroscopic stress and strain (σ , ε) as well as of nanostructure parameters. εnano is the nanoscopic elongation computed from the HDPE long period. DL is the relative change of the width of the long period distribution. DM is the relative change of the extension of the microfibrils in transverse direction.

directions, respectively. Thus, the evolution of these parameters during the experiment can be tracked automatically. Figure 11 presents the results of a quantitative nanostructure analysis for the blends which do not contain nanoclays. Elongations are illustrated by dashed lines. Bold lines show the macroscopic elongation, ε. Thin lines report the nanoscopic elongation εnano of the HDPE matrix. Circular marks indicate regions in which ε ≈ εnano . All materials reinforced by PA12 exhibit this similarity of macroscopic and nanoscopic deformation. Dashed-dotted lines show DL , the relative variation of the breadth of the long-period

Monitoring of Processes in Polymer Materials

55

distribution. The increase of all curves demonstrates increasing inhomogeneity of the long periods with increasing elongation. Dotted lines show DM , the relative variation of the microfibrillar diameter during the straining process. In all materials the elongational deformation causes the microfibrils to thin. Variation of the material composition does not cause considerable changes. This is different with the samples reinforced by PA6 (Fig. 11d-f). Here an increase of the PA6 content to 30% causes strong thinning of the microfibrils. In the PA6 blends the strong transverse decrease is going along with only moderate nanoscopic elongation εnano of the HDPE. An explanation for this finding could be microfibrillation by fracture of crystalline domains of the polyethylene. Moreover, the diagrams in Fig. 11d-f demonstrate a considerable difference (vertical arrows) between the two dashed curves. In Fig. 11e-f (εnano < ε) the nanoscopic elongation of the HDPE phase is considerably lower than the macroscopic elongation. Similarity is only observed during the initial deformation in Fig. 11f (circular mark). In the 80/20 HDPE/PA6 blend (Fig. 11d) the nanoscopic elongation of the HDPE microfibrils is considerably longer than the macroscopic elongation (εnano > ε). Although this finding appears to be unreasonable, an indication for a possible mechanism is in the strong increase of DL (Fig. 11d). This is discussed in the original paper.51

6. Conclusions Considering the present instrumental development at synchrotron radiation facilities the development of advanced data evaluation methods appears to be both promising and necessary in order to master the future data flood. The three presented methods demonstrate the potential of such work. In order to discharge the user, a part of the data evaluation may be carried out at the synchrotron facility. Such added service would require not only considerable computing power, but also additional manpower. In addition to the beamline scientist an evaluation specialist would become necessary. It would be his job to detect if raw data must be smoothed. He would have to eliminate the machine background, would generate detector masks, would center and align each scattering pattern, and would fill blind areas from consideration of symmetry. As an added service, the community of the evaluation specialists could select some standard experiments for which complete user-friendly programming environments could be built. Nevertheless, for the predominant fraction of individually designed setups it will remain necessary for the polymer scientist himself to familiarize with adapted programming techniques. This will be of particular importance, if methods shall be developed that grow with the growth of instrumental capacity. Ultimately, it is expected that an increase of the quality of results returned from scattering experiments will be closely correlated to the manpower dedicated to the programming of data evaluation modules.

Acknowledgment The author thanks the Hamburg Synchrotron Radiation Laboratory (HASYLAB) for beam time granted in the frame of project II-20080015. Development of the reported methods has been supported by the 7th framework program of the European Union (Project NANOTOUGH FP7-NMP-2007LARGE-2.1.1).

References 1. Broennimann, C.; Eikenberry, E. F.; Henrich, B.; Horrisberger, R.; H¨ulsen, G.; Pohl, E.; Schmitt, B.; Schulze-Briese, C.; Suzuki, M.; Tomizaki, T.; Toyokawa, A.; Wagner, A. “The PILATUS 1M detector,” J. Synchrotron Rad., 2006, 13, 120–130.

56

N. Stribeck

2. H¨ulsen-Bollier, G. The PILATUS 1M Detector. A Novel Large Area Pixel Detector, Ph.D. thesis, Dept. Physics, University of Erlangen-N¨urnberg, Germany 2005. 3. Stribeck, N. “Analysis of SAXS Fiber Patterns by Means of Projections,” ACS Symp. Ser., 2000, 739, 41–56. 4. Stribeck, N. X-Ray Scattering of Soft Matter, Springer, Heidelberg, New York, 2007. 5. Stribeck, N. “Scattering of Soft Condensed Matter. From Fundaments to Application,” In Applications of Synchrotron Light to Scattering and Diffraction in Materials and Life Sciences, Ezquerra, T. A.; Garci´a Guti´errez, M.; Nogales, A.; G´omez, M., eds., Springer, Berlin Heidelberg, 2009, volume 776 of Lect. Notes Phys. pp. 25–62. 6. Stribeck, N. “Deformation behavior of nanocomposites studied by X-ray scattering: Instrumentation and methodology,” In Nano- and Micromechanics of Polymer Blends and Composites, Karger-Kocsis, J.; Fakirov, S., eds., Hanser Publisher, M¨unchen, 2009, volume 1 pp. 269– 300. 7. VNI “PV-WAVE Manuals,” V 7.5, Houston, TX, 2007. 8. Stribeck, N. “Downloads,” http://www.chemie.uni-hamburg.de/tmc/stribeck/dl 2008. 9. Rajkumar, G.; AL-Khayat, H.; Eakins, F.; He, A.; Knupp, C.; Squire, J. “FibreFix—A new integrated CCP13 software package,” Fibre Diffraction Rev., 2005, 13, 11–18. 10. Bian, W.; Wang, H.; McCullough, I.; Stubbs, G. “WCEN: a computer program for initial processing of fiber diffraction patterns,” J. Appl. Cryst., 2006, 39, 752–756. 11. Stribeck, N. “On the determination of fiber tilt angles in fiber diffraction,” Acta Cryst., 2009, A65, 46–47. 12. Franklin, R. E.; Gosling, R. G. “The structure of sodium thymonucleate fibres. II. Cylindrically symmmetrical patterson function,” Acta Cryst., 1953, 6, 678–685. 13. Fraser, R. D.; Macrae, T. P.; Miller, A.; Rowlands, R. J. “Digital processing of fibre diffraction patterns,” J. Appl. Cryst., 1976, 9, 81–94. 14. Stribeck, N.; N¨ochel, U. “Direct mapping of fiber diffraction patterns into reciprocal space,” J. Appl. Cryst., 2009, 42, 295–301. 15. Mencik, Z. “Crystal structure of isotactic polypropylene,” J. Macromol. Sci. Phys., 1972, B6, 101–115. 16. Polanyi, M. “Das R¨ontgen-Faserdiagramm I. (The X-Ray Fiber-diagram I.),” Z. Phys., 1921, 7, 149–180. 17. Polanyi, M.; Weissenberg, K. “Das R¨ontgen-Faserdiagramm II. (The X-Ray fiber-diagram II.),” Z. Physik, 1923, 9, 123–130. 18. Stribeck, N.; N¨ochel, U.; Funari, S. S. “Melting and crystallization of differently oriented sets of crystallites in hard-elastic polypropylene,” Macromolecules, 2009, 42, 2093–2101. 19. Stribeck, N.; N¨ochel, U.; Almend´arez Camarillo, A.; Roth, S. V.; Dommach, M.; B¨osecke, P. “SAXS study of oriented crystallization of polypropylene from a quiescent melt,” Macromolecules, 2007, 40, 4535–4545. 20. Ruland, W. “X-ray determination of crystallinity and diffuse disorder scattering,” Acta Cryst., 1961, 14, 1180–1185. 21. Dingenouts, N.; Bolze, J.; Potschke, D.; Ballauff, M. “Analysis of polymer latexes by small-angle X-ray scattering,” Adv. Polym. Sci., 1999, 144, Epoxide Resins, Polyampholytes), 1–Epoxide Resins, Polyampholytes), 47. 22. Kieback, B.; Neubrand, A.; Riedel, H. “Processing techniques for functionally graded materials,” Mater. Sci. Eng. A Struct. Mater. Prop. Microstruct. Proces., 2003, 362, 81–106. 23. Pompe, W.; Worch, H.; Epple, M.; Friess, W.; Gelinsky, M.; Greil, P.; Hempel, U.; Scharnweber, D.; Schulte, K. “Functionally graded materials for biomedical applications,” Mater. Sci. Eng. A Struct. Mater. Prop. Microstruct. Proces., 2003, 362, 40–60. 24. Riekel, C.; Engstr¨om, P. “Diffraction and diffuse scattering from materials with microfocussed X-rays,” Nuclear Instr. Meth. Phys. Res., 1995, B97, 224–230. 25. Waigh, T. A.; Donald, A. M.; Heidelbach, F.; Riekel, C.; Gidley, M. J. “Analysis of the native structure of starch granules with small angle x-ray microfocus scattering,” Biopolymers, 1999, 49, 91–105.

Monitoring of Processes in Polymer Materials

57

26. Kolb, R.; Wutz, C.; Stribeck, N.; V. Krosigk, G.; Riekel, C. “Investigation of secondary crystallization of polymers by means of microbeam X-ray scattering,” Polymer, 2001, 42, 5257–5266. 27. Paris, O.; Li, C.; Siegel, S.; Weseloh, G.; Emmerling, F.; Riesemeier, H.; Erko, A.; Fratzl, P. “A new experimental station for simultaneous X-ray microbeam scanning for small- and wide-angle scattering and fluorescence at BESSY II,” J. Appl. Cryst., 2007, 40, s466–s470. 28. Stribeck, N.; Almendarez Camarillo, A.; N¨ochel, U.; Schroer, C.; Kuhlmann, M.; Roth, S. V.; Gehrke, R.; Bayer, R. K. “Volume-resolved nanostructure survey of a polymer part by means of SAXS microtomography,” Macromol. Chem. Phys., 2006, 207, 1239–1249. 29. Bonse, U.; Busch, F. “X-ray computed microtomography (µCT) using synchrotron radiation,” Prog. Biophys. Molec. Biol., 1996, 65, 133–169. 30. Schroer, C. G.; Kuhlmann, M.; Roth, S. V.; Gehrke, R.; Stribeck, N.; Almendarez Camarillo, A.; Lengeler, B. “Mapping the local nanostructure inside a specimen by tomographic small-angle x-ray scattering,” Appl. Phys. Lett., 2006, 88, 164102. 31. Schroer, C. G.; Kuhlmann, M.; G¨unzler, T. F.; Benner, B.; Kurapova, O.; Patommel, J.; Lengeler, B.; Roth, S. V.; Gehrke, R.; Snigirev, A.; Snigireva, I.; Stribeck, N.; Almend´arez Camarillo, A.; Beckmann, F. “Full-field and scanning microtomography based on parabolic refractive x-ray lenses,” Proc. SPIE, 2006, 6318, 6318H. 32. Kak, A. C.; Slaney, M. Principles of Computerized Tomographic Imaging; IEEE Press: New York, 1999. 33. Bracewell, R. The Fourier Transform and Its Applications; 3rd Ed. Mc Graw-Hill: New York, 1999. 34. Guinier, A.; Fournet, G. Small-Angle Scattering of X-Rays, Chapman and Hall: London, 1955. 35. Hosemann, R.; Bagchi, S. N. Direct Analysis of Diffraction by Matter; North-Holland: Amsterdam, 1962. 36. Alexander, L. E. X-Ray Diffraction Methods in Polymer Science; Wiley: New York, 1979. 37. Glatter, O.; Kratky, O., eds., Small Angle X-ray Scattering; Academic Press: London, 1982. 38. Feigin, L. A.; Svergun, D. I. Structure Analysis by Small-Angle X-Ray and Neutron Scattering; Plenum Press: New York, 1987. 39. Balt´a Calleja, F. J.; Vonk, C. G. X-Ray Scattering of Synthetic Polymers; Elsevier: Amsterdam, 1989. 40. Abel, N. H. “Aufl¨osung einer mechanischen Aufgabe,” J. Reine Angew. Math., 1826, 1, 153–157. 41. Guinier, A.; Fournet, G. “Correction of measurments of low-angle X-ray scattering,” Nature, 1947, 160, 501. 42. DuMond, J. W. M. “Method of correcting low angle X-ray diffraction curves for the study of small particle sizes,” Phys. Rev., 1947, 72, 83–84. 43. Bitter, I.; Kaufman, A. E.; Sato, M. “Penalized-distance volumetric skeleton algorithm,” IEEE Trans. Visualization and Computer Graphics, 2001, 7, 195–206. 44. Dasch, C. J. “One-dimensional tomography: a comparison of Abel, onion-peeling, and filtered backprojection methods,” Applied Optics, 1992, 31, 1146–1153. 45. Dribinski, V.; Ossadtchi, A.; Mandelshtam, V. A.; Reisler, H. “Reconstruction of Abeltransformable images: The Gaussian basis-set expansion Abel transform method,” Rev. Sci. Instr., 2002, 73, 2634–2642. 46. Stribeck, N.; N¨ochel, U.; Fakirov, S.; Feldkamp, J.; Schroer, C.; Timmann, A.; Kuhlmann, M. “SAXS-fiber computer-tomography. Method enhancement and analysis of microfibrillarreinforced composite precursors from PEBAX and PET,” Macromolecules, 2008, 41, 7637–7647. 47. Stribeck, N. “Extraction of domain structure information from small-angle X-ray patterns of bulk materials,” J. Appl. Cryst., 2001, 34, 496–503. 48. Stribeck, N.; N¨ochel, U.; Funari, S. S.; Schubert, T.; Timmann, A. “Nanostructure evolution in polypropylene during mechanical testing,” Macromol. Chem. Phys., 2008, 209, 1992–2002. 49. Wu, J.; Schultz, J. M.; Yeh, F.; Hsiao, B. S.; Chu, B. “In-situ simultaneous synchrotron small- and wide-angle X-ray scattering measurement of poly(vinylidene fluoride) fibers under deformation,” Macromolecules, 2000, 33, 1765–1777.

58

N. Stribeck

50. Stribeck, N.; N¨ochel, U.; Funari, S. S.; Schubert, T. “Tensile tests of polypropylene monitored by SAXS. Comparing the stretch-hold technique to the dynamic technique,” J. Polym. Sci. Polym. Phys., 2008, 46, 721–726. 51. Denchev, Z.; Dencheva, N.; Funari, S. S.; Motovilin, M.; Schubert, T.; Stribeck, N. “Nanostructure and mechanical properties studied during dynamical straining of microfibrillar reinforced HDPE/PA blends,” J. Polym. Sci. Part B: Polym. Phys., 2009, in print. 52. Vonk, C. G.; Kortleve, G. “X-ray small-angle scattering of bulk polyethylene,” Colloid Polym. Sci., 1967, 220, 19–24. 53. Strobl, G. R.; Schneider, M. “Direct evaluation of the electron density correlation function of partially crystalline polymers,” J. Polym. Sci., Part B: Polym. Phys., 1980, B18, 1343–1359. 54. Hall, E. L. Computer Image Processing and Recognition; Academic Press: London, 1980.

R Journal of Macromolecular Science , Part C: Polymer Reviews, 50:59–90, 2010 ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583720903503486

Ultra-Small-Angle X-ray Scattering of Polymers FAN ZHANG1,2 AND JAN ILAVSKY3 1

Materials Science and Engineering Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, USA 2 Department of Physics, Northern Illinois University, Dekalb, IL, USA 3 X-ray Science Division, Advanced Photon Source, Argonne National Laboratory, Argonne, IL, USA Ultra-small-angle X-ray scattering (USAXS) is capable of probing structural inhomogeneities in the size range of 1 to 1000 nm. Recent developments of X-ray sources and optics make USAXS increasingly relevant to polymer research. In this review, we examine the current technical state of USAXS instrumentation, and briefly introduce the method of data reduction and analysis. We emphasize USAXS’s application in areas such as polymer nanocomposites, polymer gels and solutions, polymer blends, polymer micelles and microemulsions, and colloidal sciences. Finally, we predict more USAXS studies on polymeric systems, especially those with large-scale structures or hierarchical microstructures. Keywords ultra-small-angle X-ray scattering, small angle X-ray scattering, polymers, colloids

Introduction Small-angle X-ray scattering (SAXS)1 is a nondestructive scattering technique that records elastic scattering of X-rays at scattering angles close to the direction of the incident beam. SAXS data contain information about important microstructural parameters such as the size, shape, volume, and total surface area of the scatterers, as well as characteristic distances if the scatterers are ordered or partially ordered. In a typical pinhole setup, SAXS is capable of delivering structural information between 1 and 100 nm. Much larger objects can be viewed directly with optical microscopes. The intermediate size range, however, is difficult for both microscope and SAXS, especially when the specimen is either highly absorbent or optically opaque. Ultra-small-angle X-ray scattering (USAXS) has been developed to extend the scattering q range (where its modulus q = (4π /λ)sinθ , λ is the wavelength of the X-ray, Received September 3, 2009; accepted November 22, 2009. The submitted manuscript has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (“Argonne”). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated under Contract No. DE-AC02-06CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the Government. Address correspondence to Jan Ilavsky, X-ray Science Division, Advanced Photon Source, Argonne National Laboratory, Argonne, IL, 60439, USA. E-mail: [email protected]

59

60

F. Zhang and J. Ilavsky

and 2θ is the scattering angle) of SAXS to 10−4 Å−1 and bridge the gap between SAXS and the optical microscope.2,3 USAXS features very high angular collimation in one or both collimating directions, which enables a small minimum scattering momentum that can be distinguished from the direct beam. This collimation is usually performed with single-crystal optics, first proposed by Compton nearly a century ago.4 The first experimental realization of crystal collimation was achieved in the early 1920s.5 This technique, however, suffered significant parasitic scattering from the surface scattering from the imperfect crystals and could hardly compete with slit- and curved-crystal collimation introduced by Guinier.1 In 1965, Bonse and Hart in their seminal work demonstrated that the parasitic scattering tails can be greatly suppressed with multiple reflections on channel-cut crystals.6 This result, together with the increasing availability of near-perfect Ge or Si crystals, makes double-crystal multiple-reflection collimation preferable when high angular collimation is desired. The Bonse-Hart configuration enables small-angle scattering which is commonly “behind the beamstop” in pinhole SAXS to be probed, and forms the basis of modern USAXS instruments. Many USAXS instruments have been constructed over the years using various X-ray sources, for example lab sources,7–11 second-generation synchrotron sources,8,12–14 and third-generation synchrotron sources.15–17 By making use of very long camera length (sample-to-detector distance), pinhole SAXS can also reach the low-end scattering angle offered by the Bonse-Hart configuration, albeit the total scattering range is comparatively limited.18,19 In a recent development, Cerbino et al.20 showed that analysis of near-field coherent speckles could provide scattering information in a range even lower than that afforded by Bonse-Hart cameras, though with limitations. In addition to a broad q range, USAXS also offers several distinct advantages such as high angular resolution (limited mainly by the width of the crystal rocking curve), wide dynamic range of intensity (∼8 to 9 orders of magnitude in instruments operated in thirdgeneration synchrotrons15,17), accurate energy tuning (E/E as small as 0.00015 which enables anomalous-USAXS15), and primary calibration of the X-ray scattering cross section.14 The advancement of X-ray sources, especially the availability of high-flux undulator beamlines in third-generation synchrotrons, further facilitates the quest for low photon counting statistics for samples with low scattering contrast, which makes USAXS a useful technique for obtaining bulk and ensemble-averaged structural information at the micron range and below for soft matter such as polymers, colloids, and biomaterials. Furthermore, USAXS is often unique for the study of concentrated, opaque systems or systems that are prone to structural damage during the preparation for microscopy measurements, which are common among soft materials. USAXS has been utilized in the study of polymers and other soft materials since the operation of the first USAXS instrument.21 However, when compared with pinhole SAXS, USAXS is often regarded as underutilized.22 This is largely due to the deficiencies of previous versions of USAXS instruments, where low X-ray flux and high parasitic scattering hinders the successful applications of USAXS in such systems. Recent technical improvements have resolved these obstacles and made USAXS an increasingly appealing technique for studies of soft matter in general and polymers in particular. This review will cover USAXS studies of polymers and colloids, especially those employing synchrotron radiation, to demonstrate its versatility and effectiveness. We will start with recent advancements in USAXS instrumentation and data processing, which make USAXS more relevant to soft matter research. We will emphasize USAXS applications

Ultra-Small-Angle X-ray Scattering of Polymers

61

in areas such as polymer nanocomposites, polymer gels and solutions, polymer blends, polymer micelles and microemulsions, and colloidal sciences. We hope that this review will act as a bridge to the existing instrumental capabilities and exciting sciences for which USAXS may be found suitable.

2. USAXS Instrumentation All Bonse-Hart USAXS instruments have schematic layouts similar to the one shown in Fig. 1. The main difference is the radiation source that is employed. Conventional X-ray sources, while easier to access, have disadvantages such as high beam divergence, low X-ray flux, and very limited availability of X-ray wavelength, which makes it difficult for USAXS to study specimens with low scattering contrast. Synchrotron sources, on the other hand, do not suffer these restrictions. It is especially true in the case of third-generation synchrotrons, where the X-ray energy from an undulator source is continuously tunable in a wide range (3.2 to 80 keV and above for Undulator A of the Advanced Photon Source23), X-ray flux at the sample position about 1013 photons s−1mm−1,15,16 and beam divergence about 10 microradian.24 These features contribute greatly to the sprouting of USAXS studies of polymeric systems in the past decade. After passing through beam defining slits and X-ray mirrors, the X-ray beam is collimated with multiple reflections in the collimating crystals. The angular divergence of the incident beam is fixed by the monochromator, and the angular width is decided by the full width at half maximum (FWHM) of the collimating crystal reflection curve, which is given by the dynamical diffraction theory. For a single reflection, the tails of the rocking curve roughly follow a (θ -θ B )−2 law, where θ is the diffraction angle away from the surface, and θ B is the Bragg angle for the collimating crystal. Multiple reflections between the crystal pairs, although having no effect on the FWHM of the reflection curve, greatly reduce the tail reflection following (θ -θ B )−2n, where n is the number of reflections. In this manner, a large number of reflections act to deliver an optimized signal-to-noise ratio in the scattering data. The original Bonse-Hart design employed triple and fivefold reflections on channelcut crystals,21,25 which has been adopted in many USAXS instruments.7,11,16 An alternative is to employ an even number of reflections in both the collimating and analyzing crystal pairs.8,14,15 This geometry does not change the propagation direction of the X-ray beam, and therefore makes alignment easier when energy tuning is required. We note that in recently built USAXS instruments pseudo-channelcut crystals consisting of two pieces of separate crystals are preferred. This design enables better polishing of the crystal surfaces, and therefore significantly reduces parasitic scattering from loose

Figure 1. Schematic of a Bonse-Hart USAXS instrument in the one-dimensional collimation mode.

62

F. Zhang and J. Ilavsky

or disoriented crystallites. Brief polishing and etching procedures can be found in Ilavsky et al.15 and Sztucki and Narayanan.17 The scattered beam, after being analyzed by the analyzing crystals, is recorded by the detector. USAXS employs a point detection system, such as an avalanche photodiode detector (APD),13,16 PIN photodiode,15,26 and scintillation counting detector.7,8 To ensure the reliability and reproducibility of USAXS data, the detector is required to be linear over the entire dynamic range of USAXS. For example, a low-cost, low-noise, high quantum efficiency PIN photodiode detector is shown to be linear over 10 decades of X-ray intensity.26 Several operational modes of USAXS are available. When the sample is isotropic and the X-ray flux is the bottleneck, one-dimensionally collimated (slit-smeared) USAXS is normally preferred. For an anisotropic sample, however, true scattering data cannot be recovered from slit-smearing data. An additional dimension of collimation and analyzing would be required along the direction orthogonal to the directions of both beam propagation and first collimation. This two-dimensionally collimated USAXS offers effective pinhole collimation, and is capable of recording scattered intensity as a function of both momentum transfer q and azimuthal angle χ .27 Collimation along vertical and horizontal directions, however, greatly reduces the photon flux on the detector. This geometry, therefore, is practical only with high flux sources, such as an undulator source in a third-generation synchrotron. Within the framework of one- and two-dimensionally collimated USAXS, anomalous USAXS and selected-area USAXS measurements can be made by taking advantage of the high energy resolution and high stability of both sample and high-resolution beam defining slits. Examples of these configurations can be found in Ilavsky et al.15 We mention that USAXS imaging, a size-sensitive imaging technique, can be operated in a setup very similar to that of one-dimensional collimated USAXS, except that the camera length is very small and imaging data are collected with a two-dimensional charge-coupled device (CCD) camera instead of a point detector.28 USAXS imaging has been used to investigate a wide range of systems, from polymer composites29 to plastically deformed metals.30 The imaging contrast mechanism of USAXS imaging has been explained based on wave propagation and dynamical diffraction theory. Zhang et al.31 found that refraction in the form of Porod scattering and, to a much less extent, reflection, fully account for USAXS imaging contrast. In the end of this section, we take advantage of a comparison of SAXS and USAXS scattering profiles of the same sample to illustrate the capability of USAXS, as shown in Fig. 2. The sample is a polydispersed silica suspension in water. The USAXS profile was collected at the USAXS beamline (32ID-B) at the Advanced Photon Source (APS), Argonne National Laboratory. The SAXS profile was collected with a desktop SAXS instrument. The SAXS intensity was absolutely calibrated to match the USAXS intensity. Figure 2 clearly shows that USAXS offers a broader q range and dynamic range of the intensity. In addition, the USAXS intensity is automatically calibrated (details in Section 3). The very low-q end of the USAXS scattering profile demonstrates the angular resolution of USAXS, which can be as high as 1 × 10−4 Å−1.

3. Data Reduction and Analysis Data reduction process for USAXS consists of several standard procedures such as absolute intensity calibration, desmearing (for one-dimensional collimated USAXS), and (if necessary) correction for multiple scattering.

Ultra-Small-Angle X-ray Scattering of Polymers

63

Figure 2. Comparison of USAXS and intensity-calibrated SAXS profiles of silica in water suspension.

For small-angle X-ray scattering, the scattering intensity I(q) is written as14 I (q) = 0 AtT ε

d d

(1)

where 0 is the incident X-ray flux in photons s−1 area−1, A is the illuminated sample area, t is the sample thickness, T is the sample transmission, ε is the converting efficiency of the detector, is the detector solid angle, and d (q)/d is the differential scattering cross section per unit volume per unit solid angle. It can be shown that absolute calibration of the scattering cross section only depends on the known quantities t, T, , and I0 .14 Therefore, a Bonse-Hart USAXS camera provides inherent absolute scattering intensity, which is essential for many important aspects of quantitative small-angle scattering analysis, such as obtaining the number density, volume fraction, and the specific surface area of the scatterers. Taking advantage of these properties, Zhang et al. established glassy carbon as an absolute intensity calibration standard, which is readily available for SAXS laboratories so as to incorporate absolute intensity analysis.32 For one-dimensional collimated USAXS, correction for slit smearing is required to recover the correct scattering cross section. The schematic of finite slit scattering is shown in Fig. 3. The vertical slit width 2w0 is decided by the FWHM of the crystal rocking curve. The horizontal slit length 2l0 is only restricted by the detector. The vertical slit-width 2w0 is comparable to the angular resolution of the instrument and is much smaller than 2l0 . Thus its correction is negligible. Consequently, the slit-smeared USAXS intensity is described

64

F. Zhang and J. Ilavsky

Figure 3. Schematic of finite slit scattering in one-dimensional collimated USAXS. 2θ is the scattering angle, 2l0 is the slit length, 2w0 is the slit width. Pl and Pw are the slit-length and slit-width weighing functions, respectively.

by ˜ 1 d (q) = d l0

0

l0

d 2 q + l 2 dl, d

(2)

˜ where d (q)/d is the measured, slit-smeared scattering intensity; d (q)/d is the unsmeared scattering cross section; and l is the reciprocal vector of integration in the slit direction. d (q)/d can be accurately calculated following an iterative algorithm developed by Lake.33 Because USAXS data are obtained in directions very close to that of the incident beam, multiple scattering can sometimes be significant. Multiple scattering is indicated by the apparent broadening of the rocking curve when the sample is in the beam. This effect can be corrected following a procedure described in Ilavsky et al.15 We note that multiple scattering is mostly negligible for samples with low scattering contrasts, such as polymer melts, and solutions. The reduced data are analyzed within the framework of small-angle scattering theory, which is outside the scope of this paper and will not be discussed in detail here. Interested readers are encouraged to refer to standard small-angle scattering references1,34,35 or related articles in this issue. We do note, however, that Irena, a tool suite that is developed for the support of the USAXS instrument at the Advanced Photon Source, brings together a variety of advanced small-angle scattering data modeling and evaluation tools and is in free circulation.36

4. Radiation Damage USAXS measurements making use of the Bonse-Hart configuration scan the reciprocal space point-by-point. Accordingly, samples are exposed to the X-ray beam much longer

Ultra-Small-Angle X-ray Scattering of Polymers

65

(∼10 min) than normally required with a pinhole camera. X-ray radiation damage to the sample therefore becomes a concern, especially for soft materials. It is well known that polymers degrade when exposed to intense X-ray radiation.37–41 Many chemical processes take place concurrently, each progressing at a different rate. For example, intense X-rays create photoelectric electrons, which are shown to be linearly dependent on the X-ray dose.42 For this effect, commonly known as primary radiation damage, an upper limit of the radiation damage was observed. In polymeric and biological materials, photoelectric electrons are capable of further damaging the specimen through the creation of free radicals, which lead to hydroperoxidation and consequent chain scission43 thus altering the polymeric microstructure. This process is normally referred to as secondary radiation damage. Despite the understanding of the chemical pathways, various aspects of X-ray radiation damage still have not been elucidated. Its prevention remains largely empirical. Practical mitigating approaches include monitoring a sample’s discoloration, repeating USAXS measurements to monitor the reproducibility of scattering profile as a function of time, and employing a flow-cell for polymer or colloidal suspensions.44 Extra attention needs to be paid to the prevention of radiation damage so as to ensure the quality of data.

5. Applications In this section, we present an overview of various USAXS applications in the studies of polymer gel and solution, colloidal dispersion, microemulsion, etc. The broad q range, high dynamic range, and very high angular resolution make USAXS a unique technique in these examples, which we will address. Here, we will not go into the details of individual analysis. Instead, we will devote the next section to the fundamentals and general principles of USAXS analysis.

5.1 Polymer Gels and Solutions Polymer gels form when polymers arrange into three-dimensional networks by virtue of covalent or noncovalent bonding in a second medium. Depending on the binding mechanism, polymer gels can be classified into chemical gels and physical gels. Polymer gels often exhibit hierarchical structure, with the largest dimensions on the length scale of or greater than 100 nm, which makes USAXS an appealing technique. Much effort has been conducted in the USAXS study of various polymer gels,45–66 such as hydrogels,45–49 organogels,51–54 xerogels,56, 57 and aerogels.57, 60–66 Hydrogels are composed of water-insoluble polymer chains and are highly absorbent of water. Hydrogels have a complicated structure on both the nano- and micro- scale. Due to their significant water content, hydrogels possess a degree of flexibility that resembles natural tissue. Agrawal et al.45 studied the nano- and micro-scale structures of biocompatible poly(L -lactic-acid)–poly(ethylene oxide)–poly(L -lactic acid) (PLLA-PEO-PLLA) gels. The USAXS spectra at low q were fit to a power law. The exponent monotonically increased with increasing length of crystalline PLLA blocks, which suggested that the internal structure of the hydrogel became denser during this process. Ando and Konishi46 studied the structure of transparent (VI-P) and translucent (VI-L) cellulose hydrogels prepared by coagulation and regeneration of viscose in acid solutions with and without acetone. By assuming a twophase model, the authors determined the volume fraction and average size of a high-density phase that only consists of cellulose in both types of gels.

66

F. Zhang and J. Ilavsky

When ambient conditions such as temperature, solvent quality, and electric field change, hydrogels are often subject to a first-order volume phase transition (VPT). Tirumala et al.48 studied temperature-induced VPT in neutral poly(N-isopropylacrylamide) (PIPAAm), poly(N,N-diethylacrylamide) (PDEAAm), and poly(N-isopropylmethacrylamide) (PIPMAm) hydrogels and their weakly charged counterparts prepared by copolymerizing with sodium methacrylate. USAXS results showed an abrupt VPT in both PIPAAm and PIPMAm gels, but a continuous one in PDEAAM gels. Furthermore, the VPTs were suppressed in poly(N-alkylacrylamide)s but not in PIPMAm with the addition of sodium methacrylate. The authors attributed these observed difference in VPTs to the hydrogen-bonding constraints on thermal fluctuations instead of relative hydrophobicity, as one would normally expect. Grigoriew et al.51–54 conducted a series of USAXS studies on organogels, a group of thermoreversible, viscoelastic materials formed from low-molecular-weight organic gelators. The authors found that the gel structure of monosaccharide is highly dependent on the gelator concentration.51,53 When the concentration increases, the sizes of the gelator aggregates decrease and their shapes change from disk-like to rod-like. Additionally, the sizes of the primary aggregates were found to depend strongly on the polarity of the gelators.54 Xerogel and aerogel form when the solvent is removed. Xerogels undergo unhindered shrinkage during drying, while aerogels are obtained under hypercritical conditions, which lead to no shrinkage. Both are highly porous materials and possess extremely high surface areas. USAXS is often utilized to understand the nature of the porosity in these materials. For example, a study of organosilioxane-aerosil aerogel provided evidence of two fractal structures, one built up by the organosilioxane, the other by the added silica soot.64 Schaefer et al.57 observed porosity in the nanometer range with a distinct feature of fractal in arylenebridged polysilsesquioxane xerogels and aerogels. The pore morphology shows a systematic dependence on the bridging group, but is only weakly dependent on the catalyst type, concentration, and the drying protocol. Pahl et al.66 showed that the average pore size and mass fractal dimension of resorcinol-formaldehyde (RF) aerogels change systematically as a function of resorcinol-to-catalyst ratio. Brandt and Fricke63 studied subcritically dried RF aerogels with very high catalyst concentrations. A wide range of pore size, from <1 nm to 200 nm, was reported. While SAXS is considered to be one of the best available techniques to measure the morphology of macromolecules,67 USAXS studies of polymer solutions have been few and far between. Compared with USANS, the X-ray scattering contrast is unfavorable. Compared with SAXS, USAXS features a size range that is large for individual macromolecules, and a measurement time that could easily incur radiation damage when precaution is not taken. Nonetheless, USAXS has been successfully utilized in a few cases.68–70 In one example, Li et al.69 observed a small-angle upturn in the USAXS spectra of sulfonated polystyrene ionomers and found that this upturn was not strongly dependent upon the counterions, compression-molding condition, temperature, and annealing. Instead of explaining the upturn using the Debye-Bueche model, the authors described the upturn with a power law and attributed its origin to polydispersity and irregularity when long-range order was absent. This analysis would not have been possible if not for the extended q range provided by USAXS. 5.2 Polymer Blends Polymers of different species are frequently blended together as a common industrial method to reduce cost and tune the property and functionality for a given application.

Ultra-Small-Angle X-ray Scattering of Polymers

67

In academia, polymer blends, as model systems rich in nonlinear, nonequilibrium phenomena, offer fascinating research opportunities. Depending on the miscibility of the polymeric components, polymer blends can form various structures, including microphase and macrophase domains in a phase separation process. It is therefore important to understand the morphological transitions in polymer melts and blends, and their implications on the structure-property relationship of the materials. USAXS, due to its q range, often proves useful to study the large structures in binary or ternary blends where one or more components crystallize. In a recent work, Michiewicz et al.71 studied binary blends of four different high-molecular-weight poly(styrene-b-isoprene) diblock copolymers with a lowmolecular-weight poly(styrene-b-isoprene-b-styrene) triblock copolymer and established a morphological phase diagram in the parameter space of molecular weight ratio R and blend composition. The values of R (14 < R < 43) probed in this study were significantly larger than the known ratios (R < 7). Similar to the case of 5 < R < 7, the authors found a large miscibility gap with macrophase separation into two distinct microphase-separated domains, as well as morphological transitions from lamellar to cylindrical structures. In addition, for R > 30, the blends also exhibit disordered bicontinuous and double-gyroid-like structures. The same research group also explored a ternary blend of a poly(styrene-b-isoprene) dilock copolymer, polystyrene, and polyisoprene solution-casted from cumene as a candidate of photonic crystals.72,73 The dried sample showed a well-defined peak in the reflectivity curve in the visible wavelength range (350–600 nm), which indicates a large structure formed in the blend. USAXS measurements for samples containing 20% and 40% homopolymers confirmed that the morphology of the blend was a highly ordered lamellar. The alternating layers of styrene and isoprene were found to possess a periodic dielectric structure, with their thicknesses 0.52 L and 0.48 L, respectively, where L is the lamellar thickness. Based on this structure, the band diagram was calculated, and pass- and stop-band were consequently obtained. To strengthen microphase separation and increase the inter-domain spacing of block copolymer melts, Tirumala et al.74 studied the effect of adding an additive poly(acrylic acid) homopolymer with a molar mass 1–13 times that of the poly(oxyethylene-oxypropyleneoxyethylene) copolymer on the Flory-Huggins segment-segment interaction parameter χ . While the neat copolymer was disordered, the addition of poly(acrylic acid) resulted in a well-ordered lamellar structure with an interdomain distance of 10 ± 1 nm. Powerlaw decay was consistently observed in the low-q region of slit-smeared USAXS profiles with exponent between 2 and 2.5, which is below the Porod exponent 3. This result showed that the droplet macrophase separation of homopolymers is not present in the blend regardless of its molar mass. Instead, the homopolymer was expected to uniformly distribute through the polyoxyethylene domains of the lamellae-forming blends. Kobayashi et al.75 combined light scattering (LS), USAXS, and SAXS and studied the self-assembly and morphology of l,3:2,4-bis-O-(p-methylbenzylidene)-D-sorbitol/ndibutylphthalate in the parameter space of temperature T and solute concentration PDTS in detail. In particular, USAXS was found useful for establishing the characteristic distances in the self-assembled states and the power-law slope in the gel states. Again, a mass fractal dimension was observed, which excluded the presence of a sharp interface. Combined with the previous example, this result suggests that USAXS is a powerful technique for investigating the degree of phase separation in the polymer blends.

68

F. Zhang and J. Ilavsky

5.3 Polymer Micelles and Microemulsions Surfactant molecules dispersed in selective solvents can self-assemble into various micellar morphologies such as sphere, ellipsoid, cylinder, and bilayer. Similarly, microemulsions form when two immiscible liquids are mixed together with at least one component being surface active. Polymer micelles and microemulsions often have sizes that make USAXS a suitable technique. In this section, we show a few recent USAXS applications for these systems. One example is the determination of the structure of casein micelles by Pignon et al.76 Casein micelles (calcium-protein complexes), milk fat globules, and milk sugar (lactose) are the three main biological components of milk. Despite being the subject of extensive investigation, the exact structural details of casein micelles have remained elusive.77 Using USAXS and unified analysis, as is shown in Fig. 4, the authors found two characteristic length scales for the equilibrium structure of casein, with radius of gyrations Rg about 100 and 5.6 nm pertaining to the globular micelles and their non-globular internal structure, respectively. The low-q region of the USAXS curve followed a q−4 power-law decay, which suggests the existence of a smooth interface of the large, globular micelles. The high-q region showed a decay following q−2, which was described by the Debye formula for entangled flexible polymer chains. This physical picture was in agreement with a recently proposed casein micelle model, which negated the existence of the submicellar structure within the micelle.78 In a subsequent work, Marchin et al.79 considered the environmental factors on the casein micelle structures. The casein micelles were fractionated into six fractions using a sequence of six consecutive centrifugation steps. In the low-q region, different Rgs were obtained, as expected. The globular structure of the micelles was confirmed in all cases. In the high-q region, the USAXS curves showed identical fractal-like features, which demonstrated that micelles of different fractions, albeit with different sizes, possessed the same internal structure. Furthermore, the authors studied the effects of temperature, pH, and calcium chelation on the structure of casein micelles. Thermo-treatments were found to have little influence on the structure as the USAXS curves overlapped. Reducing pH did not change the low-q region of the scattering curve, which suggested that the globular structure was independent of the acidity, at least in the range of pH probed. The change in the high-q

Figure 4. Static SAXS and USAXS measurements of casein micelle suspensions at 25◦ C and pH = 6.6. The continuous lines in the figure are obtained with unified fit.

Ultra-Small-Angle X-ray Scattering of Polymers

69

region confirmed that the internal structure of casein micelles was related to the presence of micellar calcium phosphate, which was disassociated at a lower pH value of 5.2. When the casein micelles were depleted of calcium by chelation, the USAXS curves completely changed over the entire q range and suggested the destruction of globular structure of the micelles. Di Cola et al.80 studied the morphologies of micelles formed by linear and cyclic poly(styrene-b-isoprene) (PS166 -b-PI278 ) copolymers in different solvents as a function of temperature. The micelles have one characteristic length scale of the order 100 nm, which makes USAXS suitable. For linear copolymers, the morphology of micelles formed in good solvents for PS blocks does not vary with temperature. A detailed analysis of USAXS spectra of linear copolymer in good solvents for PI blocks for temperatures above 60◦ C showed that the micelles undergo a cylinder-vesicle transition. For a cyclic copolymer, its morphology was found to be independent of both temperature and concentration. Zemb and coworkers utilized USAXS to study the structures of microemulsions.81,82 For example, Testard et al.81 examined the effect of apolar solute lindane on the microstructure of water-didodecyldimethylammoniumbromide (DDAB)-dodecane ternary microemulsion. The absolute intensity in the high-q region of USAXS curves showed a q−4 power-law dependency, demonstrating the presence of a sharp oil/water interface. The specific area measured at the Porod limit yielded the area per surfactant, which was shown to be irrelevant to the solute concentration. In addition, with increasing solute concentration, the characteristic size of the microemulsion cell was found to decrease. These structural results, together with conductivity measurements, were consistent with a disordered, open connected cylinders model. 5.4 Polymer Nanocomposites Modification of polymers with an inorganic or organic material as fillers or additives to produce polymer composites is one of the most successful examples of polymer engineering in the chemical industry. Polymer nanocomposites, in which the secondary, or in uncommon cases, ternary materials have dimensions of less than 100 nm, or structures with comparable repeating distances, offer exciting opportunities not possible with conventional polymer composite materials.83,84 Compared with macroscopic or microscopic composite fillers, the nanoscale fillers offer advantages such as ultra-high surface area per unit mass, ultra-low filler mass concentration to achieve filler percolation, and small interparticle distances in the polymer matrices. As a result, polymer nanocomposites are lighter than the conventional polymer composites and exhibit large increases in tensile modulus, strength, toughness, and heat resistance. Due to their outstanding properties, polymer nanocomposites have found many applications such as conducting polymers,85 optical materials,86 and fire retardants.87 The physical mechanism behind these applications, however, remains controversial and elusive, largely due to the complex, and often hierarchical structures of polymer nanocomposites. In polymer nanocomposites, the nanoscale fillers present an enormous amount of interfacial areas, which leads to direct interaction between polymers and the filler components. This effect introduces heterogeneity within the polymer matrix on the nanoscale and deforms the polymer from its bulk confirmation. The primary filler particles, with a length scale of tens of nanometers, form aggregates, which in turn could form agglomerates. Understanding this structural variation is crucial in terms of understanding various material behaviors of polymer nanocomposites. USAXS, because of its favorable q range, is an ideal tool for this investigation. Due to the hierarchical nature of the structure, unified analysis developed

70

F. Zhang and J. Ilavsky

by Beaucage88 is often the method of choice, which results in interpretation based on an analogy of surface or mass fractals.89,90 Narayanan et al.91 studied the structure of silica-polyvinyl acetate nanocomposites with matrices of two different molecular weights. They found that particle aggregation is independent of the molecular weight of the matrix for a fixed filler concentration and surface treatment. They also reported that various surface treatments reduce the bonding strength to the polymer matrix, as well as reduce the particle aggregation. This result was in agreement with a nonlinear viscoelastic study of the same nanocomposites above glass transition temperatures by Sternstein et al.92 Ehrburger-Dolle et al.93 studied the anisotropic structure of carbon black (CB)-filled, high-density polyethylene (HDPE) polymers. No anisotropy was observed when the CB concentration was below its percolation threshold. For concentrations above the threshold, CB aggregates with a fractal dimension less than 2 were found, and their role in the mechanical properties of CB filled nanocomposites was consistent with the model for reinforcement of rubber by fractal aggregates proposed by Witten et al.94 In a separate study of CB and polymethylmethacrylate (PMMA) nanocomposites, Levine et al.95 found that mechanically prepared samples exhibit very high electric conductivity at an extremely low CB loading percentage. USAXS measurements were used to explain this phenomenon as the formation of a three-dimensional CB-assembled nanowire network, with the mean diameter and length of the nanowire about 24 nm and up to 100 µm, respectively, which was further confirmed by USAXS imaging. Thill et al.96,97 studied spray-dried nanocomposites of silica and bromostyrene-styrene copolymer prepared by a one-step droplet drying process. The fine structure of the grains and the localization of the polymers were described. Instead of having empty pores in the dried composites, copolymers were found to fill half of the porous volume. Bellare et al.98–100 performed a series of studies on PMMA and barium sulfate nanocomposites as bone cement materials. To strengthen and provide a higher resistance to cracking, 100-nm barium sulfate particles were used instead of conventional 1–3 µm barium sulfate or zirconium oxide radiopacifiers. By comparing the analyzed total specific area with the theoretical value, they found that the nanoparticles were well dispersed in the PMMA matrix and the number of voids was limited. The stronger bonding resulting from this structure was confirmed by the increases in tensile strain-to-failure, tensile work-of-fracture, and fatigue life obtained through mechanical measurements. Chemin et al.101 studied the structure and mechanical properties of mesostructured functional hybrid coatings based on anisotropic nanoparticles dispersed in poly(hydroxylethyl methacrylate) (PHEMA). They reported that at low volume fraction, the anisotropic goethite nanorods and isotropic nanospheres form a homogeneous dispersion within the polymer matrix, which led to a strong reinforcement effect of PHEMA. Ikeda et al.102 reported an isotropic dispersion of silica nanoparticles in an in situ study of the morphological variation of silica/isoprene rubber nanocomposites. The scattering patterns change according to the stretching ratio in a manner similar to that of liquid crystalline elastomers under stress103 which indicated that the nanoparticles displaced affinely with the elastomer network. These results were further utilized as data input for reverse Monte Carlo simulation104,105 which showed that the structural change of the nanoparticles during elongation was minimal. Recenly, Kumar et al.106 observed anisotropic assembly of isotropic nanoparticles in a polymer nanocomposites system. Polystyrene chains were grafted onto the surface of spherical silica nanoparticles, the average diameter of which was about 14 nm. These

Ultra-Small-Angle X-ray Scattering of Polymers

71

Figure 5. USAXS profiles of silica grafted with different molecules mass (Mg ) of polystyrene brush. The continous lines are obtained with unified fit.

isotropic particles were implanted into isotropic polystyrene matrices of different molecular masses. USAXS, shown in Fig. 5, and TEM measurements found that after prolonged annealing, the isotropic grafted nanoparticles were capable of assembling into spherical aggregates, connected sheets, and strings, or stay well dispersed (unstructured) depending on the relative grafting density and grafted chain length. This extraordinary behavior was explained by the balance between the short-range interparticle attraction and the entropy loss of the distorted grafted polymers, and opens a new paradigm for the control of nanoparticle dispersion in a polymer matrix. Most notably, Schaefer and his coworkers68,107–126 have performed an extensive survey of various polymer nanocomposites using USAXS. The nanoscale fillers range from single-walled carbon nanotube68,124 (SWCNT) and multi-walled carbon nanotube118,123 (MWCNT) to precipitated silica.107,111,114,125,126 For these fillers, a typical small-angle scattering profile is shown in Fig. 6, where light scattering and USAXS data for Dimosil precipitated silica120 are combined to illustrate the levels of structures present in the sample. It is evident from the data that the primary particles (in this case, silica) form aggregates, which are further clustered into two types of agglomerates. Similar results were found in SWCNT124 and MWCNT118 dispersions as well, which directly showed that the nanoparticles, instead of being well dispersed, form aggregates and agglomerates. Considering that the presumption of nanoparticle reinforcement in nanocomposites is that the smaller nanoparticles offer more surface area and hence stronger bonding, these results pose a serious question for the overall effectiveness of polymer nanocomposites. For example, for the nanocomposite of MWCNT and melt-processed thermoplastic polyamide 6 (PA6), Zhao et al.117 found that the CNTs are quite flexible, regardless of the degree of chemical modification. Due to the flexibility of CNT, the yield strength of CNTs/PA6 nanocomposites is almost unchanged, and the tensile strength is only slightly increased. In a recent work, Schaefer et al.122 further showed that despite the formation of aggregates and agglomerates and the unremarkable performance in nanocomposites with hard matrices, carbon nanofibers are still capable of reinforcing soft materials (in this case, a

72

F. Zhang and J. Ilavsky

Figure 6. Combined light and USAXS scattering data for Dimosil precipated silica in the wet and dried states.

carbon nanofiber/thermoplastic polyurethane nanocomposite), although through a different mechanism. The fractal clusters formed by the nanofiber agglomerates are capable of storing elastic energy, which in turn explains the improvement in the mechanical strength. For a detailed and in-depth review of Schaefer’s work on nanocomposites, interested readers can refer to Schaefer et al.120 5.5 Colloidal Suspensions and Gels Colloidal suspensions and gels often have intrinsic sizes or characteristic distances befitting the q range of USAXS. The high-q resolution serves to obtain an accurate scattering structure factor, and the inherent absolute intensity calibration enables determination of number density of scattering particles. All these features are important to the understanding of basic statistical mechanical interactions and microstructures of such systems. Because of these reasons, USAXS has enjoyed great success in various colloidal systems, such as monocomponent dispersions,127–136 binary or ternary mixtures,137–139 colloidal gels,125,140–146 and colloidal crystals.147–151 In addition to the aforementioned advantages, USAXS allows precise measurement of the scattered intensity as q approaches 0. The normalized scattering intensity at q = 0 is related to the osmotic compressibility of the suspension. Furthermore, as previously mentioned, the structure factor of a colloidal suspension is often conveniently obtained in USAXS measurements. Therefore, USAXS profiles can be used to determine interparticle potentials or test models that govern the stability of colloidal suspension. For example, Tara et al.152 reported amorphous clustering of highly charged poly(chlorostyrene-styrene sulfonate) colloids in dilute deionized suspensions of various concentrations. In all cases, the structure factors showed a first peak and a split second peak, which indicated the formation of glasslike order despite the dilute nature of the dispersion. This research was

Ultra-Small-Angle X-ray Scattering of Polymers

73

among the first experimental evidence of long-range attraction in colloidal suspensions of low volume fraction. When the colloidal particles interact via an attractive, short-range potential, which is significantly greater than the thermal energy, a gas-liquid-type jamming transition can happen at a very low colloidal packing fraction. Pontoni et al.153 studied the static microstructure of charged silica colloidal particles in a binary mixture of 2,6 lutidine and heavy water near the fully reversible colloidal aggregation temperature TA . They found that at a low volume fraction (8% volume), the colloidal particles undergo repulsive to short-range attractive transition by a temperature dependent absorption process. Furthermore, the first maximum of the structure factor S(q) well above TA is below the Hansen-Verlet criterion 2.85, which suggests that the aggregated state has the dense, liquid-like packing feature of gas-liquid transition of colloids, instead of that of a freezing transition. In a similar, but more detailed study, Sztucki et al.154 investigated the static structure of a short-range interacting colloidal system over a large range of colloidal concentrations, including the vicinity of a reentrant glass-liquid-glass transition. The structure factor was modeled with an attractive square-well potential. The results, including the depth and range of the interparticle attraction and the structure of the aggregates, were found to be consistent with the predictions of a mode-coupling theory.155 The fine angular resolution and wide q range of USAXS make it an excellent technique for binary colloidal suspensions. Lutterbach et al.137 studied the static structure of the charge-stabilized polystyrene (PS) and perfluorinated (PFA) particles with diameters of 79 and 162 nm with a total volume fraction of 9%. PS and PFA have greatly different scattering length densities in water. Due to this property, the authors were able to extract the PFA-PFA partial structure factor and found the intensity of the first peak of this partial structure factor decreases with decreasing number fraction of PFA, which suggested the weakening of liquid-like order. The results were found to be in good agreement with predictions made by DLVO theory. USAXS is also applied in a binary suspension with large size ratio. Zhang et al.139 investigated the arrangement of highly charged zirconia nanoparticles (mean radius 2.57 nm) near the surface of negligibly charged silica microspheres (mean radius 280.11 nm), as shown in Fig. 7. The nanoparticles were shown to form a loose layer near the surface of the microspheres, with average nanoparticle-to-microsphere surface separation distance of 2 nm, which is nearly equivalent to the Debye length. The nanoparticle concentration in this layer is significantly higher than that in the bulk solution, and the average nanoparticle separation distance within the layer is about 9 times that of the nanoparticle radius. This result experimentally illustrated the static structure of a colloidal nanoparticle halo156 which has remained elusive since its discovery mainly because of the size discrepancy between the nanoparticles and microspheres. When the colloidal particles are anisotropic in shape, 2-D collimated USAXS is preferred over slit-smeared USAXS because of the complicated desmearing involved. Mock et al.157 used 2-D collimated USAXS to study the static microstructures of anisotropic polystyrene particles of various degree of anisotropy. The particle form factors of the anisotropic particles are represented by two interpenetrating spheres, with one sphere possessing a constant diameter and the other possessing a varying diameter. By changing the volume concentration, the colloids undergo a disorder-order transition. An example of anisotropic homonuclear colloids and corresponding USAXS profiles is shown in Fig. 8. Interestingly, the authors found that when the volume concentration is below 45%, the anisotropic colloids form a rotary or plastic crystal phase, where the centers of mass of the particles are ordered, but not the particle director. When the concentration is higher, the colloids form a body-center tetragonal structure, where both the center of mass and

74

F. Zhang and J. Ilavsky

Figure 7. USAXS measurements of monodisperse silica, zirconia, and the mixture of silica and zirconia. The continuous lines are least-squares fitting with user-defined mode in the framework of Irena SAS analysis package. Note the greatly different size (q) range of silica and zirconia.

Ultra-Small-Angle X-ray Scattering of Polymers

75

Figure 8. USAXS of anisotropic homonuclear colloids. The colloids, shown in Fig. 8(a), can be approximated with overlapping particles models obtained with Debye model for single particle scattering. Figure 8(b) shows its USAXS at 5% vol and its fit with Debye model.

the director are ordered. In a separate experiment, Levitz et al.158 studied the suspension of charged, disk-like synthetic clay Laponite/particles at very low ionic strength (in the order of 10−5M). By changing the concentration of the clay particles, a liquid-soft solid transition driven by electrostatic repulsive interaction was observed. The soft solid features a correlation peak at low-q. A detailed inspection of the peak showed that the clay particles form clusters, instead of being uniformly dispersed. In a rather unusual, gaseous colloidal suspension, Beaucage et al.159,160 studied the in situ nanoparticle nucleation and growth in flame aerosols at different positions inside the flame. The nanoparticles were found to form aggregates and agglomerates, the size and morphology of which were in turn mapped with the height above the burner and lateral distance from the axial center of the flame. The height above the burner was further converted to a kinetic time with the known gas flow rate. The dynamics of nanoparticle growth in flames was thus acquired. When attractive interaction exists between colloids, it is well known that the colloidal particles can aggregate to form colloidal gels which often have the characteristics of a fractal.161 For a mixture of hard-sphere colloids and nonadsorbing polymers, depletion attraction arises from the entropy-driven exclusion of the polymers from the region between closely neighbored colloids, where the range and strength of this effective attraction can be delicately tuned. Zukoski et al.141–146,162,163 performed a series of detailed and in-depth investigation of such concentrated depletion gels. In one example, Shah et al.141 studied the influence of polymer concentration and radius of gyration of the polymer on the microstructure of model hard-sphere nanocolloids, where the volume fraction of colloids was fixed at 0.40. For the given system where the radius of gyration was significantly smaller than the mean radius of the colloids, a direct homogeneous fluid to nonequilibrium gel transition was observed with increasing polymer concentration. The authors deducted the scattering structure factor S(q) by comparing the concentrated USAXS profile with corresponding dilute profile of the same particles and modeled S(q) with the polymer reference site interaction model (PRISM).164,165 The elastic modulus of the gel, yielded by

76

F. Zhang and J. Ilavsky

Figure 9. USAXS measurements of highly monodisperse silica beads. The insert shows a magnified curve near q = 3.2 × 10 −4 Å−1, where the oscillations are still clearly visible due to the very high angular resolution that USAXS provides.

this model, was found to depend on the polymer-colloid size ratio and the reduced polymer concentration in a power-law fashion. This study was extended further by Ramakrishnan et al.142 where the volume fraction dependence of the collective structure and the elastic modulus of colloidal gels of the same type was investigated. The elastic modulus extracted from PRISM was shown to have a power-law dependence on the volume fraction of the colloids, with an effective exponent decreasing with increasing depletion attraction strength. Despite the fact that the discussion in this section is confined to colloidal suspensions and gels, it is necessary to point out that USAXS is a valuable and often unique tool when investigating large particles with narrow size distribution in their ordered states due to the small angular resolution of USAXS. One example as shown in Fig. 9 is given in Ilavsky et al.,15 where USAXS profiles of monodisperse silica beads in their dilute and ordered states were compared. At q as large as 3 × 10−2 Å−1, the oscillations from the scattering form factor were still clearly visible. This feature facilitates the accurate and reliable extraction of the structure factor, and predictably, will also be useful to identify sharp diffraction peaks, if present. 5.6 Other Materials In addition to the examples shown in the previous sections, USAXS has been successfully utilized in other areas of polymeric and colloidal research, such as non-affine deformation of soft colloidal films,166 self-assembly of cellulose that is artificially synthesized via enzymatic polymerization,167 structure of dendrimers with univalent and bivalent counterions in ionic dilute solutions,168 forced assembly of two immiscible polymers produced by layer-multiplying extrusion,169 water-swollen perfluorinated ionomer membranes,170

Ultra-Small-Angle X-ray Scattering of Polymers

77

structure of hydroxypropyl cellulose as flavor-barrier membrane,171 morphology of dry lignins and size and shape of dissolved lignin particles,172 nanostructure evolution of polyethylene,173–175 the failure mechanism of glass-beads filled polystyrene,176,177 the domain-interface sharpening process during crystallization,178 and the degree of homogeneity of homogeneous dispersions.179 These examples, together with the aforementioned examples, explicitly demonstrate the wide-range applicability of USAXS. Recent technical improvements to the available instruments have made USAXS an increasingly attractive tool to study large scale and hierarchical structures in complex fluids. As a result, more effort has been made to employ the unique capability of USAXS to understand physical, chemical, or engineering properties of these materials. There is no doubt that USAXS, as a proven structure-characterization technique, will continue to tackle complex, yet important and interesting scientific problems in soft condensed matter.

6. State of Analysis for USAXS Data USAXS instruments, when compared with pinhole SAXS instruments, provide a wider range of scattering vector as well as a larger dynamic range of scattering intensity. The information content of USAXS data, therefore, is more complete. In principle, the results obtained from proper analyses of USAXS data ought to be more accurate and more representative of the microstructures studied. However, it sometimes seems that the exact opposite is true. Developing a model which would describe the microstructure over up to four decades in size, such as that in the case of USAXS study of binary colloidal mixture with great size disparity139 is not a simple feat; and many studies significantly fail in this regard. In this review we have reported the wide range of USAXS applications in polymers, while realizing that the quality of data analysis varies. So far, we have avoided pointing out specific flaws in data analysis to keep the review easier to follow. In this section we will identify the major challenges facing USAXS data analysis, and will offer some general suggestions regarding good practices in this area. We will also provide a few examples where analysis may have fallen short. While ab initio small angle scattering (SAS) analysis can still be seen in literature, it has become more popular to construct SAS analysis based on existing SAXS analysis packages. The availability of these packages not only makes the analysis more convenient, but also establishes a degree of confidence to the analysis, considering that the software packages, which are tested and employed by many users, are less error-prone. Some packages were developed on the basis of the Igor Pro commercial scientific software (Wavemetrics, 2008), for example, NIST analysis tools,180 Motofit,181 and Irena.36 Self-standing SAS packages developed with more common programming languages (Fortran, C, Java, Pascal, etc.) also exist, such as Scatter,182 SASFit,183 Fish,184 and ATSAS, a suite of software developed by Svergun et al.185–189 The exact functionalities of these packages vary, and there are usually specific areas where they excel. For example, ATSAS186 is the undisputed standard for bio-SAXS analysis. The wide availability of these packages may lead the users to enjoy the seemingly large freedom regarding curve fitting, and sometimes base their analysis on convenience, rather than facts. It is worth being aware that proper usage of any SAS analysis package or any SAS analysis method in general, stresses on the reliance of the users to be knowledgeable about the specific problem and capable of choosing an appropriate tool/method. This is more evident for USAXS analysis because most SAS analysis packages except Irena are not primarily developed to be used with USAXS data.

78

F. Zhang and J. Ilavsky

As stated above, the main challenge to USAXS data analysis lies in the fact that USAXS probes a wide range of scattering vector. Therefore it is necessary to design either one model to describe the microstructure over such a wide range of sizes—potentially from Angstrom to micron, or to find a valid approach to split the data into different segments, analyze each segment with an appropriate method, and then combine the results in a meaningful way to describe the microstructure. For example, Unified fit is developed for the latter case88,90 in which the microstructure is modeled as a sequence of hierarchical structural levels. These levels could be independent or could also possess complex relationships among themselves. One needs to have a fundamentally sound understanding of the sample to assign the true meaning of each structural level in terms of the underlying microstructure. Having clearly stated and defendable basis for the chosen analysis method is of essential importance for fitting of any small-angle scattering data. It is especially this case for USAXS data analysis, as its q range is extensive and various portions of data may need to be analyzed with different assumptions. It is also crucial to realize that different methods (packages) may have different foundations. For instance, ATSAS assumes that the scatterers are mono-sized and mono-shaped; volume size-distribution analysis assumes that the scatterers have the same shape but varying sizes, and fractal analysis assumes self-similarity of the microstructure over an extended length scale. A good example of proper USAXS analysis with several methods combined can be found in the investigation of the structure of nanocomposites by Schaefer and Justice,120 where scattering from primary particles was analyzed by Guinier analysis or size-distribution analysis, and the structure of large-scale aggregates and agglomerates were obtained with fractal analysis. We do note, however, that in some cases multiple analysis methods are used without befitting justification.51,54 One example is shown in Fig. 10, which is an attempt to fit USAXS data from apolar and polar gels by fractals analysis.54 The data do not aid this analysis by presenting clearly separated regions with different power-law dependence. In the same articles, pair distribution function and Guinier analysis are also employed, again without discussion about their applicability. Here we strengthen the misuse of fractal analysis when there is no clearly defined fractal region and/or the power law slope obtained strongly depends on the selection of the fitting interval. Another common issue in USAXS analysis is the inability to distinguish between highly asymmetric particles (long rods, high aspect-ratio disks, etc.) and fractals. It is commonly recognized that for these particles the region bound by two “Guinier type regions” features a power-law slope of -1 (randomly oriented rod) or -2 (disk or sheet-like object). But it is much less realized that it is also possible to have particles with these shapes to cause the scaling to have a nearly arbitrary power law component. This possibility is manifested in recent USAXS studies of aerogels.190,191 These materials are constructed by solid materials in various shapes (sheet-like or rod-like) and not fractals. Their USAXS profiles, however, exhibit varying power-law slopes. These results suggest that simple fractal analysis of USAXS data such as in the case of silica aerogels192 may require further discussion to eliminate other possible microstructures. USAXS analysis is often not unique. Without constraints, data fitting can be meaningless. It is therefore important to keep in mind that the more one knows, the more one can learn from USAXS, or SAS in general. We are glad that most USAXS applications so far show deep appreciation of these principles. We also recognize that improvement could also be made, and awareness of these principles should be steadily promoted especially as USAXS instrumentation has greatly matured and USAXS, as a technique, is ready to make greater contribution to polymer research.

Ultra-Small-Angle X-ray Scattering of Polymers

79

Figure 10. USAXS data of (a) apolar gels and (b) polar gels with fractal analysis.

7. Summary and Outlook Small-angle X-ray scattering is a very useful technique to study the size, shape, and structural inhomogeneities of polymers. USAXS, by taking advantage of a wide q range, acts to bridge pinhole SAXS and light scattering, and provides access to large-scale structures and hierarchical structures. Recent developments in high-flux sources and crystals with low parasitic surface scattering enhance the capability of USAXS to probe samples with low scattering contrast. Many applications of USAXS have been found in areas such as

80

F. Zhang and J. Ilavsky

polymer gels and solutions, polymer nanocomposites, polymer blends, polymer micelles and microemulsions, and colloidal sciences. We acknowledge that despite the success that USAXS enjoys, it remains largely a static scattering approach due to point-scanning detection in the reciprocal space. Efforts are currently under way to perform on-fly USAXS scans, which will reduce the scan time by an order of magnitude (from ∼10 min to ∼1 min). This will have two major implications—1) for polymers and biomaterials, radiation damage will become less of a concern; 2) probes of slow kinetics will be possible. Undoubtedly this improvement will open a window into future scientific discoveries. There is also a plan at APS to incorporate a short pinhole camera that covers the q range from 0.05–1 Å−1 with the existing Bonse-Hart setup to facilitate automatic measurement of a “complete” small-angle X-ray scattering profile. This combination not only greatly improves the q resolution of USAXS at high qs, but also allows probing from the molecular structure to the self-assembled microstructure, and will be especially useful to characterize structures with multiple length scales. USAXS, as a technique, has reached a stage where technical developments have made it ready to greatly contribute to research into polymers and other soft materials. USAXS has, and will continue to enjoy success in areas such as nanocomposites and colloidal sciences. It is also foreseeable that USAXS will be adopted as an ideal technique to study biomaterials with hierarchical structures and to answer the related open questions.

Acknowledgment Research at the Advanced Photon Source, Argonne National Laboratory is supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-06CH11357.

References 1. Guinier, A.; Fournet, G. Small-Angle Scattering of X-Rays: John Wiley & Sons: New York, 1955; p 268. 2. Allen, A. J.; Jemian, P. R.; Black, D. R.; Burdette, H. E.; Spal, R. D.; Krueger, S.; Long, G. G. “Ultra-small-angle x-ray-scattering to bridge the gap between visible-light scattering and standard small-angle scattering cameras,” Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment, 1994, 347, 487–490. 3. Allen, A. J. “Characterization of ceramics by x-ray and neutron small-angle scattering,” J Am Ceram Soc, 2005, 88, 1367–1381. 4. Compton, A. H. “The intensity of x-ray reflection, and the distribution of the electrons in atoms,” Physical Review, 1917, 9, 29. 5. Davis, B.; Stempel, W. M. “An experimental study of the reflection of x-rays from calcite,” Physical Review, 1921, 17, 608. 6. Bonse, U.; Hart, M. “Tailless x-ray single-crystal reflection curves obtained by multiple reflection,” Appl Phys Lett, 1965, 7, 238–240. 7. Lambard, J.; Zemb, T. “A triple-axis bonse hart camera used for high-resolution small-angle scattering,” Journal of Applied Crystallography, 1991, 24, 555–561. 8. Chu, B.; Li, Y.; Gao, T. “A bonse-hart ultrasmall angle x-ray scattering instrument employing synchrotron and conventional x-ray sources,” Rev Sci Instrum, 1992, 63, 4128–4133. 9. Lambard, J.; Lesieur, P.; Zemb, T. “A triple axis double crystal multiple reflection camera for ultra small-angle x-ray-scattering,” Journal De Physique I, 1992, 2, 1191–1213. 10. Rieker, T. P.; Hubbard, P. F. “The university of new mexico/sandia national laboratories smallangle scattering laboratory,” Review of Scientific Instruments, 1998, 69, 3504–3509.

Ultra-Small-Angle X-ray Scattering of Polymers

81

11. Matsuoka, H.; Nakatani, Y.; Yamakawa, M.; Ise, N., “Ultra-small-angle x-ray-scattering study of the structure of colloidal and polymeric systems,” Journal De Physique Iv, 1993, 3, 451–454. 12. Chu, B.; Yeh, F.; Li, Y.; Harney, P. J.; Rousseau, J.; Darovsky, A.; Siddons, D. P. “Construction of a high-energy bonse–hart ultrasmall-angle x-ray scattering instrument,” Review of Scientific Instruments, 1994, 65, 3233–3237. 13. Smolsky, I. L.; Liu, P.; Niebuhr, M.; Ito, K.; Weiss, T. M.; Tsuruta, H., “Biological small-angle x-ray scattering facility at the stanford synchrotron radiation laboratory,” Journal of Applied Crystallography, 2007, 40, S453–S458. 14. Long, G. G.; Jemian, P. R.; Weertman, J. R.; Black, D. R.; Burdette, H. E.; Spal, R. “Highresolution small-angle x-ray-scattering camera for anomalous scattering,” J Appl Crystallogr, 1991, 24, 30–37. 15. Ilavsky, J.; Jemian, P. R.; Allen, A. J.; Zhang, F.; Levine, L. E.; Long, G. G. “Ultra-small-angle x-ray scattering at the advanced photon source,” Journal of Applied Crystallography, 2009, 42, 469–479. 16. Narayanan, T.; Diat, O.; Bosecke, P. “Saxs and usaxs on the high brilliance beamline at the esrf,” Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment, 2001, 467, 1005–1009. 17. Sztucki, M.; Narayanan, T. “Development of an ultra-small-angle x-ray scattering instrument for probing the microstructure and the dynamics of soft matter,” Journal of Applied Crystallography, 2007, 40, S459–S462. 18. Shinohara, Y.; Kishimoto, H.; Inoue, K.; Suzuki, Y.; Takeuchi, A.; Uesugi, K.; Yagi, N.; Muraoka, K.; Mizoguchi, T.; Amemiya, Y. “Characterization of two-dimensional ultra-small-angle x-ray scattering apparatus for application to rubber filled with spherical silica under elongation,” Journal of Applied Crystallography, 2007, 40, S397–S401. 19. Krosigk, G. V.; Cunis, S.; Gehrke, R.; Kranold, R. “Recent developments and asaxs measurements at the ultra small angle x-ray scattering instrument of hasylab,” Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment, 2001, 467, 1088–1091. 20. Cerbino, R.; Peverini, L.; Potenza, M. A. C.; Robert, A.; Bosecke, P.; Giglio, M. “X-rayscattering information obtained from near-field speckle,” Nature Physics, 2008, 4, 238– 243. 21. Bonse, U.; Hart, M. “Small angle x-ray scattering by spherical particles of polystyrene and polyvinyltoluene,” Z fur Physik, 1966, 189, 151–162. 22. Bhatia, S. R. “Ultra-small-angle scattering studies of complex fluids,” Current Opinion in Colloid & Interface Science, 2005, 9, 404–411. 23. Lai, B.; Khounsary, A.; Savoy, R.; Moog, L.; Gluskin, E. Undulator a characteristics and specifications; ANL/APS/TB-3; Advanced Photon Source, Argonne National Laboratory, 1993. 24. Yang, B. X.; Lumpkin, A. H. In Visualizing electron beam dynamics and instabilities with synchrotron radiation at the aps, 21st Particle Accelerator Conference (PAC), Knoxville, TN, May 16–20, 2005; Knoxville, TN, 2005; pp 2603–2607. 25. Bonse, U.; Hart, M. “Tailless x-ray single-crystal reflection curves obtained by multiple reflection,” Applied Physics Letters, 1965, 7, 238–240. 26. Jemian, P. R.; Long, G. G. “Silicon photodiode detector for small-angle x-ray scattering,” J Appl Cryst, 1990, 23, 430–432. 27. Ilavsky, J.; Allen, A. J.; Long, G. G.; Jemian, P. R. “Effective pinhole-collimated ultrasmallangle x-ray scattering instrument for measuring anisotropic microstructures,” Review of Scientific Instruments, 2002, 73, 1660–1662. 28. Levine, L. E.; Long, G. G. “X-ray imaging with ultra-small-angle x-ray scattering as a contrast mechanism,” Journal of Applied Crystallography, 2004, 37, 757–765. 29. Levine, L. E.; Long, G. G.; Ilavsky, J.; Gerhardt, R. A.; Ou, R.; Parker, C. A. “Self-assembly of carbon black into nanowires that form a conductive three dimensional micronetwork,” Applied Physics Letters, 2007, 90, 014101.

82

F. Zhang and J. Ilavsky

30. Long, G. G.; Levine, L. E.; Fields, R. J. “X-ray scattering and imaging from plastically deformed metals,” Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing, 2001, 309, 28–31. 31. Zhang, F.; Long, G. G.; Levine, L. E.; Ilavsky, J.; Jemian, P. R. “Quantitative characterization of the contrast mechanisms of ultra-small-angle x-ray scattering imaging,” Journal of Applied Crystallography, 2008, 41, 416–427. 32. Zhang, F.; Ilavsky, J.; Long, G.; Quintana, J.; Allen, A.; Jemian, P., “Glassy carbon as an absolute intensity calibration standard for small-angle scattering,” Metallurgical and Materials Transactions A. DOI: 10.1007/s11661-009-9950-x. 33. Lake, J. A. “An iterative method of slit-correcting small-angle x-ray data,” Acta Cryst, 1967, 23, 191–194. 34. Glatter, O.; Kratky, O. Small-Angle X-Ray Scattering; Academic Press: London, 1982; p 515. 35. Lindner, P.; Zemb, T. In Neutron, X-Ray and Light Scattering: Introduction to An Investigative Tool for Colloidal and Polymeric Systems, European Workshop on Neutron, X-ray and Light Scattering as an Investigative Tool for Colloidal and Polymeric Systems, Bombannes, France, 1991, 1990; Lindner, P.; Zemb, T., Eds. North-Holland, Amsterdam: Bombannes, France, 1990; p 375. 36. Ilavsky, J.; Jemian, P. R. “Irena: Tool suite for modeling and analysis of small-angle scattering,” Journal of Applied Crystallography, 2009, 42, 347–353. 37. Kusy, R. P.; Greenberg, A. R. “Determination of radiation-damage in isotactic poly(methyl methacrylate) by x-ray-diffraction,” Macromolecules, 1978, 11, 1051–1053. 38. Kusy, R. P.; Greenberg, A. R. “X-ray-diffraction technique to assess radiation-damage in semicrystalline polymers .1. Isotactic poly(methyl methacrylate),” Journal of Polymer Science Part a-Polymer Chemistry, 1979, 17, 3969–3978. 39. Rightor, E. G.; Hitchcock, A. P.; Ade, H.; Leapman, R. D.; Urquhart, S. G.; Smith, A. P.; Mitchell, G.; Fischer, D.; Shin, H. J.; Warwick, T. “Spectromicroscopy of poly(ethylene terephthalate): Comparison of spectra and radiation damage rates in x-ray absorption and electron energy loss,” Journal of Physical Chemistry B, 1997, 101, 1950–1960. 40. Kuwamoto, S.; Akiyama, S.; Fujisawa, T. “Radiation damage to a protein solution, detected by synchrotron x-ray small-angle scattering: Dose-related considerations and suppression by cryoprotectants,” Journal of Synchrotron Radiation, 2004, 11, 462–468. 41. McGeehan, J. E.; Carpentier, P.; Royant, A.; Bourgeois, D.; Ravelli, R. B. G. “X-ray radiationinduced damage in DNA monitored by online Raman,” Journal of Synchrotron Radiation, 2007, 14, 99–108. 42. Teng, T. Y.; Moffat, K. “Primary radiation damage of protein crystals by an intense synchrotron x-ray beam,” Journal of Synchrotron Radiation, 2000, 7, 313–317. 43. Shlyapnikov, Y. A.; Kriryushkin, S. G.; Mar’in, A. P. Antioxidative Stabilization of Polymers: CRC Press: Boca Raton, FL 1996. 44. Allen, A. J.; Hackley, V. A.; Jemian, P. R.; Ilavsky, J.; Raitano, J. M.; Chan, S. W. “In situ ultra-small-angle x-ray scattering study of the solution-mediated formation and growth of nanocrystalline ceria,” Journal of Applied Crystallography, 2008, 41, 918–929. 45. Agrawal, S. K.; Sanabria-DeLong, N.; Jemian, P. R.; Tew, G. N.; Bhatia, S. R. “Microto nanoscale structure of biocompatible pla-peo-pla hydrogels,” Langmuir, 2007, 23, 5039– 5044. 46. Ando, H.; Konishi, T. “Structure analysis of regenerated cellulose hydrogels by small-angle and ultra-small-angle x-ray scattering,” Physical Review E, 2000, 62, 727–733. 47. Kang, J. M.; Beers, K. J. “Synthesis and characterization of pcl-b-peo-b-pcl-based nanostructured and porous hydrogels,” Biomacromolecules, 2006, 7, 453–458. 48. Tirumala, V. R.; Ilavsky, J.; Ilavsky, M. “Effect of chemical structure on the volume-phase transition in neutral and weakly charged poly(n-alkyl(meth)acrylamide) hydrogels studied by ultrasmall-angle x-ray scattering,” Journal of Chemical Physics, 2006, 124, 234911.

Ultra-Small-Angle X-ray Scattering of Polymers

83

49. Ando, H.; Konishi, T.; Ise, N.; Saga, H.; Saito, H. “Structural analysis of cellulose hydrogel by small-angle and ultra-small-angle x-ray scattering”, In Statistical Physics, Tokuyama, M.; Stanley, H. E., Eds. 2000; Vol. 519, pp 205–207. 50. de Moor, P.; Beelen, T. P. M.; van Santen, R. A.; Tsuji, K.; Davis, M. E. “Saxs and usaxs investigation on nanometer-scaled precursors in organic-mediated zeolite crystallization from gelating systems,” Chemistry of Materials, 1999, 11, 36–43. 51. Grigoriew, H.; Gronkowski, J. “Usaxs studies of monosaccharide gels. Ii. The common features of structural changes,” Journal of Non-Crystalline Solids, 2006, 352, 5492–5497. 52. Grigoriew, H.; Luboradzki, R.; Cunis, S. “In situ studies of monosaccharide gelation using the small-angle x-ray scattering time-resolved method,” Langmuir, 2004, 20, 7374–7377. 53. Grigoriew, H.; Luboradzki, R.; Gronkowski, J. “Usaxs studies of monosaccharide gels. I. Dependence of the glucofuranose-based gel structure on the gelator concentration,” Journal of Non-Crystalline Solids, 2006, 352, 3052–3057. 54. Grigoriew, H.; Temeriusz, A.; Chmielewska, D.; Gronkowski, J.; Mirkowska, M. “Usaxs study of influence of gelator polarity on organogel structure,” Journal of Sol-Gel Science and Technology, 2007, 44, 249–254. 55. Klausener, P.; Woermann, D. “Structure and transport properties of cation exchange gel membranes: Facilitated transport of ethene with silver ions as carriers,” Journal of Membrane Science, 2000, 168, 17–27. 56. Boury, B.; Corriu, R. J. P.; Delord, P.; Le Strat, V. “Structure of silica-based organic-inorganic hybrid xerogel,” Journal of Non-Crystalline Solids, 2000, 265, 41–50. 57. Schaefer, D. W.; Beaucage, G.; Loy, D. A.; Shea, K. J.; Lin, J. S. “Structure of arylenebridged polysilsesquioxane xerogels and aerogels,” Chemistry of Materials, 2004, 16, 1402– 1410. 58. Skillas, G.; Agashe, N.; Kohls, D. J.; Ilavsky, J.; Jemian, P.; Clapp, L.; Schwartz, R. J.; Beaucage, G. “Relation of the fractal structure of organic pigments to their performance,” Journal of Applied Physics, 2002, 91, 6120–6124. 59. Takenaka, M.; Kobayashi, T.; Saijo, K.; Tanaka, H.; Iwase, N.; Hashimoto, T.; Takahashi, M. “Comparison in fractal dimension between those obtained from structure factor and viscoelasticity of gel networks of 1,3 : 2,4-bis-o-(p-methylbenzylidene)-d-sorbitol in polystyrene melt at gel point,” Journal of Chemical Physics, 2004, 121, 3323–3328. 60. Berthon, S.; Barbieri, O.; Ehrburger-Dolle, F.; Geissler, E.; Achard, P.; Bley, F.; Hecht, A. M.; Livet, F.; Pajonk, G. M.; Pinto, N.; Rigaci, A.; Rochas, C. “Dls and saxs investigations of organic gels and aerogels,” Journal of Non-Crystalline Solids, 2001, 285, 154–161. 61. Brandt, R.; Petricevic, R.; Probstle, H.; Fricke, J. “Acetic acid catalyzed carbon aerogels,” Journal of Porous Materials, 2003, 10, 171–178. 62. Berthon-Fabry, S.; Langohr, D.; Achard, P.; Charrier, D.; Djurado, D.; Ehrburger-Dolle, F. “Anisotropic high-surface-area carbon aerogels,” Journal of Non-Crystalline Solids, 2004, 350, 136–144. 63. Brandt, R.; Fricke, J. “Acetic-acid-catalyzed and subcritically dried carbon aerogels with a nanometer-sized structure and a wide density range,” Journal of Non-Crystalline Solids, 2004, 350, 131–135. 64. Marliere, C.; Woignier, T.; Dieudonne, P.; Primera, J.; Lamy, M.; Phalippou, J. “Two fractal structures in aerogel,” Journal of Non-Crystalline Solids, 2001, 285, 175–180. 65. Woignier, T.; Lamy, M.; Marliere, C.; Dieudonne, P.; Primera, J.; Beurroies, I.; Phalippou, J. “Large scale porous structure in gels and relationship with their plastic behavior,” Journal of Sol-Gel Science and Technology, 2003, 26, 143–147. 66. Pahl, R.; Bonse, U.; Pekala, R. W.; Kinney, J. H. “Saxs investigations on organic aerogels,” Journal of Applied Crystallography, 1991, 24, 771–776. 67. Chu, B.; Hsiao, B. S. “Small-angle x-ray scattering of polymers,” Chemical Reviews, 2001, 101, 1727–1761.

84

F. Zhang and J. Ilavsky

68. Brown, J. M.; Anderson, D. P.; Justice, R. S.; Lafdi, K.; Belfor, M.; Strong, K. L.; Schaefer, D. W., “Hierarchical morphology of carbon single-walled nanotubes during sonication in an aliphatic diamine,” Polymer, 2005, 46, 10854–10865. 69. Li, Y. J.; Peiffer, D. G.; Chu, B. “Long-range inhomogeneities in sulfonated polystyrene ionomers,” Macromolecules, 1993, 26, 4006–4012. 70. Rzayev, J. “Synthesis of polystyrene—polylactide bottlebrush block copolymers and their melt self-assembly into large domain nanostructures,” Macromolecules, 2009, 42, 2135–2141. 71. Mickiewicz, R. A.; Ntoukas, E.; Avgeropoulos, A.; Thomas, E. L. “Phase behavior of binary blends of high molecular weight diblock copolymers with a low molecular weight triblock,” Macromolecules, 2008, 41, 5785–5792. 72. Urbas, A.; Sharp, R.; Fink, Y.; Thomas, E. L.; Xenidou, M.; Fetters, L. J. “Tunable block copolymer/homopolymer photonic crystals,” Advanced Materials, 2000, 12, 812–814. 73. Edrington, A. C.; Urbas, A. M.; DeRege, P.; Chen, C. X.; Swager, T. M.; Hadjichristidis, N.; Xenidou, M.; Fetters, L. J.; Joannopoulos, J. D.; Fink, Y.; Thomas, E. L. “Polymer-based photonic crystals,” Advanced Materials, 2001, 13, 421–425. 74. Tirumala, V. R.; Daga, V.; Bosse, A. W.; Romang, A.; Ilavsky, J.; Lin, E. K.; Watkins, J. J. “Wellordered polymer melts with 5 nm lamellar domains from blends of a disordered block copolymer and a selectively associating homopolymer of low or high molar mass,” Macromolecules, 2008, 41, 7978–7985. 75. Kobayashi, T.; Takenaka, M.; Saijo, K.; Hashimoto, T., “Self-assembly and morphology of gel networks in 1,3 : 2,4-bis-o-(p-methylbenzylidene)-d-sorbitol/n-dibutylphthalate,” Journal of Colloid and Interface Science, 2003, 262, 456–465. 76. Pignon, F.; Belina, G.; Narayanan, T.; Paubel, X.; Magnin, A.; Gesan-Guiziou, G. “Structure and rheological behavior of casein micelle suspensions during ultrafiltration process,” Journal of Chemical Physics, 2004, 121, 8138–8146. 77. Farrell, H. M.; Malin, E. L.; Brown, E. M.; Qi, P. X. “Casein micelle structure: What can be learned from milk synthesis and structural biology?” Current Opinion in Colloid & Interface Science, 2006, 11, 135–147. 78. Holt, C. “Structure and stability of bovine casein micelles,” Advances in Protein Chemistry, 1992, 43, 63–151. 79. Marchin, S.; Putaux, J. L.; Pignon, F.; Leonil, J. “Effects of the environmental factors on the casein micelle structure studied by cryo transmission electron microscopy and small-angle x-ray scattering/ultrasmall-angle x-ray scattering,” Journal of Chemical Physics, 2007, 126, 045101. 80. Di Cola, E.; Lefebvre, C.; Deffieux, A.; Narayanan, T.; Borsali, R. “Micellar transformations of poly(styrene-b-isoprene) block copolymers in selective solvents,” Soft Matter, 2009, 5, 1081–1090. 81. Testard, F.; Zemb, T. “Solute effect on connectivity of water-in-oil microemulsions,” Langmuir, 2000, 16, 332–339. 82. Barnes, I. S.; Hyde, S. T.; Ninham, B. W.; Derian, P. J.; Drifford, M.; Zemb, T. N. “Small-angle x-ray-scattering from ternary microemulsions determines microstructure,” Journal of Physical Chemistry, 1988, 92, 2286–2293. 83. Giannelis, E. P. “Polymer layered silicate nanocomposites,” Advanced Materials, 1996, 8, 29–35. 84. Vaia, R. A.; Giannelis, E. P. “Polymer nanocomposites: Status and opportunities,” Mrs Bulletin, 2001, 26, 394–401. 85. Gangopadhyay, R.; De, A. “Conducting polymer nanocomposites: A brief overview,” Chemistry of Materials, 2000, 12, 608–622. 86. Beecroft, L. L.; Ober, C. K. “Nanocomposite materials for optical applications,” Chemistry of Materials, 1997, 9, 1302–1317. 87. Porter, D.; Metcalfe, E.; Thomas, M. J. K. “Nanocomposite fire retardants—a review,” Fire and Materials, 2000, 24, 45–52. 88. Beaucage, G. “Approximations leading to a unified exponential power-law approach to smallangle scattering,” J Appl Crystallogr, 1995, 28, 717–728.

Ultra-Small-Angle X-ray Scattering of Polymers

85

89. Schaefer, D. W. “Polymers, fractals, and ceramic materials,” Science, 1989, 243, 1023–1027. 90. Beaucage, G. “Small-angle scattering from polymeric mass fractals of arbitrary mass-fractal dimension,” Journal of Applied Crystallography, 1996, 29, 134–146. 91. Narayanan, R. A.; Thiyagarajan, P.; Zhu, A. J.; Ash, J.; Shofner, M. L.; Schadler, L. S.; Kumar, S. K.; Sternstein, S. S. “Nanostructural features in silica-polyvinyl acetate nanocomposites characterized by small-angle scattering,” Polymer, 2007, 48, 5734–5741. 92. Sternstein, S. S.; Zhu, A. J. “Reinforcement mechanism of nanofilled polymer melts as elucidated by nonlinear viscoelastic behavior,” Macromolecules, 2002, 35, 7262–7273. 93. Ehrburger-Dolle, F.; Hindermann-Bischoff, M.; Livet, F.; Bley, F.; Rochas, C.; Geissler, E. “Anisotropic ultra-small-angle x-ray scattering in carbon black filled polymers,” Langmuir, 2001, 17, 329–334. 94. Witten, T. A.; Rubinstein, M.; Colby, R. H. “Reinforcement of rubber by fractal aggregates,” Journal De Physique Ii, 1993, 3, 367–383. 95. Levine, L. E.; Long, G. G.; Ilavsky, J.; Gerhardt, R. A.; Ou, R.; Parker, C. A., “Self-assembly of carbon black into nanowires that form a conductive three dimensional micronetwork,” Applied Physics Letters, 2007, 90, 014101. 96. Thill, A.; Spalla, O. “Influence of templating latex on spray dried nanocomposite powders studied by small angle scattering,” Journal of Colloid and Interface Science, 2005, 291, 477–488. 97. Sen, D.; Spalla, O.; Thill, A. In Nanocomposite Powders by Evaporation Driven Self Assembly during Spray Drying: Some New Insights on the Structure Modifications, Symposium on Doped Nanopowders held at the 2006 Fall E-MRS Meeting, Warsaw, POLAND, Sep 04–08, 2006; Lojkowski, W.; Blizzard, J. R.; Narkiewicz, U.; Fidelus, J. D., Eds. Warsaw, Poland, 2006; pp 73–80. 98. Gomoll, A. H.; Bellare, A.; Fitz, W.; Thornhill, T. S.; Scott, R. D.; Jemian, P. R.; Long, G. G. In A Nano-Composite Poly(Methyl-Methacrylate) Bone Cement, Symposium on Nanophase and Nanocomposite Materials III held at the 1999 MRS Fall Meeting, Boston, Ma, Nov 29-Dec 02, 1999; Komarneni, S.; Parker, J. C.; Hahn, H., Eds. Boston, Ma, 1999; pp 399–404. 99. Bellare, A.; Gomoll, A. H.; Fitz, W.; Scott, R. D.; Thornhill, T. S. In Improving the Fatigue Properties of Poly(Methyl Methacrylate) Orthopaedic Cement containing Radiopacifier Nanoparticles, 4th International Conference on Processing and Manufacturing of Advanced Materials, Madrid, Spain, Jul 07–11, 2003; Chandra, T.; Torralba, J. M.; Sakai, T., Eds. Madrid, Spain, 2003; pp 3133–3138. 100. Gomoll, A. H.; Fitz, W.; Scott, R. D.; Thornhill, T. S.; Bellare, A. “Nanoparticulate fillers improve the mechanical strength of bone cement,” Acta Orthopaedica, 2008, 79, 421–427. 101. Chemin, N.; Rozes, L.; Chaneac, C.; Cassaignon, S.; Bourhis, E. L.; Jolivet, J.-P.; Spalla, O.; Barthel, E.; Sanchez, C. “Structure and mechanical properties of mesostructured functional hybrid coatings based on anisotropic nanoparticles dispersed in poly(hydroxylethyl methacrylate),” Chemistry of Materials, 2008, 20, 4602–4611. 102. Ikeda, Y.; Yasuda, Y.; Yamamoto, S.; Morita, Y. “Study on two-dimensional small-angle xray scattering of in situ silica filled nanocomposite elastomer during deformation,” Journal of Applied Crystallography, 2007, 40, S549–S552. 103. Zhang, F.; Heiney, P. A.; Srinivasan, A.; Naciri, J.; Ratna, B., “Structure of nematic liquid crystalline elastomers under uniaxial deformation,” Physical Review E, 2006, 73, 021701. 104. Hagita, K.; Arai, T.; Kishimoto, H.; Umesaki, N.; Shinohara, Y.; Amemiya, Y. “Two-dimensional pattern reverse monte carlo method for modelling the structures of nano-particles in uniaxial elongated rubbers,” Journal of Physics-Condensed Matter, 2007, 19, 230017. 105. Hagita, K.; Arai, T.; Kishimoto, H.; Umesaki, N.; Suno, H.; Shinohara, Y.; Amemiya, Y. “Structural changes of silica particles in elongated rubber by two-dimensional small-angle xray scattering and extended reverse monte carlo analysis,” Rheologica Acta, 2008, 47, 537–541. 106. Akcora, P.; Liu, H.; Kumar, S. K.; Moll, J.; Li, Y.; Benicewicz, B. C.; Schadler, L. S.; Acehan, D.; Panagiotopoulos, A. Z.; Pryamitsyn, V.; Ganesan, V.; Ilavsky, J.; Thiyagarajan, P.; Colby, R. H.; Douglas, J. F. “Anisotropic self-assembly of spherical polymer-grafted nanoparticles,” Nat Mater, 2009, 8, 354–359.

86

F. Zhang and J. Ilavsky

107. Schaefer, D. W.; Chen, C. Y. “Structure optimization in colloidal reinforcing fillers: Precipitated silica,” Rubber Chemistry and Technology, 2002, 75, 773–793. 108. Schaefer, D. W.; Suryawanshi, C.; Pakdel, P.; Ilavsky, J.; Jemian, P. R., “Challenges and opportunities in complex materials: Silica-reinforced elastomers,” Physica A, 2002, 314, 686–695. 109. Schaefer, D. W.; Vu, B. T. N.; Mark, J. E. “The effect of interphase coupling on the structure and mechanical properties of silica-siloxane composites,” Rubber Chemistry and Technology, 2002, 75, 795–810. 110. Rajan, G. S.; Sur, G. S.; Mark, J. E.; Schaefer, D. W.; Beaucage, G. “Preparation and characterization of some unusually transparent poly(dimethylsiloxane) nanocomposites,” J Polym Sci Pol Phys, 2003, 41, 1897–1901. 111. Suryawanshi, C. N.; Pakdel, P.; Schaefer, D. W. “Effect of drying on the structure and dispersion of precipitated silica,” Journal of Applied Crystallography, 2003, 36, 573–577. 112. Vu, B. T. N.; Mark, J. E.; Schaefer, D. W. “Interfacial modification for controlling silicapolysiloxane interactions and bonding in some elastomeric composites,” Compos Interface, 2003, 10, 451–473. 113. Petrovic, Z. S.; Cho, Y. J.; Javni, I.; Magonov, S.; Yerina, N.; Schaefer, D. W.; Ilavsky, J.; Waddon, A. “Effect of silica nanoparticles on morphology of segmented polyurethanes,” Polymer, 2004, 45, 4285–4295. 114. Schaefer, D. W.; Justice, R. S.; Koemer, H.; Vaia, R. A.; Zhao, J.; Yang, M.; Vale, J. In Large-Scale Morphology of Dispersed Layered Silicates, MRS 2004. 115. Justice, R. S.; Schaefer, D. W.; Vaia, R. A.; Tomlin, D. W.; Bunning, T. J. “Interface morphology and phase separation in polymer-dispersed liquid crystal composites,” Polymer, 2005, 46, 4465–4473. 116. Vale, J. M.; Justice, R. S.; Schaefer, D. W.; Mark, J. E., “Calcium alginate barrier films modified by montmorillonite clay,” J Macromol Sci Phys, 2005, B44, 821–831. 117. Zhao, C. G.; Hu, G. J.; Justice, R.; Schaefer, D. W.; Zhang, S. M.; Yang, M. S.; Han, C. C. “Synthesis and characterization of multi-walled carbon nanotubes reinforced polyamide 6 via in situ polymerization,” Polymer, 2005, 46, 5125–5132. 118. Zhao, J.; Schaefer, D. W.; Shi, D. L.; Lian, J.; Brown, J.; Beaucage, G.; Wang, L. M.; Ewing, R. C. “How does surface modification aid in the dispersion of carbon nanofibers?” Journal of Physical Chemistry B, 2005, 109, 23351–23357. 119. Zhao, L. H.; Clapsaddle, B. J.; Satcher, J. H.; Schaefer, D. W.; Shea, K. J. “Integrated chemical systems: The simultaneous formation of hybrid nanocomposites of iron oxide and organo silsesquioxanes”,Chem Mater, 2005, 17, 1358–1366. 120. Schaefer, D. W.; Justice, R. S. “How nano are nanocomposites?” Macromolecules, 2007, 40, 8501–8517. 121. Chen, C. G.; Justice, R. S.; Schaefer, D. W.; Baur, J. W. “Highly dispersed nanosilica-epoxy resins with enhanced mechanical properties,” Polymer, 2008, 49, 3805–3815. 122. Schaefer, D. W.; Zhao, J.; Dowty, H.; Alexander, M.; Orler, E. B. “Nanofibre reinforcement of soft materials,” Soft Matter, 2008, 4, 2071–2079. 123. Schaefer, D.; Brown, J. M.; Anderson, D. P.; Zhao, J.; Chokalingam, K.; Tomlin, D.; Ilavsky, J. “Structure and dispersion of carbon nanotubes,” Journal of Applied Crystallography, 2003, 36, 553–557. 124. Schaefer, D. W.; Zhao, J.; Brown, J. M.; Anderson, D. P.; Tomlin, D. W. “Morphology of dispersed carbon single-walled nanotubes,” Chemical Physics Letters, 2003, 375, 369–375. 125. Schaefer, D. W.; Rieker, T.; Agamalian, M.; Lin, J. S.; Fischer, D.; Sukumaran, S.; Chen, C. Y.; Beaucage, G.; Herd, C.; Ivie, J. “Multilevel structure of reinforcing silica and carbon,” Journal of Applied Crystallography, 2000, 33, 587–591. 126. Green, D. L.; Lin, J. S.; Lam, Y. F.; Hu, M. Z. C.; Schaefer, D. W.; Harris, M. T. “Size, volume fraction, and nucleation of stober silica nanoparticles,” Journal of Colloid and Interface Science, 2003, 266, 346–358. 127. Matsuoka, H.; Kakigami, K.; Ise, N.; Kobayashi, Y.; Machitani, Y.; Kikuchi, T.; Kato, T. “Ultra-small-angle x-ray-scattering study—preliminary experiments in colloidal suspensions,”

Ultra-Small-Angle X-ray Scattering of Polymers

128.

129. 130.

131. 132.

133.

134. 135.

136. 137.

138.

139.

140.

141. 142. 143. 144. 145. 146. 147.

87

Proceedings of the National Academy of Sciences of the United States of America, 1991, 88, 6618–6619. Morvan, M.; Espinat, D.; Lambard, J.; Zemb, T., “Ultrasmall-angle and small-angle x-rayscattering of smectite clay suspensions,” Colloids and Surfaces a-Physicochemical and Engineering Aspects, 1994, 82, 193–203. Arora, A. K.; Tata, B. V. R., “Interactions, structural ordering and phase transitions in colloidal dispersions,” Advances in Colloid and Interface Science, 1998, 78, 49–97. Shalkevich, A.; Stradner, A.; Bhat, S. K.; Muller, F.; Schurtenberger, P. “Cluster, glass, and gel formation and viscoelastic phase separation in aqueous clay suspensions,” Langmuir, 2007, 23, 3570–3580. Smith, W. E.; Zukoski, C. F. “Hard structured particles: Suspension synthesis, characterization, and compressibility,” Langmuir, 2004, 20, 11191. Matsuoka, H.; Kakigami, K.; Ise, N. “Ultra-small-angle x-ray-scattering study of the structure of colloidal dispersions,” Proceedings of the Japan Academy Series B-Physical and Biological Sciences, 1991, 67, 170–175. Peyre, V.; Spalla, O.; Belloni, L.; Nabavi, M. “Stability of a nanometric zirconia colloidal dispersion under compression: Effect of surface complexation by acetylacetone,” Journal of Colloid and Interface Science, 1997, 187, 184–200. Yoshida, H.; Yamanaka, J.; Koga, T.; Ise, N.; Hashimoto, T. “Novel crystallization process in dilute ionic colloids,” Langmuir, 1998, 14, 569–574. Yoshida, H.; Yamanaka, J.; Koga, T.; Koga, T.; Ise, N.; Hashimoto, T. “Transitions between ordered and disordered phases and their coexistence in dilute ionic colloidal dispersions,” Langmuir, 1999, 15, 2684–2702. Harada, T.; Matsuoka, H. “Ultra-small-angle x-ray and neutron scattering study of colloidal dispersions,” Current Opinion in Colloid & Interface Science, 2004, 8, 501–506. Lutterbach, N.; Versmold, H.; Reus, V.; Belloni, L.; Zemb, T., “Charge-stabilized liquidlike ordered binary colloidal suspensions. 1. Ultra-small-angle x-ray scattering characterization,” Langmuir, 1998, 15, 337. Anderson, B. J.; Gopalakrishnan, V.; Ramakrishnan, S.; Zukoski, C. F. “Scattering for mixtures of hard spheres: Comparison of total scattering intensities with model,” Phys Rev E, 2006, 73, 031407. Zhang, F.; Long, G. G.; Jemian, P. R.; Ilavsky, J.; Milam, V. T.; Lewis, J. A. “Quantitative measurement of nanoparticle halo formation around colloidal microspheres in binary mixtures,” Langmuir, 2008, 24, 6504–6508. Sztucki, M.; Narayanana, T.; Beaucage, G. “In situ study of aggregation of soot particles in an acetylene flame by small-angle x-ray scattering,” Journal of Applied Physics, 2007, 101, 114304 Shah, S. A.; Chen, Y. L.; Ramakrishnan, S.; Schweizer, K. S.; Zukoski, C. F. “Microstructure of dense colloid-polymer suspensions and gels,” J Phys-Condens Mat, 2003, 15, 4751–4778. Ramakrishnan, S.; Chen, Y. L.; Schweizer, K. S.; Zukoski, C. F. “Elasticity and clustering in concentrated depletion gels,” Phys Rev E, 2004, 70, 040401. Ramakrishnan, S. A. S. S.; Chen, Y. L.; Schweizer, K. S.; Zukoski, C. F. “Scattering studies of the structure of colloid-polymer suspensions and gels,” Langmuir, 2003, 19, 5128–5136. Ramakrishnan, S.; Zukoski, C. F., “Microstructure and rheology of thermoreversible nanoparticle gels,” Langmuir, 2006, 22, 7833–7842. Ramakrishnan, S.; Gopalakrishnan, V.; Zukoski, C. F., “Clustering and mechanics in dense depletion and thermal gels,” Langmuir, 2005, 21, 9917–9925. Shah, S. A.; Chen, Y. L.; Schweizer, K. S.; Zukoski, C. F. “Viscoelasticity and rheology of depletion flocculated gels and fluids,” J Chem Phys, 2003, 119, 8747–8760. Matsuoka, H.; Harada, T.; Yamaoka, H. “An exact evaluation of salt concentration-dependence of interparticle distance in colloidal crystals by ultra-small-angle x-ray-scattering,” Langmuir, 1994, 10, 4423–4425.

88

F. Zhang and J. Ilavsky

148. Matsuoka, H.; Harada, T.; Kago, K.; Yamaoka, H., “Exact evaluation of the salt concentration dependence of interparticle distance in colloidal crystals by ultra-small-angle x-ray scattering .2. The universality of the maximum in the interparticle distance salt concentration relationship,” Langmuir, 1996, 12, 5588–5594. 149. Konishi, T.; Ise, N. “Orientation and lattice structure of microcrystals of colloidal silica particles in dispersions observed by ultra-small-angle x-ray scattering,” Langmuir, 1997, 13, 5007–5010. 150. Harada, T.; Matsuoka, H.; Ikeda, T.; Yamaoka, H., “An exact evaluation of salt concentration dependence of interparticle distance in colloidal crystals by ultra-small-angle x-ray scattering. 3. Confirmation of solid-liquid transition by three-dimensional paracrystal analysis,” Langmuir, 1999, 15, 573–577. 151. Harada, T.; Matsuoka, H.; Ikeda, T.; Yamaoka, H. “Exact evaluation of the salt concentration dependence of interparticle distance in colloidal crystals by ultra-small-angle x-ray scattering. 4. Effect of counterion species and the possibility of new factors for colloidal crystal formation,” Langmuir, 2000, 16, 1612–1619. 152. Tata, B. V. R.; Yamahara, E.; Rajamani, P. V.; Ise, N. “Amorphous clustering in highly charged dilute poly(chlorostyrene-styrene sulfonate) colloids,” Physical Review Letters, 1997, 78, 2660–2663. 153. Pontoni, D.; Narayanan, T.; Petit, J. M.; Grubel, G.; Beysens, D. “Microstructure and dynamics near an attractive colloidal glass transition,” Physical Review Letters, 2003, 90, 188301. 154. Sztucki, M.; Narayanan, T.; Belina, G.; Moussaid, A.; Pignon, F.; Hoekstra, H. “Kinetic arrest and glass-glass transition in short-ranged attractive colloids,” Phys Rev E, 2006, 74, 051504. 155. Menon, S. V. G.; Manohar, C.; Rao, K. S. “A new interpretation of the sticky hard-sphere model,” J Chem Phys, 1991, 95, 9186–9190. 156. Tohver, V.; Smay, J. E.; Braem, A.; Braun, P. V.; Lewis, J. A. “Nanoparticle halos: A new colloid stabilization mechanism,” Proceedings of the National Academy of Sciences of the United States of America, 2001, 98, 8950–8954. 157. Mock, E. B.; Zukoski, C. F. “Determination of static microstructure of dilute and concentrated suspensions of anisotropic particles by ultra-small-angle x-ray scattering,” Langmuir, 2007, 23, 8760–8771. 158. Levitz, P.; Lecolier, E.; Mourchid, A.; Delville, A.; Lyonnard, S. “Liquid-solid transition of laponite suspensions at very low ionic strength: Long-range electrostatic stabilisation of anisotropic colloids,” Europhysics Letters, 2000, 49, 672–677. 159. Kammler, H. K.; Beaucage, G.; Kohls, D. J.; Agashe, N.; Ilavsky, J. “Monitoring simultaneously the growth of nanoparticles and aggregates by in situ ultra-small-angle x-ray scattering,” Journal of Applied Physics, 2005, 97, 054309. 160. Beaucage, G.; Kammler, H. K.; Mueller, R.; Strobel, R.; Agashe, N.; Pratsinis, S. E.; Narayanan, T. “Probing the dynamics of nanoparticle growth in a flame using synchrotron radiation,” Nat Mater, 2004, 3, 370–374. 161. Schaefer, D. W.; Martin, J. E.; Wiltzius, P.; Cannell, D. S., “Fractal geometry of colloidal aggregates,” Physical Review Letters, 1984, 52, 2371–2374. 162. Gopalakrishnan, V.; Shah, S. A.; Zukoski, C. F. “Cage melting and viscosity reduction in dense equilibrium suspensions,” J Rheol, 2005, 49, 383–400. 163. Gopalakrishnan, V.; Zukoski, C. F. “Microstructure of equilibrium fluid clusters in colloidpolymer suspensions,” Physical Review E (Statistical, Nonlinear, and Soft Matter Physics), 2007, 75, 021406–9. 164. Fuchs, M.; Schweizer, K. S. “Structure of colloid-polymer suspensions,” J Phys-Condens Mat, 2002, 14, R239–R269. 165. Fuchs, M.; Schweizer, K. S. “Macromolecular theory of solvation and structure in mixtures of colloids and polymers,” Phys Rev E, 2001, 64, 021514. 166. Men, Y.; Rieger, J.; Roth, S. V.; Gehrke, R.; Kong, X. “Non-affine structural evolution of soft colloidal crystalline latex films under stretching as observed via synchrotron x-ray scattering,” Langmuir, 2006, 22, 8285–8288.

Ultra-Small-Angle X-ray Scattering of Polymers

89

167. Tanaka, H.; Koizumi, S.; Hashimoto, T.; Kurosaki, K.; Kobayashi, S., “Self-assembly of synthetic cellulose during in-vitro enzymatic polymerization process as studied by a combined small-angle scattering method,” Macromolecules, 2007, 40, 6304–6315. 168. Ohshima, A.; Konishi, T.; Yamanaka, J.; Ise, N. “Ordered” Structure in ionic dilute solutions: Dendrimers with univalent and bivalent counterions,” Physical Review E, 2001, 64, 051808. 169. Ania, F.; Puente-Orench, I.; Calleja, F. J. B.; Khariwala, D.; Hiltner, A.; Baer, E.; Roth, S. V. “Ultra-small-angle x-ray scattering study of pet/pc nanolayers and comparison to afm results,” Macromolecular Chemistry and Physics, 2008, 209, 1367–1373. 170. Gebel, G.; Moore, R. B. “Small-angle scattering study of short pendant chain perfuorosulfonated ionomer membranes,” Macromolecules, 2000, 33, 4850–4855. 171. Heitfeld, K. A.; Schaefer, D. W. “Structure-property relationships in flavour-barrier membranes with reduced high-temperature diffusivity,” Soft Matter, 2009, 5, 156–163. 172. Vainio, U.; Maximova, N.; Hortling, B.; Laine, J.; Stenius, P.; Simola, L. K.; Gravitis, J.; Serimaa, R. “Morphology of dry lignins and size and shape of dissolved kraft lignin particles by x-ray scattering,” Langmuir, 2004, 20, 9736–9744. 173. Turell, M. B.; Bellare, A. “A study of the nanostructure and tensile properties of ultra-high molecular weight polyethylene,” Biomaterials, 2004, 25, 3389–3398. 174. Stribeck, N.; Bayer, R.; von Krosigk, G.; Gehrke, R. “Nanostructure evolution of oriented high-pressure injection-molded poly(ethylene) during heating,” Polymer, 2002, 43, 3779– 3784. 175. Wang, Z.-G.; Hsiao, B. S.; Stribeck, N.; Gehrke, R. “Nanostructure evolution of isotropic high-pressure injection-molded uhmwpe during heating,” Macromolecules, 2002, 35, 2200– 2206. 176. Karl, A.; Cunis, S.; Gehrke, R.; Von Krosigk, G.; Lode, U.; Luzinov, I.; Minko, S.; Pomper, T.; Senkovsky, V.; Voronov, A.; Wilke, W. “Investigation of failure mechanisms in polymer composites by simultaneous measurement of ultra-small-angle scattering and acoustic emission during the deformation. I. Method,” Journal of Macromolecular Science-Physics, 1999, B38, 901–912. 177. Minko, S.; Karl, A.; Senkovsky, V.; Pomper, T.; Cunis, S.; Gehrke, R.; Von Krosigk, G.; Lode, U.; Luzinov, I.; Voronov, A.; Wilke, W. “Investigation of failure mechanisms in polymer composites by simultaneous measurement of ultra-small-angle scattering and acoustic emission during the deformation. Ii. Evaluation of the interface strength,” Journal of Macromolecular Science-Physics, 1999, B38, 913–929. 178. Xiao, Z.; Ilavsky, J.; Long, G. G.; Akpalu, Y. A., “How do orientation fluctuations evolve to crystals?” In Lecture Notes in Physics: Progress in Understanding of Polymer Crystallization, Reiter, G.; Strobl, G. R., Eds. Springer: Berlin Heidelberg, 2007; Vol. 714, pp 117–132. 179. Ise, N.; Konishi, T.; Tata, B. V. R. “How homogeneous are “homogeneous dispersions” Counterion-mediated attraction between like-charged species,” Langmuir, 1999, 15, 4176–4184. 180. Kline, S. R. “Reduction and analysis of sans and usans data using igor pro,” Journal of Applied Crystallography, 2006, 39, 895–900. 181. Nelson, A. “Co-refinement of multiple-contrast neutron/x-ray reflectivity data using motofit,” Journal of Applied Crystallography, 2006, 39, 273–276. 182. Forster, F.http://www.chemie.uni-hamburg.de/pc/sfoerster/software.html 183. Kohlbrecher, J.; Bressler, I.http://kur.web.psi.ch/sans1/SANSSoft/sasfit.html 184. Heenan, R.http://www.small-angle.ac.uk/small-angle/Software/FISH.html 185. Svergun, D. I.http://www.embl-hamburg.de/ExternalInfo/Research/Sax/software.html 186. Konarev, P. V.; Petoukhov, M. V.; Volkov, V. V.; Svergun, D. I. “Atsas 2.1, a program package for small-angle scattering data analysis,” Journal of Applied Crystallography, 2006, 39, 277– 286. 187. Konarev, P. V.; Volkov, V. V.; Sokolova, A. V.; Koch, M. H. J.; Svergun, D. I. “Primus: A windows pc-based system for small-angle scattering data analysis,” Journal of Applied Crystallography, 2003, 36, 1277–1282.

90

F. Zhang and J. Ilavsky

188. Svergun, D.; Barberato, C.; Koch, M. H. J. “Crysol—a program to evaluate x-ray solution scattering of biological macromolecules from atomic coordinates,” Journal of Applied Crystallography, 1995, 28, 768–773. 189. Semenyuk, A. V.; Svergun, D. I. “Gnom—a program package for small-angle scattering dataprocessing,” Journal of Applied Crystallography, 1991, 24, 537–540. 190. Barty, A.; Marchesini, S.; Chapman, H. N.; Cui, C.; Howells, M. R.; Shapiro, D. A.; Minor, A. M.; Spence, J. C. H.; Weierstall, U.; Ilavsky, J.; Noy, A.; Hau-Riege, S. P.; Artyukhin, A. B.; Baumann, T.; Willey, T.; Stolken, J.; van Buuren, T.; Kinney, J. H. “Three-dimensional coherent x-ray diffraction imaging of a ceramic nanofoam: Determination of structural deformation mechanisms,” Physical Review Letters, 2008, 101, 055501. 191. Kucheyev, S. O.; Toth, M.; Baumann, T. F.; Hamza, A. V.; Ilavsky, J.; Knowles, W. R.; Saw, C. K.; Thiel, B. L.; Tileli, V.; van Buuren, T.; Wang, Y. M.; Willey, T. M. “Structure of lowdensity nanoporous dielectrics revealed by low-vacuum electron microscopy and small-angle x-ray scattering,” Langmuir, 2007, 23, 353–356. 192. Marliere, C.; Despetis, F.; Etienne, P.; Woignier, T.; Dieudonne, P.; Phalippou, J. “Very largescale structures in sintered silica aerogels as evidenced by atomic force microscopy and ultra-small angle x-ray scattering experiments,” Journal of Non-Crystalline Solids, 2001, 285, 148–153.

R Journal of Macromolecular Science , Part C: Polymer Reviews, 50:91–111, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583720903503494

Preferred Orientation in Polymer Fiber Scattering CHRISTIAN BURGER, BENJAMIN S. HSIAO, AND BENJAMIN CHU Chemistry Department, Stony Brook University, Stony Brook, NY Fiber symmetry is one of the most important sample geometries encountered in both wide-angle x-ray scattering (WAXS) and small-angle x-ray scattering (SAXS) of polymers, applicable both to natural polymers like collagen or cellulose and to many synthetic polymers that come in fiber form or otherwise exhibit cylindrical rotational symmetry. The structural information to be determined in scattering experiments from such fiber systems includes both the structure of the individual structural unit and qualitative and quantitative information about the preferred orientation state of the ensemble. Existing approaches and new developments to analyze fiber scattering patterns are rigorously reviewed. Special emphasis is placed on the calculation of complete SAXS and WAXS fiber scattering patterns, and various practical examples including collagen and cellulose fibers as well as fibers based on copolymers of polyethylene and polypropylene are discussed. Keywords fiber diffraction, wide-angle x-ray scattering, small-angle x-ray scattering, collagen, cellulose, polypropylene, polyethylene

1. Introduction Structure analysis by scattering experiments can be performed on isotropic samples (powders, isotropic bulk, solutions), single crystals, and on systems that are neither isotropic nor perfectly oriented but show preferred orientation of the constituting structural units. The goal of structure analysis of powders, solutions, and single crystals usually is information about the structure itself (e.g., a crystal unit cell or a particle shape). Structure analysis of systems with preferred orientation additionally generates qualitative and quantitative information about the preferred orientation that can be linked to mechanical and other material properties. Frequently, the structure of the actual structural unit is already known and the preferred orientation information becomes the only output of such experiments. The mathematical treatment of preferred orientation in scattering experiments as lined out in this review applies equally to small-angle x-ray scattering (SAXS) and to wide-angle x-ray scattering (WAXS). Among the systems with preferred orientation, systems with cylindrical rotational symmetry, also known as “fiber symmetry” but not limited to actual fibers, play a particularly important role. On the one hand, this is because many systems of interest, e.g., synthetic and natural polymer fibers, show this type of sample geometry. On the other hand, the combination of point focused X-ray beam and 2D detector allows to record (almost) the Received October 18, 2009; accepted November 22, 2009. Address correspondence to Christian Burger, Chemistry Department, Stony Brook University, Stony Brook, NY 11794-3400. E-mail: [email protected]

91

92

C. Burger et al.

Figure 1. Schematic representation of a fiber scattering experiment, showing a calculated WAXS fiber pattern for α-isotactic polypropylene.

complete information contained in 3D reciprocal space in a single 2D detector frame, see Fig. 1, with the only limitations being a possible information loss at large angles near the meridian due to the curvature of the Ewald sphere1 and, of course, the obvious cut-off at large scattering angles due to finite detector size. The ability to capture the complete scattering information in a single detector frame allows unique time-resolved scattering experiments with polymer fibers as a function of a parameter like temperature, stretchingratio, relaxation time, etc. that would not be practicable otherwise. Figure 2 shows schematic sketches of various possible geometrical arrangements in systems with fiber symmetry. In all cases, we consider an individual structural unit (depicted as a decorated cylinder in the sketch) replicated throughout 3D real space. Under the assumption of “simple fiber symmetry,” the structural unit itself shows cylindrical symmetry about its own symmetry axis (depicted as individual arrow for each cylinder), either by its own nature (e.g., SAXS of whole semi-crystalline polymer stacks) or because of an actual rotational average (e.g., WAXS of the crystalline lamellae inside a semi-crystalline polymer fiber). The symmetry axis of each structural unit forms an orientation angle β with the main fiber axis (large vertical arrow in Fig. 2), resulting in a distribution of orientation angles known as orientation distribution function (ODF) g(β) depending a single angle β, assumed to 0 on the pole and π/2 = 90◦ on the equator. β is one the three Euler angles α, β, γ characterizing the orientation of an individual structural unit. The average over the azimuthal angle α of the fiber (cf. Fig. 5) is the condition for the presence of any kind of fiber symmetry, “simple” or otherwise, and the average over the azimuthal angle γ of the individual structural unit is a consequence of “simple fiber symmetry” as discussed above. The case of “general fiber symmetry,” see e.g. ref. 2, with a bivariate ODF g(β, γ ) is considerably more complicated but not very common in praxis, so that we limit this review to the situation of “simple fiber symmetry.” Note also that the orientation angle β of each structural unit and its position in the sample should be uncorrelated in a way not to generate any constructive interference which is usually valid in good approximation. As Fig. 2 shows, simple fiber symmetry is not limited to the prevalent case of parallel orientation depicted in

Preferred Orientation in Polymer Fiber Scattering

93

Figure 2. Schematic sketches of various possible geometrical arrangements of individual structural units (depicted as decorated stacks), tilted with respect to the main fiber axis (vertical arrow).

Fig. 2(a) with an ODF peaked about β = 0, but other situations with ODFs centered about β = π/2 = 90◦ (Fig. 2(b)) or an arbitrary oblique angle (Fig. 2(c)) are possible. The majority of polymer fiber SAXS and WAXS reports found in the literature does not fully exploit the large amount of information contained in those fiber patterns. The data evaluation is frequently limited to obtaining Hermans’ orientation parameter P2 ,3 a single numeric parameter defined in (4·7) describing the width of the ODF, hopefully taking the limitations discussed in section 5. and, for equatorial arcs, (4·9) into account. We hope to show in this review that a complete calculation of the whole SAXS or WAXS fiber scattering pattern allowing a semi-quantitative or even quantitative comparison with the experimental data is the most satisfactory data analysis approach and discuss its technical details. As can be seen in Fig. 1, the dominant feature of preferred orientation in fiber patterns from sufficiently ordered samples, e.g., semi-crystalline polymer fibers, are characteristic arc-shaped peak profiles. One might expect that these arcs could be modelled using a factorization of radial and angular distributions both exhibiting some sort of bell-shape distribution. Under certain conditions this is correct, and the details will be discussed in section 5. It is important to note that while the radial component can often be modeled using traditional peak profiles like Lorentzians or Gaussians, this is usually not a valid approach for the angular component. Figure 3 shows a comparison of the resulting fiber averages of a tilted polar point and a tilted equatorial ring. It is clear, that the precession average of the tilted equatorial ring generates a qualitatively different type of intensity distribution, as would any precessing ring at an off-axis position. Thus, there are geometric relationships between the individual peak profiles of a fiber pattern and their ODF that need to be taken into account for a consistent description of fiber scattering. The correct mathematical treatment of preferred orientation can be difficult, as is shown by the fact that one of the most frequently cited approaches to deal with this problem was recently shown to be incorrect,4 see section 2.

2. Transformation between Axial and Equatorial Profiles The strongest density fluctuation in uniaxially oriented arrangements of polymer molecules as well as of nematic liquid crystals is due to the lateral packing of elongated structural units and results in equatorial arcs that usually dominate the fiber patterns. In this case,

94

C. Burger et al.

Figure 3. Tilting a polar point and an equatorial ring by an orientation angle β and performing a fiber average leads to significantly different results: the polar point generates a simple ring while the precession average of the tilted equatorial ring generates a non-uniform intensity band about the equator.

the angular profile of the equatorial arcs is related to but not identical to the ODF of the structural units. The relationship was first given by Kratky in 1933.5 In Fig. 3, the properly normalized (see section 3.) angular profile of the polar point tilted from its axial position by an angle β in Fig. 3 can be described by a δ-function: δ(φ − β) sin β

(2·1)

The corresponding intensity profile of the equatorial band extending over the interval π/2 ± β is given by the croissant function4, 5 2 π −1 Re(sin2 φ − cos2 β)−1/2

(2·2)

where taking the real part Re removes the imaginary values of the square root for angles φ < π/2 − β outside the equatorial band. The complete transformation between arbitrary axial distributions gax and equatorial distributions geq can now be pieced together as a superposition of croissant functions: π/2 δ(φ − β) gax (φ) = sin β dβ gax (β) sin β 0 π/2 geq (φ) = 2 π −1 gax (β) Re(sin2 φ − cos2 β)−1/2 sin β dβ (2·3) 0

Throughout this section, the axial profile gax can describe either a meridional peak profile of the ODF itself. When the axial profile gax is written as a function of x = cos φ and the equatorial profile geq is written as a function of y = sin φ, gax (φ) ≡ gˆ ax (cos φ) ≡ gˆ ax (x),

(2·4)

geq (φ) ≡ gˆ eq (sin φ) ≡ gˆ eq (y),

(2·5)

Preferred Orientation in Polymer Fiber Scattering

95

they are related by an Abel transformation,6 gˆ eq (y) =

2 π

y

gˆ ax (x) (y 2 − x 2 )−1/2 dx.

(2·6)

0

(2·6) can sometimes be used to analytically calculate the equatorial profile in cases where an analytical solution for the the full transformation kernel F (φ, φ ) to be discussed in section 3. can not be found. For example, the equatorial profile (6·18) for the Maier-Saupe ODF (6·15) can be derived4 from (2·6) without knowledge of the general kernel (6·17). A particularly interesting example for an equatorial profile in a stretched polyethylene 2D fiber WAXS is shown in Fig. 4(a), with a 3D relief detail of the equatorial profile shown in Fig. 4(b). The puzzling aspect of this pattern is that the typical bell shape normally observed in such profiles is replaced by a rather flat plateau. Since the Abel transformation (2·6) constitutes a fractional integral of order 1/2, it can be conjectured that applying another Abel transform to the resulting croissant-shaped distribution leads to the desired plateaushaped distribution with flat top and smooth edges, which is indeed the case.7 A possible physical interpretation of having a croissant function as an ODF is sketched in Fig. 4(c) in terms of a continuous variation of the preferred orientation across the fiber cross-section, presumably due to internal shearing effects during fiber stretching.7 The known inversion of the Abel transformation, d gˆ ax (x) = dx

x

gˆ eq (y) (x 2 − y 2 )−1/2 y dy,

(2·7)

0

Figure 4. Experimental polyethylene fiber WAXS showing equatorial profiles with usual “box patterns” having flat plateaus rather than bell shapes, and their interpretation in terms of an axial ODF given in form of a croissant function, leading to a structural model with a continuous variation of the preferred orientation across the fiber cross-section. Reproduced with permission from ref. 7.

96

C. Burger et al.

could be used to determine the ODF from an equatorial peak as discussed in ref. 3 for cellulose fibers. However, the inversion (2·7) constitutes an “ill-posed problem” and its numerical treatment requires some care.8, 9 Apparently unaware of Kratky’s original work, Leadbetter and Norris in 1979 attempted to independently rederive (2·3) in the context of scattering from nematic liquid crystals,10 leading to one of the most frequently cited and practically employed approaches to analyze equatorial scattering from systems with preferred orientation. Their result differs from (2·3) for the identical problem. It was recently shown that Leadbetter and Norris’s solution was incorrect, and the exact location of the error in their derivation was pointed out.4 Unfortunately, this important correction has not attracted much attention, and the invalid Leadbetter and Norris approach appears to be in continuous and widespread use.

3. General Transformation for Arbitrary Profile As discussed in the introduction, the preferred orientation of a system in simple fiber symmetry is fully described by an ODF g(β) depending on a single angle β describing the distribution of the tilt angles of the individual structural units with respect to the fiber axis. As a probability density distribution, g(β) is normalized, and we choose the following normalization,

2π

α=0

π

g(β) sin β dβ dα = 4 π

=⇒

π/2

g(β) sin β dβ = 1,

(3·1)

0

β=0

where the simplification results from symmetry considerations. We found the normalization (3·1) to be the most convenient but note that it differs from the one chosen in ref. 11 which in turn differs from the one in ref. 12. For the isotropic case, we have g(β) = 1. Let I (s, φ ) be the intensity distribution of the individual structural unit and J (s, φ) be the resulting intensity distribution of the preferentially oriented ensemble. Here s = 2 λ−1 sin θ is the absolute value of the scattering vector s, λ is the wavelength, and 2θ is the scattering angle. Note the two different polar angles φ and φ in their own coordinate frames of the structural unit and the fiber, respectively. In order to calculate the intensity distribution J of the fiber-averaged ensemble, we need to average over the intensity distributions I of the individual structural units, properly weighted with the ODF, and correctly taking the relationships between all involved angles into account. We abbreviate I (s, φ ) as I (φ ) and J (s, φ) as J (φ) for constant s. J (φ) =

1 4π

2π α=0

π

I (φ ) g(β) sin β dβ dα

(3·2)

β=0

From the spherical trigonometric relationships cos φ = cos φ cos β + sin φ sin β cos α cos β = cos φ cos φ + sin φ sin φ cos η

(3·3)

we have the Jacobian11, 12 sin β dβ dα = sin φ dφ dη

(3·4)

Preferred Orientation in Polymer Fiber Scattering

97

Figure 5. Spherical-trigonometric relationships between the scattering vector s and the primary axes of the intensity distribution of the structural unit I (s, φ ) and the oriented ensemble J (s, φ), respectively.12

and can substitute the integration variables in (3·2) from α and β to η and φ : 2π π 1 I (φ ) g(β) sin φ dφ dη 4 π η=0 φ =0 2π 1 1 π = I (φ ) g(β) dη sin φ dφ 2 φ =0 2 π η=0 π/2 ≡ I (φ ) F (φ, φ ) sin φ dφ

J (φ) =

(3·5)

0

where the integration kernel F (φ, φ ) is given by 1 π F (φ, φ ) = g(β) dη π 0

(3·6)

and symmetry considerations allowed to adjust the upper integration limits in both (3·5) and (3·6). It is clear from (3·3) that the transformation kernel is symmetric in φ and φ : F (φ, φ ) = F (φ , φ).

(3·7)

The kernel F (φ, φ ) has an intuitive physical meaning: A narrow peak (approximated as a δ-function) as part of the intensity distribution of the structural unit I (φ ) at a given angle φ0 transforms into a peak profile J (φ) described by this kernel: I (φ ) =

δ(φ − φ0 ) sin φ0

=⇒

J (φ) = F (φ, φ0 ).

(3·8)

98

C. Burger et al.

Since the kernel itself can describe a valid peak shape, it must be normalized,

π/2

F (φ, φ0 ) sin φ dφ = 1

(3·9)

0

for arbitrary φ0 , so that we also have

π/2 π/2 0

F (φ, φ ) sin φ dφ sin φ dφ = 1,

(3·10)

0

and

π/2

J (s, φ) sin φ dφ =

0

π/2

I (s, φ ) sin φ dφ ,

(3·11)

0

i.e., the preferred orientation effect only redistributes the existing intensity but leaves the invariant constant, as it should be. Setting φ = 0 in (3·6) with (3·3) restores the ODF itself, F (φ, 0) = g(φ)

(3·12)

so that the ODF could in some approximation be directly determined if a meridional peak could be experimentally observed (which is usually not the case for WAXS from most polymer fibers in typical 2θ -ranges). The relationship between F (φ, π/2) = geq (φ) and the ODF g(β) = gax (β) was discussed in section 2. For the isotropic case, we have: g(β) = 1 =⇒ F (φ, φ ) = 1.

(3·13)

Note that the integral transformation (3·5) is a linear operation, so that more complicated ODFs could be superimposed as linear combinations, ˜ g(β) =

N

fn gn (β),

n=1

F˜ (φ, φ ) =

N

fn Fn (φ, φ ),

(3·14)

n=1

where the normalization condition (3·1) requires that N n=1 fn = 1. The most frequent application of (3·14) together with (3·13) is the addition of an isotropic fraction f to an arbitrary ODF: ˜ g(β) = f + (1 − f ) g(β), F˜ (φ, φ ) = f + (1 − f ) F (φ, φ ).

(3·15)

See ref. 13,14, and Fig. 10 for examples of the application of (3·15) to WAXS from cellulose fibers.

Preferred Orientation in Polymer Fiber Scattering

99

4. Legendre Expansions In a similar way as a convolution integral simplifies upon Fourier or Laplace transformation, the integral transformation (3·5) can be reduced to a product when all involved functions are expanded into series of Legendre polynomials.12, 15, 16 Due to symmetry constraints, only those Pnm have non-zero coefficients where m = 0 and n is even. Hence, let g be expanded as g(β) =

∞

an P2n (cos β),

(4·1)

n=0

where the coefficient an is given by

π/2

an = (1 + 4 n)

g(β) P2n (cos β) sin β dβ,

(4·2)

0

using the orthogonality relation

π/2

P2m (cos φ) P2n (cos φ) sin φ dφ =

0

δmn . 1 + 4n

(4·3)

Note that our normalization differs from the one used in ref. 12. The kernel F (φ, φ ) can now be written as12, 15 F (φ, φ ) =

∞

an P2n (cos φ) P2n (cos φ ).

(4·4)

n=0

The Legendre polynomials are the eigenfunctions of the integral operator (3·5),

π/2

P2n (cos φ ) F (φ, φ ) sin φ dφ =

0

an P2n (cos φ). 1 + 4n

(4·5)

Noting that P0 (cos φ ) = 1, we can understand the non-obvious normalization (3·9),

π/2

F (φ, φ ) sin φ dφ = a0 P0 (cos φ) = 1,

(4·6)

0

that appears to be difficult to prove by other means, as is trying to solve this integral for explicit kernels F (φ, φ ) like those in (6·13) or (6·17). Due to the normalization of g, a0 = 1. The next coefficient a1 is related to Hermans’ orientation parameter P2 ,3 also known as the nematic order parameter, P2 = 0

π/2

a1 3 cos2 β − 1 g(β) sin β dβ = , 2 5

(4·7)

which assumes the values of P2 = 1 for perfect parallel orientation, P2 = 0 for totally random orientation, and P2 = −1/2 for perfect perpendicular orientation, respectively. Setting φ = π/2 in (4·4), we obtain the expansion F (φ, π/2) =

∞ (−1)n (2 n)! n=0

4n n! n!

an P2n (cos φ),

(4·8)

100

C. Burger et al.

from which it can be seen that P2,π/2 = −

P2,0 , 2

(4·9)

so that an axial P2 , i.e., Hermans’ orientation parameter of the ODF, can be obtained from a P2 extracted from an equatorial arc by multiplying with −2, if the approximation (5·14) holds. When I and J are expanded into Legendre series corresponding to (4·1) with parameters bn and cn , respectively, the integral transformation (3·5) reduces to the product cn =

an bn 1 + 4n

(4·10)

which could be inverted to a division if either I for known g or g for known I is to be retrieved from a given J .12 When applied to experimental data, Legendre expansions will primarily be useful when all involved functions are broad. For high degrees of orientation and/or narrow peak widths of I , the convergence of the Legendre expansions will be poor and numerical stability problems can arise.

5. Factorizable Normalized Peak Profiles We consider WAXS or SAXS from a fiber system with sufficient translational periodic order such that discrete scattering peaks are generated. The intensity distribution I of a discrete peak of the structural unit will have certain widths based on crystallite size and disorder effects. For convenience and without significant loss of generality, we assume this intensity distribution I (but not, in general, the orientation-smeared arc-shaped intensity distribution J ) to be factorizable into a product of normalized distributions. In cylindrical coordinates, s12 = s sin φ , s3 = s cos φ , we have I (s) = I (s12 , s3 ) = Iint H12 (s12 ) H3 (s3 ) where the individual H distributions shall be normalized, ∞ H12 (s12 ) 2 π s12 ds12 = 1,

(5·2)

0 ∞ −∞

so that

(5·1)

H3 (s3 ) ds3 = 1,

(5·3)

I (s) d3 s = Iint .

(5·4)

Let H0 be a 1D normalized bell-shaped distribution of unity integral width, e.g. a Gaussian, H0 (t) = exp(−π t 2 ),

(5·5)

H0 (t) = (1 + π 2 t 2 )−1 .

(5·6)

or Lorentzian distribution,

Preferred Orientation in Polymer Fiber Scattering

101

Figure 6. Various collagen SAXS patterns showing meridional reflections with different degrees of preferred orientation. Reproduced with permission from ref. 17.

Possible candidates for the functions H12 and H3 in (5·1) with centers a and adjustable widths b12 or b3 could be: H3 (s3 ) = b3−1 H0(s3 /b3 ) for a = 0 H3 (s3 ) = (2 b3 )−1 ± H0[(s3 ± a)/b3 ] H12 (s12 ) ∼ = (2 π b12 s12 )

−1

(5·7) for a > 0

H0[(s12 − a)/b12 ]

for a b

(5·8) (5·9)

Note that (5·9) is an approximation that tends to be very good as long as there is no significant overshoot of the shifted H0 towards the negative region of the abscissa. For an H12 centered about s12 = 0, a construct based on a 1D distribution H0 will, in general, not be applicable so that one of the following distributions could be used, 2 −2 2 H12 (s12 ) = b12 exp −π s12 b12 , (5·10) −3/2 −2 2 2 H12 (s12 ) = b12 1 + 2 π s12 b12 . (5·11) Figure 6 shows meridional SAXS of various collagen samples with different degrees of preferred orientation. The undistorted intensity distribution of a single peak is assumed to be factorizable as in (5·1). For each peak, the integrated intensity Iint , the peak position s3 = a, the lateral width b12 and the longitudinal width b3 are parameters to be retrieved. For meridional collagen SAXS, it can further be assumed, that the positions a are equidistant and that b12 is constant, reducing the overall number of parameters of interest. The orientation distribution g is taken to be peaked around β = 0. Figure 6(a) shows an example of a system with a very high degree of preferred orientation, combined with a relatively broad lateral width b12 . In this case, the effect of the preferred orientation is barely noticeable, does not lead to an appreciable curvature of the reflections, and can in general be neglected.

102

C. Burger et al.

Figure 7. Comparison of (a) experimental 1st order meridional fish bone collagen SAXS (the figure shows a near-beamstop detail of figure 6(b)) and (b) a calculated peak profile J , using (3·5) with (6·13), where I is given by (5·1) with (5·10) and (5·8) with (5·5), a = 1, b12 = 1.3, b3 = 0.09, p = 40. Note that the maximum of the peak distribution is not exactly located at s = (0, a −1 ) which could have implications for the use of such patterns for calibration purposes. Reproduced with permission from ref. 17.

Figure 6(b) shows the situation for an intermediate degree of preferred orientation and broad b12 . The reflections are appreciably curved but that curvature is not simply circular with its center at the origin of reciprocal space as is the case in Fig. 6(c). It appears that there is no good approximation to treat this case satisfactorily so that the exact relationship (3·5) needs to be integrated numerically, as shown in Fig. 7 using the transformation kernel based on Onsager’s ODF (6·10) and parameters as indicated.17 Note how the spine of the peak in Fig. 7 is curved between the dotted layer line (representing the situation of Fig. 6(a)) and the dashed unit circle (representing the situation of Fig. 6(c)). Note also that the center of the peak is not at the ideal position of the reciprocal period. In the limit of a very broad peak, the reciprocal period is shifted toward the point of inflection at the lower angle flank of the peak. It is also of interest to note, that the separation of preferred orientation and lateral constant width information that normally requires a sequence of peaks can here be carried out for a single peak profile. Figure 8(c) shows a nice example for the final result of an analysis of both the equidistant meridional reflections as well as the broad and diffuse equatorial butterfly pattern due to disordered mineral stacks in mineralized intramuscular herring bone.18 In Fig. 6(c), we have the combination of a low degree of preferred orientation and a lateral width b12 that is small enough to not noticeably distort the reflections from their circular arc shapes. These reflections should, in good approximation, be factorizable into their radial and angular contributions, as discussed in the next section. In view of (3·11), Iint could in principle be obtained by integration over the orientationally smeared peak profile J , but a peak fitting procedure with Iint as adjustable parameter is usually a preferable alternative, especially if non-constant backgrounds and/or overlapping peaks are present. Furthermore, when the scattering angles are large enough that the curvature of the Ewald sphere can no longer be neglected so that its correction leads to a region of missing information around the meridian,1 a peak fit will interpolate this missing region and takes its contribution to the total integral into account, as shown for the example of SAXS from intramuscular fish bone in Fig. 9. In WAXS of polymer fibers measured up to suffiently large scattering angles, the situation can occur that the missing region about the meridian due to the curvature of the Ewald sphere is not acceptable because complete reflections near the meridian are not

Preferred Orientation in Polymer Fiber Scattering

103

Figure 8. Comparison of experimental and calculated SAXS patterns for mineralized herring bone, taking both the equidistant meridional reflections and the equatorial butterfly pattern into account. Reproduced with permission from ref. 19, see also ref. 18 for a detailed account on this analysis.

observable and not correctable in a way shown in Fig. 9. In this case, a single exposure in normal beam geometry into a single 2D detector frame is no longer sufficient; the sample needs to be tilted and multiple exposures are required. An example mapping an undistorted 2D section through the 3D intensity distribution of polyacrylonitrile (PAN) fibers using a modified fiber diffractometer in symmetric transmission with a point detector was given by Liu and Ruland.20 Consider a meridional peak in parallel orientation (g peaked around β = 0) or an equatorial peak in perpendicular orientation (g peaked around β = π/2) and assume that the lateral width b12 of I given by (5·1) is small compared to the width contributed by the

Figure 9. SAXS of unmineralized intramuscular fish bone and the effect of the curvature of the Ewald sphere. Even a small (and practically unavoidable) tilt angle of 0.83◦ leads to a significant effect in the corrected figure (b) that cannot be neglected for the higher meridional orders. The calculated figure (c) interpolates the missing region about the meridian and produces the correct integrated intensities. Reproduced with permission from ref. 17.

104

C. Burger et al.

orientation distribution g, as in Fig. 6(c). Under these conditions, H12 can be approximated by a δ-function, so that (3·5) can be solved to yield a factorizable function for J :

π/2

J (s, φ) = 0

δ(s sin φ ) H3 (s cos φ ) F (φ, φ ) sin φ dφ π s 2 sin φ

H3 (s) H3 (s) F (φ, 0) = g(φ). 2 π s2 2 π s2

=

(5·12)

(5·12) shows that, under these conditions, the orientation distribution g can be directly obtained as the angular part of the peak shape. Here and below in this section, the distributions J are normalized,

∞ π/2

0

J (s, φ) 4 π s 2 ds sin φ dφ = 1,

(5·13)

0

so that the integrated intensity Iint (5·4) could be obtained as a scaling parameter in a fit without the need for any integrations. Note that the information about b12 is lost in (5·12) and cannot be retrieved from a pattern like the one in Fig. 6(c). The corresponding case for an equatorial peak in parallel orientation or a meridional peak in perpendicular orientation, where I is given by (5·1) with (5·7) and b3 is small compared to the orientation width, leads to π/2 H12 (s sin φ ) δ(s cos φ ) F (φ, φ ) sin φ dφ J (s, φ) = 0

H12 (s) F (φ, π/2). = 2s

(5·14)

Note that the angular part F (φ, π/2) of such a factorizable peak is not identical to the orientation distribution g but related to it in a way, that does not require the complete knowledge of the kernel F (φ, φ ), as discussed in the previous section 2.. If an off-axis peak can be approximated by a polar factorization with negligible width in φ , the following J is obtained:

π/2

J (s, φ) = 0

=

H0 [(s − a)/b] δ(φ − φ0 ) F (φ, φ ) sin φ dφ 4 π s2 b sin φ

H0 [(s − a)/b] F (φ, φ0 ). 4 π s2 b

(5·15)

Note that the normalization condition (5·13) applied to (5·15) requires that the non-obvious (3·9) holds. The approximations (5·12), (5·14), and (5·15) are usually sufficient for the vast majority of wide-angle fiber patters, provided that 1. the assumptions of simple fiber symmetry are valid, 2. the peaks are narrow enough due to sufficient crystallite sizes and long-range order, and 3. the long-range order is 3D in nature. Under these conditions, and when the kernel F (φ, φ ) can be given in analytical form, the complete fiber pattern can be modeled without the need for numerical integrations.

Preferred Orientation in Polymer Fiber Scattering

105

6. Analytic Transformation Kernels Analytic expressions for the general transformation kernel F (φ, φ ) are very convenient, but only a small number of empirical ODFs actually allow an analytical integration of (3·5). We will usually write the ODF as a function of cos β and substitute (3·3) in the form cos β = cos φ cos φ + sin φ sin φ cos η ≡ x + y cos η

(6·1)

so that the resulting kernel F (φ, φ ) is given as a function of x = cos φ cos φ and y = sin φ sin φ . 6.1 Poisson Kernel The classical choice for an ODF with a narrow head and long tails similar to a Lorentzian distribution is the Poisson kernel:11 gPK (β) =

p1/2 (1 + p) −1 [(1 + p)2 − 4 p cos2 β] atanh(p1/2 )

(6·2)

(6·2) describes an ODF peaked about β = 0 for p > 0 and an ODF peaked about β = π/2 for p < 0, respectively. The limits p = 1 and p = −1 generate infinitely narrow ODFs while p = 0 describes the isotropic case. Hermans’ orientation parameter P2 using (4·7) is given by: P 2,PK =

p1/2 1 3 (1 + p) 1+p− − . 4p atanh(p1/2 ) 2

(6·3)

The kernel F (φ, φ ) can be obtained in analytical form:

−1/2 p1/2 (1 + p + 2 p−1/2 x)2 − 4 p y 2 FPK (φ, φ ) = 2 atanh(p1/2 )

−1/2

, + (1 + p − 2 p−1/2 x)2 − 4 p y 2

(6·4)

which is equivalent, apart from our differing normaliztion, to the form given in ref. 11, but is more consistent within the notation used throughout this section. Note that (6·4) constitutes the rare case of an analytical kernel for both parallel and perpendicular orientation. However, the Lorentzian-like peak shape with very long tails may not be appropriate for all practically encountered systems. 6.2 Cosp and Sinp Orientation Distributions The classical choices for Gaussian-like ODFs with broad heads and short tails are11 gCP (β) = (1 + p) |cos β|p

(6·5)

for parallel and gSP (β) =

2 [(3 + p)/2] sinp β π 1/2 (1 + p/2)

(6·6)

106

C. Burger et al.

for perpendicular orientation. Hermans’ orientation parameters P2 are given by: P 2,CP =

3 (1 + p) 1 − 2 (3 + p) 2

(6·7)

P 2,SP =

1 3 − , 2 (1 + p) 2

(6·8)

and

respectively. For | cos β|p with integer p, a solution for the kernel F (φ, φ ) based on a binomial expansion is given in ref. 11. A previously unknown solution for arbitrary p is FCP (φ, φ ) =

i (p + 2) [f (x + y, x − y) − f (x − y, x + y)]

(p + 3/2)

π 1/2

with f (u, v) = |u|p u1/2 v −1/2 2 F1 (1/2, 1 + p; 3/2 + p; u/v),

(6·9)

where 2 F1 is a hypergeometric function. An analytical solution for the kernel F (φ, φ ) for sinp β with arbitrary p does not appear to be feasible. 6.3 Onsager’s Orientation Distribution Inspection of (3·6) with (6·1) leads to the straightforward conclusion that a simple solution of the integral could be obtained if g(β) ∼ exp(p cos β) since exp[p (x + y cos η)] can be factorized and the remaining η-dependent part can be integrated. Taking symmetry and normalization into account leads to gON (β) = p csch(p) cosh(p cos β),

(6·10)

where csch(p) = 1/ sinh(p). The ODF (6·10) was first used by Onsager21 as an empirical trial function for the hard-rod fluid. Note that (6·10) remains perfectly well behaved for broad distributions, i.e. small p, while |cos β|p develops a singularity at β = π/2. (6·10) can be expanded into a series of Legendre polynomials and modified spherical Bessel functions in (z) = [π/(2 z)]1/2 In (z) where In is the modified Bessel function of the first kind of order n, gON (β) = p csch(p)

∞

(1 + 4 n) i2n (p) P2n (cos β),

(6·11)

n=0

from which the coefficient an of the Legendre expansion (4·1) can be extracted, e.g. to use it in (4·10). Hermans’ orientation parameter P2 is given by

(6·12) P 2,ON = 1 − 3 p−1 coth(p) − p−1 . The kernel F (φ, φ ) has a particularly simple shape: FON (φ, φ ) = p csch(p) cosh(p x) I0 (p y)

(6·13)

where I0 is the modified Bessel function of the first kind of order 0. For φ = π/2, this takes the form FON (φ, π/2) = p csch(p) I0 (p sin φ).

(6·14)

Preferred Orientation in Polymer Fiber Scattering

107

Figure 10. Comparison of experimental and calculated WAXS for two samples of dissolved and respun cellulose fibers showing predominantly a cellulose II crystal structure. Reproduced with permission from ref. 14.

Curiously, an equivalent form of (6·13) was found in ref. 22 as an approximate solution for a Gaussian orientation distribution, in our notation given by g(β) ∼ exp(−p β 2 /2), in the limit of large p, not noting its connection to Onsager’s function (6·10). The combination of a reasonable shape for the ODF (6·10) and a simple form for the kernel (6·13) makes this ODF the preferred choice when it is applicable (i.e., parallel orientation), and we have made extensive use of it to generate calculated fiber patterns for both natural and synthetic polymer fibers. Examples are shown in Fig. 10 for cellulose and in Fig. 11 for a polypropylene copolymer. In both cases, the calculation of the full WAXS fiber pattern was based on (5·15) with (3·15) and (6·13). The factorizable exp(p cos β) = exp(p x) exp(y cos η) (6·10) can, unfortunately, not be easily adapted to perpendicular orientation; trying exp(p sin β) does not lead to an analytical solution for the kernel so that other ODFs are preferred for perpendicular orientation.

Figure 11. Comparison of experimental and calculated WAXS for a sheared sample of a propylenebutylene statistical copolymer showing fiber symmetry. The lower left quadrant of the 2D WAXS pattern in the left shows a calculated WAXS pattern for pure α-iPP, the upper right quadrant shows a calculated WAXS pattern for pure γ -iPP, and the two remaining noisy quadrants show the experiment data. The 3D relief plot in the right shows calculated WAXS for a superposition of 20% α-iPP and 80% γ -iPP in the two smooth quadrants and the experimental data in the noisy quadrants. We found such quadrant plots to be very instructive for the comparison of calculated and experimental fiber patterns. Reproduced with permission from ref. 23.

108

C. Burger et al.

6.4 Maier-Saupe Orientation Distribution The Maier-Saupe ODF24 is defined as gMS (β) = c exp(p cos2 β) = c exp(−p sin2 β),

(6·15)

where the normalization constant c is given by c=

2 p1/2 , π 1/2 erfi p1/2

(6·16)

x erfi(x) = erf(i x)/ i is the imaginary error function, erf(x) = 2 π −1/2 0 t 2 dt, and c = c exp(p). In analogy to the Poisson kernel, (6·15) is written such that positive p describes parallel orientation (g peaked about β = 0) and negative p describes perpendicular orientation (g peaked about β = π/2). (6·15) is a fairly straightforward choice for an orientation distribution and has been used frequently before, e.g. in ref. 25 to describe the preferred orientation in system containing the tobacco-mosaic-virus, but all treatment of the general integral (3·5) has been numerical so far. While a fully closed form solution for the kernel can still not be given, the following series, which can be derived using the expansion exp(z cos φ) = ∞ n=−∞ In (z) cos(n φ), where In is the modified Bessel function of the first kind of order n, FMS (φ, φ ) = c exp[p(x 2 + y 2 /2)] I0 (2 p x y) I0 (p y 2 /2) +2

∞

I2n (2 p x y) In (p y 2 /2) , (6·17)

n=1

shows excellent convergence behavior, so that in typical situations a very small number of terms of the sum need to be evaluated, and the calculation of full fiber patterns using (6·17) becomes computationally feasible. For φ = π/2, (6·17) reduces to (6·18) F (φ, π/2) = c exp p2 sin2 φ I0 p2 sin2 φ . An example for perpendicular orientation, i.e., ODF peaked about β = π/2, that cannot be treated using Onsager’s ODF but can be treated using the Maier-Saupe ODF with negative parameter p is found for a-axis or b-axis orientation in polyethylene fibers,26 see Fig. 12.

7. Off-Axis Centered ODFs, Four-Point Patterns So far, we have only considered ODFs centered about β = 0 for parallel orientation (Fig. 2(a)) or ODFs centered about β = π/2 for perpendicular orientation (Fig. 2(b)). As sketched in Fig. 2(c), ODFs centered about oblique tilt-angles are also possible. SAXS from lamellar systems with such ODFs leads to characteristic four-point patterns. At first sight one might be tempted to create the ODF by shifting an existing ODF: ˜ g(β) ∼ g(β − β0 ).

(7·1)

However, it quickly becomes apparent that ODFs of the type (7·1) are difficult to normalize and in general not recommended. The correct approach is to use one of the analytic kernels

Preferred Orientation in Polymer Fiber Scattering

109

Figure 12. Calculated WAXS fiber patterns for orthorhombic (a–c) and monoclinic (d) polyethylene with symmetry and orientation as indicated, using (5·12) and (5·15) with (5·5) and (6·17), radial integral width b = 0.15 nm−1 , axis labeling s in nm−1 , see ref. 26 for details.

discussed in section 6. as the ODF: ˜ g(β) = F (φ, β0 )

(7·2)

which, in view of (3·9), automatically ensures the correct normalization. An analytical kernel F (φ, φ ) corresponding to the ODF defined in (7·2) can usually not be found so that using Legendre expansions as discussed in section 4. becomes the most favorable approach, especially if the Legendre expansion coefficients of (7·2) can be given in analytical form which is the case for the kernel based Onsager’s ODF in view of (6·11).

8. Conclusions It was the goal of this review to show that a rigorous treatment of preferred orientation effects in SAXS and WAXS from natural and synthetic polymer fibers and other samples with fiber symmetry is possible. The preferred data analysis approach is the calculation of the complete fiber pattern, taking both the parameters of the individual structural units (crystal structures, crystallite or particle sizes and shapes) and of their orientation arrangement in the ensemble into account. Practical examples for the calculation of whole fiber patterns and their comparison to experimental scattering data were shown for natural collagen and cellulose fibers as well as fibers based on copolymers of polyethylene and polypropylene.

Acknowledgments The authors wish to thank Drs. Dufei Fang, Hongwen Zhou, and Lixia Rong for their assistance with the X-ray measurements and data analysis; Drs. Melvin Glimcher and Lila Graham of Harvard Medical School for providing the bone samples. We gratefully acknowledge financial support from the National Institutes of Health (BC) and the National Science Foundation (BH).

References 1. Fraser, R. D. B.; Macrae, T. P.; Miller, A.; Rowlands, R. J. “Digital processing of fiber diffraction patterns,” J. Appl. Cryst., 1976, 9, 81–94. 2. Plaetschke, R.; Ruland, W. “Preferred orientation of the internal structure of carbon layers in carbon-fibers,” Prog. Coll. Polym. Sci., 1985, 71, 140–144.

110

C. Burger et al.

3. Hermans, J. J.; Hermans, P. H.; Vermaas, D.; Weidinger, A. “Quantitative evaluation of orientation in cellulose fibres from the x-ray fibre diagram,” Rec. Trav. Chim. Pays-Bas-J. Roy. Neth. Chem. Soc., 1946, 65, 427–447. 4. Burger, C.; Ruland, W. “Evaluation of equatorial orientation distributions,” J. Appl. Cryst., 2006, 39, 889–891. 5. Kratky, O. “Zum Deformationsmechanismus der Faserstoffe, I.,” Kolloid Z., 1933, 64, 213–222. 6. Bracewell, R. The Fourier Transform and its Applications, 3rd ed., pp. 351–356 McGraw-Hill, New York, 1999. 7. Zuo, F.; Burger, C.; Hsiao, B. S.; Chen, H.; Chiu, D.; Lai, S.-Y. “Equatorial box patterns in WAXD during uniaxial deformation of olefin block copolymer fibers,” Macromolecules, 2009. submitted. 8. Seitsonen, S. “Determination of orientation distributions in fibres and sheets,” J. Appl. Cryst., 1968, 1, 82. 9. Seitsonen, S. “Calculation of orientation distributions in fibres and sheets,” J. Appl. Cryst., 1973, 6, 44–44. 10. Leadbetter, A. J.; Norris, E. K. “Distribution functions in 3 liquid-crystals from x-ray-diffraction measurements,” Mol. Phys., 1979, 38, 669–686. 11. Ruland, W.; Tompa, H. “Effect of preferred orientation on intensity distribution of (hk) interferences,” Acta Cryst. A, 1968, 24, 93–99. 12. Ruland, W. “Elimination of effect of orientation distributions in fiber diagrams,” Colloid Polym. Sci., 1977, 255, 833–836. 13. Chen, X. M.; Burger, C.; Fang, D.; Ruan, D.; Zhang, L.; Hsiao, B. S.; Chu, B. “X-ray studies of regenerated cellulose fibers wet spun from cotton linter pulp in naoh/thiourea aqueous solutions,” Polymer, 2006, 47, 2839–2848. 14. Chen, X. M.; Burger, C.; Wan, F.; Zhang, J.; Rong, L. X.; Hsiao, B. S.; Chu, B.; Cai, J.; Zhang, L. “Structure study of cellulose fibers wet-spun from environmentally friendly naoh/urea aqueous solutions,” Biomacromolecules, 2007, 8, 1918–1926. 15. Deas, H. D. “The diffraction of x-rays by a random assemblage of molecules having partial alignment,” Acta Cryst. A, 1952, 5, 542–546. 16. Lovell, R.; Mitchell, G. R. “Molecular-orientation distribution derived from an arbitrary reflection,” Acta Cryst. A, 1981, 37, 135–137. 17. Burger, C.; Zhou, H. W.; Sics, I.; Hsiao, B. S.; Chu, B.; Graham, L.; and Glimcher, M. J. “Smallangle x-ray scattering study of intramuscular fish bone: collagen fibril superstructure determined from equidistant meridional reflections,” J. Appl. Cryst., 2008, 41, 252–261. 18. Burger, C.; Zhou, H. W.; Wang, H.; Sics, I.; Hsiao, B. S.; Chu, B.; Graham, L.; and Glimcher, M. J. “Lateral packing of mineral crystals in bone collagen fibrils,” Biophysical Journal, 2008, 95, 1985–1992. 19. Zhou, H. W.; Burger, C.; Sics, I.; Hsiao, B. S.; Chu, B.; Graham, L.; and Glimcher, M. J. “Smallangle x-ray study of the three-dimensional collagen/mineral superstructure in intramuscular fish bone,” J. Appl. Cryst., 2007, 40, S666–S668. 20. Liu, X. D.; Ruland, W. “X-ray studies on the structure of polyacrylonitrile fibers,” Macromolecules, 1993, 26, 3030–3036. 21. Onsager, L. “The effects of shape on the interaction of colloidal particles,” Ann.NY Acad.Sci., 1949, 51, 627–659. 22. Holmes, K. C.; Leigh, J. B. “Effect of disorientation on intensity distribution of non-crystalline fibers. 1. Theory,” Acta Cryst. A, 1974, 30, 635–638. 23. Mao, Y.; Burger, C.; Thurman, D. W.; Tsou, A. H.; Hsiao, B. S. “Shear-induced crystallization of propylene-butylene random copolymer revealed by 2D wide-angle x-ray scattering analysis: experiment and simulation,” Macromolecules, 2009. submitted. 24. Maier, W.; Saupe, A. “Eine einfache molekular-statistische Theorie der nematischen kristallinflussigen Phase 1.,” Z. f. Naturforschung A, 1959, 14, 882–889.

Preferred Orientation in Polymer Fiber Scattering

111

25. Oldenbourg, R.; Wen, X.; Meyer, R. B.; Caspar, D. L. D. “Orientational distribution function in nematic tobacco-mosaic-virus liquid-crystals measured by x-ray-diffraction,” Phys. Rev. Lett., 1988, 61, 1851–1854. 26. Keum, J. K.; Burger, C.; Zuo, F.; Hsiao, B. S. “Probing nucleation and growth behavior of twisted kebabs from shish scaffold in sheared polyethylene melts by in situ x-ray studies,” Polymer, 2007, 48, 4511–4519.

Polymer Reviews, 50:113–143, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583721003698853

Reviews Polymeric Membranes for Chiral Separation of Pharmaceuticals and Chemicals AKON HIGUCHI,1,2,3 MIHO TAMAI,4 YI-AN KO,5 YOH-ICHI TAGAWA,4 YUAN-HSUAN WU,6 BENNY D. FREEMAN,6 JUN-TANG BING,7 YUNG CHANG,8 AND QING-DONG LING3,5 1

Department of Chemical and Materials Engineering, National Central University, No. 300 Jung da Rd., Chung-Li, Taoyuan 32001, Taiwan 2 Department of Reproduction, National Research Institute for Child Health and Development, 2-10-1 Okura, Setagaya-ku, Tokyo 157-8535, Japan 3 Cathay Medical Research Institute, Cathay General Hospital, No. 32, Ln 160, Jian-Cheng Road, Hsi-Chi City, Taipei 221, Taiwan 4 Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, B-51 4259 Nagatsuta-cho, Midori-ku, Yokohama, Kanagawa 226-8501, Japan 5 Graduate Institute of Systems Biology and Bioinformatics, National Central University, No. 300 Jung da Rd., Chung-Li, Taoyuan 32001, Taiwan 6 Department of Chemical Engineering, The University of Texas at Austin Center for Energy and Environmental Resources, 10100 Burnet Road, Building 133, Austin, TX 78758, United States 7 Obstetrics & Gynecology Division, Li Shin Hospital, No. 77 Guang Tai Rd., Ping-Zhen, Taoyuan 32405, Taiwan 8 Department of Chemical Engineering, R&D Center for Membrane Technology, Chung Yuan Christian University, 200, Chung-Bei Rd., Chungli, Taoyuan 320, Taiwan The optical resolution or chiral separation of one specific enantiomer from others is in demand for the production of pharmaceuticals because many pharmaceuticals exist as stereoisomers, with each enantiomer having different biological activity. There is considerable demand for separation techniques appropriate for the large-scale resolution of chiral molecules. Chiral separation of racemic mixtures of pharmaceuticals through chiral or achiral polymeric membranes with or without a chiral selector represents a promising system for future commercial application. This article reviews several polymeric materials for the chiral separation of pharmaceuticals. Several chiral separation Received December 10, 2009; accepted January 11, 2010. Address correspondence to Akon Higuchi, Department of Chemical and Materials Engineering, National Central University, Jhongli, Taoyuan 32001, Taiwan. E-mail: [email protected]

113

114

A. Higuchi et al. membranes were prepared from chiral polymers where enantioselectivity was generated from chiral carbons in the main chain. However, it is rather difficult to generate excellent chiral separation membranes from chiral polymers alone, because racemic penetrants mainly encounter the flexible side chains of the membrane polymers. Therefore, chiral separation membranes were also prepared using polymers with a chiral branch. Furthermore, several molecules have been used for specific interactions between the molecules and specific pharmaceuticals or drugs in chiral separation membranes. Cyclodextrins, crown ether derivatives, albumin, and DNA are commonly used as stereoselective ligands in chiral separation membranes. Finally, this article discusses future trends in polymeric materials for chiral separation membranes. Keywords chiral separation, optical resolution, polymeric membranes, pharmaceutical, dialysis

1. Introduction Terrestrial life utilizes only the L enantiomers of amino acids; this is known as the homochirality of life.1 The ability to optically resolve one specific enantiomer from others is important for the production of pharmaceuticals and food products2 because many pharmaceuticals, nutraceuticals, and agricultural chemicals have their stereoisomers, with each enantiomer having different biological activity. In some cases, only one of the isomers has the preferable activity, while the other chiral form may produce undesirable and/or toxic side effects. For example, it has been reported that the severe teratogenic side effects of the drug thalidomide may reside exclusively in the S-enantiomer.3 The (S,S)-diastereomer of ethambutol is effective in the treatment of tuberculosis, but the (R,R)-diastereomer can cause blurred vision, eye pain, and might result in complete blindness.4,5 The FDA and the Committee for Proprietary Medicinal Products (CPMP) now require pharmaceutical companies to produce only a single enantiomer as the therapeutic agent or to clearly demonstrate the appropriateness of using a racemic mixture.3,6,7 This requirement has resulted in considerable demand for separation techniques appropriate for the large-scale resolution (purification) of chiral molecules.3 Worldwide, the market for chiral fine chemicals sold as single enantiomers was $6.63 billion in 2000, and the market is expected to grow at a rate of 13.2% annually, reaching $16.0 billion in 2007.8 There is a growing need for separation techniques appropriate for the large-scale resolution of chiral molecules, although many single enantiomer drugs are produced by stereoselective synthesis. In particular, relatively low-cost pharmaceuticals cannot be produced by stereoselective synthesis. The most widely used methods for the separation of racemic mixtures are diastereomeric salt crystallization,9,10 column chromatography,11–15 and stereoselective enzyme catalysis.16,17 Liquid membranes with immobilized chiral ligands have also been used for chiral separation,3,18–20 although these techniques could be difficult to apply in commercial systems because of the instability of the liquid membranes. An alternative approach is to use an affinity ultrafiltration system in which a large stereoselective ligand is added to the bulk solution to selectively bind, and thus retain, one of the stereoisomers.3 Chiral separation of pharmaceuticals through polymeric membranes with an immobilized chiral selector could be very promising for commercial systems in the future. This paper summarizes polymeric materials for the chiral separation of pharmaceuticals and discusses the future trend of polymeric materials for chiral separation membranes.

2. Permeation and Selective Theory (Analysis) for Chiral Separation Chiral separation membranes preferentially allow a specific enantiomer to adsorb to or diffuse into the membrane. This specificity is generated by chiral recognition sites in the

Membranes for Chiral Separation

115

membranes such as chiral side chains, chiral backbones, or immobilized chiral selectors in polymeric chiral separation membranes. These enantioselective membranes act as selective barriers in the resolution process, and they preferentially transport one enantiomer due to the stereospecific interaction between the enantiomer and chiral recognition sites.21 The transport process of enantiomers through the membranes can be categorized as filtration, dialysis, electrodialysis, and pervaporation, depending on the main driving force of the permeation of enantiomers through the membranes, i.e., pressure gradient, concentration difference, electric field difference, or vapor difference, respectively. The flux of enantiomers, J i , can be defined in the dialysis membranes and filtration membranes as follows: Ji = Qi /At,

(1)

where Qi is the mass of solute i (R-enantiomer or S-enantiomer) allowed to permeate for a given time t, and A is the effective membrane area. The permeability coefficient for the dialysis process with no electric field gradient through the membrane is defined as Pi = Ji L/(Cf − Cp ),

(2)

where Pi is the permeation coefficient of the solute i, L is the membrane thickness, and Cf and Cp refer to the concentrations in the feed solution (solution at the upstream side) and permeate solution (solution at downstream side), respectively. A solution-diffusion mechanism determines the permeation of enantiomers through the homogeneous dense membranes and is described as follows: P = DS,

(3)

where D and S refer to the diffusion coefficient and sorption coefficient (solubility), respectively. The diffusion coefficient D is a kinetically determined coefficient influenced by the membrane and enantiomer characteristics and the interaction between the two. The sorption coefficient is a thermodynamically determined parameter defined as the ratio of the concentration in the membrane (Cm ) to that in the solution (Co ), as shown in Eq. [4]. S = Cm /Co

(4)

The separation factor α is calculated from the concentration of the upstream side and downstream side, and is defined as follows: α = (Cp (R)/Cp (S)/(Cf (R)/Cf (S))

(5)

α = (Cp (S)/Cp (R)/(Cf (S)/Cf (R)),

(6)

or

where Cf (R) and Cf (S) are the concentrations of the R-enantiomer and S-enantiomer in the feed solution (solution at upstream side), respectively. Cp (R) and Cp (S) are the concentrations of the R-enantiomer and S-enantiomer in the permeate solution (solution at downstream side), respectively. The concentrations in the upstream side, Cf (S) and Cf (R),

116

A. Higuchi et al.

are the same in some cases. In this case, α reduces to; a = Cp (S)/Cp (R) or Cp (R)/Cp (S).

(7)

The enantioselectivity of transport through the membrane can be divided into two factors, solubility selectivity and diffusion selectivity. α = P (R)/P (S) = D(R)S(R)/[D(S)S(S)]

(8)

α = P (S)/P (R) = D(S)S(S)/[D(R)S(R)],

(9)

or

where D(R) and D(S) are the diffusion coefficients of the R-enantiomer and S-enantiomer, respectively. S(R) and S(S) are the solubility coefficients of the R-enantiomer and S-enantiomer, respectively. The chiral selectivity of transport through membranes is also evaluated in terms of the enantiomeric excess (ee) of permeates.21 The ee value is defined as the ratio of the concentration difference over the total concentration of both enantiomers in the permeate.22 ee = [Cp (R) − Cp (S)]/[Cp (R) + Cp (S)]

(10)

ee = [Cp (S) − Cp (R)]/[Cp (S) + Cp (R)].

(11)

or

When the concentrations in the feed side Cf (S) and Cf (R) are the same, the separation factor can be calculated from ee using the following equation: α = (1 + ee)/(1 − ee).

(12)

3. Resolution Mechanism through the Membranes The mechanism of chiral separation on polymeric membranes can be categorized as diffusion-selective membranes and sorption-selective membranes.23 Diffusion-selective membranes are usually made of an intrinsically chiral polymer without specific foreign chiral selectors, for example albumin or other proteins, chiral polysaccharide chains or segments, DNA, crown ether derivatives, and oligopeptides. Sorption-selective membranes can be made by embedding or immobilizing chiral selectors in polymer membranes or on the membrane surfaces and these membranes have less selective diffusion but show highly selective sorption. Examples of chiral selectors include crown ether derivatives,24 cyclodextrin, albumin and other proteins, and DNA. In most cases of chiral separation through polymeric membranes, there is a trade-off between diffusion selectivity and solution selectivity; the membranes showing diffusion selectivity for one chiral isomer have sorption selectivity for the opposite chiral isomer. Therefore, the permeation selectivity is determined by whichever selectivity is higher, sorption selectivity or diffusion selectivity. In general, chiral separation membranes with sorption selectivity should be designed with no diffusion selectivity, and vice versa. One method that can be used to reduce the diffusion selectivity of sorption selective concentration-driven permeation chiral separation membranes (dialysis membranes) is to apply an electrical potential that makes the concentration-driven permeation of chiral pharmaceuticals or chemicals a potential-driven permeation. Potentialdriven permeation generally results in the same diffusion coefficient for each isomer.

Membranes for Chiral Separation

117

The driving force for the permeation and separation of chiral pharmaceuticals and chemicals is the concentration difference between feed and permeate solutions for the dialysis method, and a pressure-driven force for ultrafiltration and nanofiltration. Most studies of chiral separation membranes have been performed in dialysis membranes. The main disadvantages of the dialysis method are that the concentration of the final product is more dilute than that of the feed solution, and that permeation is extremely slow. Due to these disadvantages, chiral separation in industrial applications may require ultrafiltration or nanofiltration through chiral separation membranes. In addition to dialysis and filtration, pervaporation via membranes is also useful for the chiral separation process, where the driving force of the permeation is a vapor pressure difference. Shinohara and Aoki et al. reported chiral separation of racemates of 1,3-butanediol, 2-butanol, and their derivatives by pervaporation through a (+)-poly{1[dimethyl(10-pinanyl)silyl]-1-propyne} membrane.25 They reported a permeation rate of 1.19 × 10−3 gm/h and 41.7%ee for racemic 1,3-butanediol. Enantioselective vapor permeation is also effective for chiral separation if the racemic compounds are more or less volatile. Only a few examples have been reported of chiral separation of pharmaceuticals and racemates using the pervaporation method25 although it is expected that this technology will be used more often in the future.

4. Chiral Separation Membranes Prepared from the Chiral Main Chain of Polymers Several chiral separation membranes were prepared from chiral polymers where enantioselectivity was generated from chiral carbons in the main chain. Poly(γ -methyl-Lglutamate),26–28 alginate,29,30 chitosan,29 cellulose,31 and their derivatives (see Fig. 1) are typically used as chiral polymers for the preparation of chiral separation membranes. Examples of chiral separation membranes prepared from a chiral polymer main chain are summarized in Tables 1 and 2, and are reviewed as follows.

Figure 1. Chemical scheme of PMLG (a), sodium alginate (b), cellulose (c) and chitosan (d) used as chiral polymers for preparation of chiral separation membranes.

118

a

Tryptophan Tryptophan Tryptophan Tryptophan Tyrosine Tryptophan Tryptophane Tryptophan Tryptophan Phenylalanine Tryptophan Tryptophan Tryptophan Tryptophan Tryptophan Tryptophan Tryptophan Ac-tryptophan

Targeted molecule

Separation factor 1.4 3.0 2.6 60.0 5.0 2.6 3.4 3.4 99.0 1.1 8.0 14.0 7.0 1.4 1.9 1.1 1.2 2.0

Flux or permeability coefficient (methoda,b) J = 10−6 gm/m2h atm (UF) P = 8.4 × 10−6 g m−1/h (ED) P = 4.2 × 10−8 cm2s−1 (D) P = 4 × 10−7 mol/m2h (D) P = 10−5 mol/m2h (D) P = 2.2 × 10−8 m2s−1 (D) J = 7.5 mg/m2h (UF) J = 24.8 mg/m2h (UF) J = 6.4 mg/m2h (UF) J = 2.11 × 10−12 m2/sec (UF) (UF) (UF) (UF) P = 5.02 × 10−9 cm2/sec (D) P = 1.87 × 10−9 cm2/sec (D) P = 6.04 × 10−9 cm2/sec (D) P = 4.71 × 10−9 cm2/sec (D) P = 4.9 × 10−6 cm/sec (ED)

Method; D indicates dialysis, ED indicates electro dialysis, UF indicates ultrafiltration.

PMLG derivatives PMLG derivatives PMLG derivatives PMLG derivatives PMLG derivatives Crosslinked PMLG Crosslinked alginate Crosslinked alginate Crosslinked chitosan Crosslinked L-phenylalanine Plasma polymerized l-menthol Plasma polymerized d-camphor Plasma polymerized l-menthol Plasma polymerized terpene Plasma polymerized terpene Plasma polymerized terpene Plasma polymerized terpene Polyamide

Chiral sites of membranes

Table 1 Chiral separation through polymeric membranes having chiral main chain

27 26 26 28 28 71 29 30 72 34 73 74 74 75 75 75 75 76

Ref.

Membranes for Chiral Separation

119

Table 2 Chiral separation through polymeric membranes having helical structure of main chain Chiral sites of membranesa Poly(OPSPA) Poly(OPSPA) Poly(CPSPA) Poly(HPSPA) Poly(HPSPA) Copoly(CHPSPA) Copoly(CHPSPA) Copoly(PSDPA) Copoly(PSDPA) Copoly(PSDPA) Cellulose acetate butyrate MTSPOE

Targeted molecule

Flux or permeability coefficient (methodb)

Phenylalanine Tryptophan Phenylalanine Phenylalanine Phenylalanine Phenylalanine Phenylalanine Tryptophan Tryptophan Tryptophan 2-phenyl-1-propanol

P = 1.11 × 10−14 m2/h (D) P = 1.34 × 10−14 m2/h (D) P = 4.34 × 10−14 m2/h (D) P = 2.9 × 10−14 m2/h (D) P = 3.02 × 10−14 m2/h (D) P = 3.94 × 10−14 m2/h (D) P = 3.90 × 10−14 m2/h (D) P = 5.51 × 10−12 m2/h (D) P = 5.15 × 10−12 m2/h (D) P = 3.60 × 10−12 m2/h (D) J = 1.19 × 10−6 g/hr atmcm2 (UF) P = 13.72 m2h−1 (UF)

Separation factor Ref. 640 2.9 7.9 2.6 2.1 4.1 3.5 1.4 1.6 3.4 19.0

37 37 67 67 67 67 67 66 66 66 31

1.5

77

a Poly(OPSPA); Poly(p-(oligopinanylsiloxanyl)phenylacetylene), Poly(CPSPA); Poly(chiral pinanylsiloxanyl)phenylacetylene), Poly(HPSPA); Poly(hydroxylpinanylsiloxanyl)phenylacetylene), Copoly(CHPSPA); Copoly(chiral hydroxylpinanylsiloxanyl)phenylacetylene), Copoly(PSDPA); Copoly(pinanylsilyl)diphenylacetylene), MTSPOE; 1,2-bis(2-methyl-1-triethylsiloxy-1-propeny loxy)ethane derivatives. b Method; D indicates dialysis, UF indicates ultrafiltration.

Aoki et al. prepared chiral separation membranes from poly(γ -methyl-L-glutamate) (PMLG) chiral main chains with achiral side chains of 3-(pentamethyldisiloxanyl)propyl groups.27 The enantioselectivity and permeation rate through the modified PMLG membranes were higher than those through unmodified PMLG membranes by ultrafiltration. The enantioselective permeation continued for more than 160 hrs. Increasing the disiloxane side chain content increased the permeation rate while maintaining the enantioselectivity. The α-helix content of the membranes did not have a significant effect on the enantioselectivity or permeation rate. This result indicates that the higher-order structure did not affect the enantioselective permeation. The enantioselectivity was ascribed to the asymmetric carbons of the main chain rather than to the α-helix conformation of the membrane materials. The adsorption enantioselectivity of this membrane favored D(R)-tryptophan, while the diffusion and permeation selectivity favored L(S)-tryptophan. Therefore, enantioselective permeation was caused by the suppression of D-tryptophan. The authors of this study concluded that the siloxane region was small enough for the asymmetric centers in the main chain to interact with the permeating solutes and to enantioselectively recognize the permeating solute.27 Thoelen used poly(γ -methyl-L-glutamate) (PMLG) membranes transesterified with R 30 for the chiral separation of tryptophan.26 The membranes showed Igepal and Brij enantioselectivity towards tryptophan in pressure driven permeation. An initial enantiomeric excess of 20% was obtained, but a decline in selectivity was observed during the course of the permeation experiment. This decrease was ascribed to the saturation of the enantioselective recognition sites in the membrane based on the mechanism of affinity membranes.

120

A. Higuchi et al.

Changing the driving force into an electrical potential difference solved this problem, resulting in a constant separation factor of 3 for PMLG membranes in the separation of acethyl-tryptophan. Based on these results, electrodialysis with membranes showing affinity to specific enantiomers appears to be a highly promising membrane process for chiral separation, even though the membranes do not show steady state enantioseparation in conventional concentration-driven or pressure-driven membrane processes.26 Photocontrol of membrane chiral recognition has also been developed.32 Yashima et al. prepared membranes of cellulose and amylose derivatives bearing a photoresponsive [4-(phenylazo)phenyl]carbamate residue incorporated at the 2,3,6-,6-, or 2,3-positions of the glucose units.32 Enantioselective adsorption of several neutral racemates (i.e., trans2,3-diphenyloxirane, 1,2,2,2-tetraphenylethanol, hydrochloric salt of oxprenolol, and 1-(9anthryl)-2,2,2-trifluoroethanol) on the photoresponsive membranes was investigated during the course of trans-cis isomerization of the pendant azobenzene residues. The chiral recognition ability was influenced by the trans or cis content, and the trans isomers of the polysaccharide derivatives showed higher enantioselectivity than the cis isomers. The photo-responsiveness of the chiral recognition of these polymers was discussed on the basis of circular dichroism (CD) data, lH nuclear magnetic resonance (NMR) spectroscopic data, and molecular modeling.32 Tris(phenylcarbamate)s of cellulose and amylose were reported to possess the conformations of left-handed 3-fold (3/2) and 4-fold (4/1) helices, respectively, on the basis of X-ray analysis.32,33 Therefore, the photoresponsive cellulose membranes bearing the [4(phenylazo)phenyl]carbamate residue probably have similar conformations.32 According to this hypothesis, the intramolecular hydrogen bonds along the polysaccharide backbone can participate in the formation of the rigid conformations, which should only be possible for the trans isomers. From the structures of trans- and cis- photoresponsive cellulose having [4-(phenylazo)phenyl]carbamate calculated by molecular mechanics, the pendant [4-(phenylazo)phenyl] carbamate residues in the trans-conformation were arranged regularly, while those of the cis-conformation were not because of the bent structure.32 Moreover, the carbamate residues of the photoresponsive cellulose in the cis-conformation are unfavorably concealed behind the bent cis-azobenzene moieties, so that the carbamate residues are not able to interact effectively with racemates.32 These molecular modeling results explain the decrease in adsorption ability and the lower chiral recognition ability of the cis-conformation of the photoresponsive cellulose membranes. Kim et al. prepared crosslinked sodium alginate and chitosan membranes with glutaraldehyde for chiral separation of racemic tryptophan and tyrosine by ultrafiltration.29 Both crosslinked sodium alginate and chitosan membranes were found to be applicable for the chiral separation of racemic tryptophan and tyrosine by a pressure driven process. When a crosslinked chitosan membrane with a 70% swelling index was used for the chiral separation of a racemic tryptophan mixture (0.49 mmol/l aqueous solution), over 98% of enantiomeric excess (e.e.) and 6.4 mg/(m2h) of flux were obtained.29 The presence of five chiral carbons located on the ring structure of sodium alginate and chitosan seemed to induce a chiral environment in the membrane, which is similar to the function of cellulose and its derivatives, making the membrane enantioselective. With an increasing degree of crosslinking, the membrane showed higher enantioselectivity by increasing the interaction between the chiral environment of the membrane and penetrating chiral isomers.29 The other factors that could decrease the intermolecular interaction between the membrane and solutes, such as an increase in operating pressure and an increase in the concentration of the feed solution, acted against the enantioselectivity.29

Membranes for Chiral Separation

121

Masawaki et al. prepared crosslinked L(S)-phenylalanine with glutaraldehyde, and UF membranes were prepared from the blend of crosslinked L-phenylalanine and polysulfone.34 The membrane showed diffusion selectivity of D(R)-phenylalanine, while sorption selectivity of L-phenylalanine was found in concentration-driven permeation. In total, the membrane showed preferential permeation of D-phenylalanine over L-phenylalanine due to diffusion selectivity, and the separation factor was reported to be 1.25 to 4.0 in both concentration-driven and pressure-driven permeation.34

5. Chiral Separation Membranes Prepared from Polymers with a Chiral Branch It is rather difficult to generate excellent chiral separation membranes from chiral polymers alone, because racemic penetrants mainly encounter the flexible side chains of the membrane polymers. Therefore, chiral separation membranes were prepared using polymers with a chiral branch. Some examples of chiral separation membranes prepared from polymers with a chiral branch are shown in Table 3 and reviewed as follows. Lee and Frank prepared polypeptide-modified poly(vinylidene fluoride) (PVDF) membranes for the separation of chiral molecules in ultrafiltration.35 Poly(γ -benzyl-Lglutamates) (PBLG) were vapor-deposited on the PVDF membranes, and the PBLG on PVDF membranes were modified through debenzylation or an ester exchange reaction to produce poly(L-glutamic acid) (PLGA) and polyglutamates with triethylene glycol monomethyl ether side chains (PLTEG). The enantioselectivities for chiral α-amino acids (tryptophan, phenylalanine and tyrosine) and chiral drugs (propranolol, atenolol, and ibuprofen) were measured by concentration-driven experiments, and were found to range from 1.04 to 1.47.35 The selectivity increased with the helical content of PLGA immobilized on PVDF membranes. The enantioselectivity was observed to be higher for chemically grafted polypeptide-modified PVDF membranes compared to polypeptidephysisorbed PVDF membranes. This difference is attributed to the higher molecular weight and density of the polypeptide chains, which enhance the interaction between the chiral compounds and the surface-bound polypeptides.35 Gumi et al. prepared chiral separation membranes from polysulfone grafted with N-dodecyl-4(R)-hydroxy-L-proline as a chiral selector.36 The membranes prepared from the chiral derivatized polysulfone showed a separation factor of 1.1 in the dialysis permeation of racemic propranol, where S-propranol preferentially permeated through the membranes.36 Aoki et al. prepared chiral separation membranes from poly[p-(oligopinanylsiloxanyl) phenylacetylene]s with a chiral helical main chain. Membranes with a high chiral helicity in the main chain showed good enantioselectivity for racemic phenylalanine, tryptophan, valine, and 2-phenethyl alcohol based on diffusion selectivity, while there was almost no sorption selectivity.37 Chiral separation of racemic amino acids through polymeric membranes with saccharide side chains was also reported. Satoh prepared chiral separation membranes of polyacrylonitrile-graft-(1->6)-2,5-anhydro-3,4-di-O-methyl-D-glucitol.38 The permeation rates of the amino acids increased in the order of phenylglycine < phenylalnine < tryptophan, according to the molecular size of the permeating solutes. The permeation rate of the D-isomer was found to be higher than that of the L-isomer for the amino acids that were evaluated, although the L-isomer was more highly adsorbed than the D-isomer. Therefore, the chiral separation by these membranes was caused by diffusion selectivity.

122 Tryptophan, tyrosine, serine Tryptophan Phenylalanine Tyrosine Propranol Atenolol Ibuprofen Glutamic acid

Tryptophan Phenylalanine 1,3-butanediol 2-butanol Tryptophan (D) Phenylalanine (D) 2-BuOH 2-BuOH 1,3-Butanediol Phenylglycine perchlorate Phenylalanine perchlorate

Targeted molecule

(D) P = 1.95 × 10−7 cm2/s (D) P = 1.35 × 10−7 cm2/s (D) P = 1.90 × 10−7 cm2/s (D) P = 1.87 × 10−7 cm2/s (D) P = 1.56 × 10−7 cm2/s (D) P = 2.39 × 10−7 cm2/s (D)

P = 1.34 × 10−14 m2/h (D) P = 1.11 × 10−14 m2/h (D) P = 1.19 × 10−3 g/mh (EV) P = 8.37 × 10−4 g/mh (EV) P = 4.72 × 10−6 gm/m2h (D) P = 1.65 × 10−6 gm/m2h (D) P = 837 × 10−6 gm/m2h (D) P = 8.37 × 10−4 gm/m2h (PV) P = 1.19 × 10−4 gm/m2h (PV) P = 4.91 × 10−7 cm2/min (D) P = 4.01 × 10−7 cm2/min (D)

Flux or permeability coefficient (methodb)

1.5 1.3 1.1 1.2 1.1 1.2 1.1

2.9 640.0 2.4 2.6 2.2 3.4 2.6 2.6 2.4 1.2 1.1 1.1

Separation factor

37 37 25 25 78 78 78 25 25 38 38 38 79 35 35 35 35 35 35 36

Ref.

a Poly(DPSPP); Poly[(−)-1-p-[dimethyl(10-pinanyl)silyl]phenyl-2-phenylacetylene], Poly(DPSP); Poly{1-dimethyl(10-pinanyl)sily]-1-propyne}, PANg-D-glucitol; Polyacrylonitrile-graft-(1->6)-2,5-anhydro-D-glucitol, PMLG; Poly(L-glutamate), PSf; Polysulfone. b Method; D indicates dialysis, ED indicates electro dialysis, PV indicates pervaporation, UF indicates ultrafiltration.

Poly(DPSPP) Poly(DPSPP) Poly(DPSPP) Poly(DPSPP) Poly(DPSP) Poly(DPSP) Poly(DPSP) Poly(DPSP) Poly(DPSP) PAN-g-D-glucitol PAN-g-D-glucitol PAN PMLG derivatives PMLG derivatives PMLG derivatives PMLG derivatives PMLG derivatives PMLG derivatives PMLG derivatives PSf having myrtenal-derived terpenoid

Chiral sites of membranesa

Table 3 Chiral separation through polymeric membranes having chiral side chain

Membranes for Chiral Separation

123

6. Chiral Separation Membranes with Immobilized Stereoselective Ligands as Chiral Selectors and Recognition Sites Several groups have used specific interactions between large biological molecules and specific pharmaceuticals or drugs for chiral separation. Cyclodextrins39–45 and crown ether derivatives24 (see Fig. 2) are commonly used as stereoselective ligands in chiral separation membranes. The idea of using affinity binding of specific enantiomers by large biological macromolecules such as proteins and DNA in ultrafiltration dates back to the early work by Higuchi et al.46 on the purification of a tryptophan isomer by exploiting its binding to albumin on the membrane. Chiral separation using membranes with immobilized large molecules as chiral selectors can work by three mechanisms: (1) affinity membranes, (2) selective sorption membranes, and (3) selective diffusion membranes. The chiral separation mechanism of affinity membranes is based on the selective adsorption of specific isomers compared to the other isomers. Since these membranes cannot obtain steady state chiral separation, chiral separation work using affinity membranes was not included in this review. Several examples of chiral separation membranes with immobilized stereoselective ligands as chiral selectors and recognition sites based on selective sorption and/or diffusion membranes are summarized in Tables 4 and 5, and described in detail in the following sections. 6.1 Immobilized Cyclodextrin Membranes Native cyclodextrins (CD) are cyclic oligosaccharides consisting of six to eight D-(+)glucopyranose units that provide three-point interactions for the chiral recognition of various organic molecules by hydrophobic interaction with the CD cavity and two hydrogen bonds

Figure 2. Chemical scheme of alpha-cyclodextrin (a), beta-cyclodextrin (b), gamma-cyclodextrin (c), 18-crown-6 (d), dibenzo-18-crown-6 (e), and diaza-18-crown-6 (f) used as stereoselective ligands in chiral separation membranes.

124 Flux or permeability coefficient (methoda) (ED) P = 2.94 × 10−7 cm2s−1 (D) P = 0.249 × 10−7 cm2s−1 (D) P = 2.70 × 10−7 cm2s−1 (D) (D) P = 1.53 × 10−7 cm2/s (D) P = 1.66 × 10−7 cm2/s (D) P = 0.029 mg/cm2h (D) J = 6.7 × 10−4 mol/m2h (D) (D) (D) P = 1.75 × 10−8 m/s (D)

Targeted molecule D-4-hydroxyphenylglycine Phenylalanine Tryptophan Histidine Chlorthalidone Phenylalanine Tryptophan Tryptophan Lactic acid Propranolol Propranolol Propranolol

Method; D indicates dialysis,. ED indicates electro dialysis, UF indicates ultrafiltration.

β-cyclodextrin β-cyclodextrin β-cyclodextrin β-cyclodextrin β-cyclodextrin β-cyclodextrin β-cyclodextrin β-cyclodextrin N-3,5-dinitrobenzoyl-L-alanine-octylester N-hexadecyl-L-hydroxyproline L-di-n-dodecyltartrate N-dodecyl-4(R)-hydroxy-L-proline

Chiral sites of materials in membranes

Table 4 Chiral separation through polymeric membranes having small chiral molecules

1.3 1.4 1.3 1.2 1.2 1.3 1.1 1.5 15.9 1.0 1.0 1.1

Separation factor

80 41 41 41 39 23 42 40 81 82 82 36

Ref.

Membranes for Chiral Separation

125

Table 5 Chiral separation through polymeric membranes having large chiral molecules Chiral sites of materials in membranesa Crosslinked BSA Crosslinked BSA Crosslinked BSA Crosslinked BSA DNA-cellulose DNA-cellulose DNA-cellulose DNA-cellulose DNA DNA DNA-chitosan DNA-chitosan D-amino acid oxidase apoenzyme a

Targeted molecule

Flux or permeability coefficient (methodb)

Separation factor

Ref.

Phenylalanine Leucine Tryptophan Tryptophan Tryptophan Tryptophan Phenylalanine Phenylalanine Phenylglycine Phenylalanine Phenylalanine Phenylalanine Phenylalanine

1.5 m3/m2 day kg (UF) 1.5 m3/m2 day kg (UF) 1.5 m3/m2 day kg (UF) (UF) 6.5 × 10−3 m3/m2 day kg (UF) (UF) 6.5 × 10−3 m3/m2 day kg (UF) 6.5 × 10−3 m3/m2 day kg (UF) 6.5 × 10−3 m3/m2 day kg (UF) (UF) 1.3 × 10−3 m3/m2 day kg (UF) 1.7 × 10−3 m3/m2 day kg (UF) 2.2 × 10−5 cm2/sec (D)

1.3 1.1 1.4 8.7 1.2 1.3 1.6 1.6 1.1 1.2 2.7 1.8 3.3

48 48 48 83 2 2 2 55 2 51 56 56 49

BSA; bovine serum albumin. Method; D indicates dialysis, UF indicates ultrafiltration.

b

with the hydroxyl groups at the opening of the CD.43 Xiao and Chung prepared immobilized β-cyclodextrin (CD) membranes using commercially available cellulose (CA) dialysis membranes with a molecular weight cutoff (MWCO) of 1000 and investigated the chiral separation of a racemic tryptophan solution through the immobilized CD membranes.42 Their dialysis transport experiments showed that D-tryptophan preferentially permeated through the immobilized CD membranes, obtaining an enantioselectivity of around 1.10. Note that the L-tryptophan binds CD at a higher affinity (binding constant of 0.043 mM−1 in CD solution) than D-tryptophan (0.031 mM−1 in CD solution).42 This means that the chiral separation of tryptophan is based on diffusion selectivity in the immobilized CD membranes.42 Compared with other chiral selector-immobilized membranes, CD-functionalized membranes have a lower cost and might have wider applicability and higher tolerance in various environments. However, chiral separation through immobilized CD membranes has the disadvantage of low selectivity because native cyclodextrins have limited chiral recognition ability and limited flexibility, which are important to enable interaction with the enantiomers. Chemical modification of CD and the preparation of CD derivatives might be an interesting topic of research, and chiral separation can be performed using immobilized CD-derivative membranes. Xiao et al. prepared acetylated-β-cyclodextrinimmobilized CA dialysis membranes for chiral separation of racemic tryptophan.23 The acetylated CD-immobilized membrane exhibits enantioselectivity in the range of 1.26–1.33 depending on the acetylation time, while native CD-immobilized membranes only show an enantioselectivity of 1.11. The improvement in enantioselectivity after acetylation was mainly attributed to the improved discrimination ability of acetylated CD and the decrease

126

A. Higuchi et al.

in membrane pore size.23 From molecular modeling simulations, the acetylation of hydroxyl groups was suggested to result in a CD conformation with toroidal distortions, creating a greater steric hindrance for phenylalanine interaction. As a result, compared to the original CD, the acetylated CD has less effective binding but better discrimination of enantiomers. Molecular modeling simulations also indicated that the energy drop was only 3 kcal/mol between different enantiomers before and after the binding of phenylalanine with an unmodified CD, while the energy drop increased to 10 kcal/mol when the acetylated CD was employed as the chiral selector, which contributed to the higher recognition of L-phenylalanine as a chiral selector.23 Ishihara prepared a copolymer of acrylnitrile (PAN) and aminoethylated β-CD and used it to make chiral separation membranes.41 The membranes showed preferential permeation of L-phenylalanine over D-phenylalanine due to diffusion selectivity with a separation factor of 1.40, while the membranes showed preferential sorption of D-phenylalanine over L-phenylalanine, the opposite tendency to that of native CD and acetylated CD. Therefore, these results suggest that the selective sorption of CD to specific enantiomers such as the R-isomer or L-isomer can change depending on the chemical modification of CD. The appropriate chemical modification of CD will likely be important for the development of chiral separation membranes with immobilized CD derivatives. 6.2 Immobilized Albumin Membranes Serum albumin is reported to have a high-affinity binding site for L-tryptophan. Its binding constant was reported to be 4.4 × 104 M−1 by Kragh-Hansen,47 and weakly bound Dtryptophan has been reported to displace L-tryptophan.48 This evidence prompted Higuchi et al. to develop a method for chiral separation of racemic amino acids using ultrafiltration of solutions of various racemic amino acids through immobilized serum albumin membranes, making use of the binding site of bovine serum albumin (BSA) to the L-isomer.46,48 These authors reported that the immobilized BSA membranes efficiently demonstrated chiral separation of not only racemic tryptophan but also leucine and phenylalanine.48 The target molecule for chiral separation through the immobilized albumin membranes should be amino acids with aromatic or hydrophobic side groups because of the specific binding site characteristics of albumin. The chiral separation of low molecular weight pharmaceuticals such as ibuprofen using the immobilized BSA membranes should be an interesting research topic in the future. 6.3 Immobilized Apoenzyme Membranes Enzymes are well-known to have high substrate selectivity, understood with the lock and key model. Enzymes could be a good selection for the recognition of chiral molecules in chiral separation membranes. However, enzymes not only recognize the specific enantiomer, but also catalyze a chemical reaction on the molecule. Since apoenzyme requires a cofactor to perform the enzymatic reaction, unwanted chemical conversion of the substrate molecule in enzyme-based separation can be circumvented when using this molecular-recognition agent. Lakshmi and Martin investigated enantioseparation using apoenzymes immobilized in porous polymeric membranes.49 They found that the membranes selectively transport the specific substrate molecule without the unwanted chemical conversion of the molecule. When D-amino acid oxidase apoenzyme was loaded in the membranes, facilitated transport of D-phenylalanine relative to L-phenylalanine was observed, with the maximum D- versus L-penylalanine selectivity coefficient reported to be from 3.3 to 4.9.49

Membranes for Chiral Separation

127

6.4 Immobilized Antibody Membranes Antibodies can specifically bind to a variety of targets, known as antigens, depending on the variants of antibodies. The interaction between the antibody and antigen is extremely high, and allows the antibody to identify and bind only their unique antigen in the midst of the millions of different molecules from an induced fit binding.50 Recognition of the antigen by the antibody could be available for the recognition of chiral molecules in chiral separation membranes. However, the binding constants for the antibody are generally too large to bind with the specific antigen reversibly, which is undesirable because the chiral separation membrane must ultimately release the target molecules so that they can be collected in the permeate solution. Lee et al.50 solved this problem by tuning the binding constant of the antibody with the addition of dimethyl sulfoxide (DMSO) to the racemic feed solution and the permeate solution. They prepared chiral separation membranes immobilized antibody for 4-[3-(4fluorophenyl)-2-hydroxy-1-[1,2,4]triazol-1-yl-propyl]-benzonitrile, which is a drug serving as an inhibitor of aromatase enzyme activity.50 This molecule has two chiral centers and thus four stereoisomers: RR, SS, SR, and RS. The antibody used selectively binds the RS relative to the SR enantiomer, and the membranes based on the Fab fragment of this antibody were used to separate this enantiomeric pair. Nanopore alumina membranes were used as host membranes for immobilization of the anti-RS molecules. The membranes having pores with diameters of 20 and 35 nm were used for these studies.50 A sol-gel template synthesis method was used to deposit silica nanotubes within the pores of the alumina membranes. The inside walls of the silica nanotubes were then reacted with a silane that terminated in an aldehyde functional group, which reacts spontaneously with free amino sites on the antibody they used.50 An average separation factor of 2.0 was obtained for the membranes prepared from the alumina membranes with a pore diameter of 35 nm, when both the RS and SR enantiomers were dissolved in 10% DMSO-phosphate buffer saline (PBS) buffer at pH 8.5 as a feed solution. When the effect of the concentration of the enantiomers in the racemic feed solution on the flux was investigated, they observed a Langmuirian-shaped flux curve for the RS enantiomer but not for the SR enantiomer, which suggests a facilitated transport of the RS enantiomer.50 The selectivity in a facilitated transport process can be increased by shutting down the nonfacilitated (diffusional) transport of the unwanted chemical species. This can be accomplished by decreasing the pore size in the membrane in porous membranes. When Lee et al. used alumina membranes having pores 20 nm in diameter as base membranes for the immobilization of the antibody, the separation factor was reported to be increased to 4.5.50 The porous polymeric membranes immobilized antibodies could be one of the candidates for the chiral separation membranes immobilized proteins. 6.5 Immobilized DNA Membranes DNA has been discovered to have several novel functions aside from carrying genetic information, such as electron transfer and DNA enzymatic activity.51 DNA can also intercalate some enantiomers with a binding constant that depends on the stereoenantiomer.52 DNA is contained in several common protein preparations as an impurity on the order of ppb.53 Therefore, Higuchi et al. investigated the effect of the DNA in the albumin solution on the chiral separation of amino acids by ultrafiltration in 2002.54 These results generated the idea that DNA has a binding site for enantiomers, with the binding constant

128

A. Higuchi et al.

depending on the stereoenantiomer. Higuchi et al. established that DNA can be used as a biomacromolecular chiral selector in chiral separation technology.52–56 When affinity ultrafiltration experiments were performed for the chiral separation of racemic phenylalanine using DNA as chiral selectors in the feed solution with a DNA concentration less than 0.5 ppm, D(R)-phenylalanine was preferentially present in the collected permeate solution, although the separation factor fluctuated from 0.5 to 20 depending on the permeation time. This fluctuation was explained by the fact that DNA sometimes releases phenylalanine in a dilute DNA solution (i.e., 0.01–0.5 ppm) due to the conformational change of DNA over time.52 Higuchi et al. prepared immobilized DNA membranes from cellulose dialysis membranes with different pore sizes and investigated the effect of the pore size on chiral separation through immobilized DNA membranes.52,54,55 They found that D-phenylalanine preferentially permeated through the immobilized DNA membranes with pore sizes <2.0 nm (MWCO <5000), while L(S)-phenylalanine preferentially permeated through the immobilized DNA membranes with a pore size >2.0 nm (MWCO of the base membranes >5000) as shown in Fig. 3.55 The pore size of the immobilized DNA membranes regulated preferential permeation of the stereoenantiomer through the membranes. The immobilized DNA membranes adsorbed L-phenylalanine preferentially, independent of the pore size.52 Furthermore, in a concentration gradient and electric field, DNA was able to permeate through membranes with a pore size of 5 nm but did not permeate through membranes with a pore size of 1.5 nm. Therefore, in membranes with a pore size <2 nm, it was estimated that DNA was immobilized on the surface of the membranes and not inside the pores, while in membranes with a pore size >2 nm, DNA was expected to be immobilized on the surfaces as well as inside the pores. Considering the above results, Fig. 4 shows the model of chiral separation by the immobilized DNA membranes.55

Figure 3. Dependence of the separation factor in the permeate (open circle) and the concentrate (closed circle) solutions through the immobilized DNA membranes on the MWCO of the base membranes and on the pore size of the immobilized DNA membranes at pH 7.0 and 25◦ C.55

Membranes for Chiral Separation

129

Figure 4. Chiral separation model in the ultrafiltration of racemic phenylalanine solution through the immobilized DNA membranes.2

Zhang et al. also reported that DNA can be used as a chiral selector for the chiral separation of racemic tyrosine and tryptophan by affinity ultrafiltration through dialysis membranes with MWCO = 12,000–14,000.57 They reported that D-enantiomers were preferentially present in the permeate solution while L-enantiomers were preferentially present in the feed solution.57 Higuchi and other researchers55,57 showed that the immobilized DNA membranes are potentially useful for chiral separation. Chiral separation by these immobilized DNA membranes is based on the interaction between DNA and a specific stereoenantiomer because no chiral separation was found in the cellulose membranes without bound DNA. The immobilized DNA membranes were categorized as channel-type membranes and not as affinity membranes as for affinity ultrafiltration using albumin.46,48,58 The membrane pore size and the binding affinity of the specific stereoenantiomer should be the most important factors for the preparation of channel-type membranes. The immobilized albumin membranes reported in the literature46,48,58 did not work as channel-type membranes but as affinity membranes, although the membranes have a similar pore size (e.g., MWCO = 13,000) to the membranes used in DNA immobilized membranes. This is due to the strong binding affinity of L-amino acids to albumin. When the binding affinity of the specific stereoenantiomer to the membranes is too strong, the specific enantiomer cannot permeate through the membrane, adsorbing instead. The weaker binding affinity of DNA to L-amino acids compared to that of albumin makes the immobilized DNA membranes channel-type membranes.55

130

A. Higuchi et al.

7. Chiral Separation by Non-Chiral Membranes The central dogma that chirality can only be generated from chiral molecules could be extended to predict that membranes without any chiral molecules cannot contribute to chiral separation. Therefore, most researchers have developed chiral separation membranes prepared from natural chiral polymers, synthetic chiral polymers from asymmetric synthesis, polymers with chiral branches, or immobilized chiral molecules, proteins, polysaccharides, and DNA. However, chiral separation membranes can be prepared from non-chiral polymers with “chiral memory,” accomplished with molecularly imprinted membranes or membranes with a chiral memory of the polymer helicity. Some examples of those chiral separation membranes are sumarized in Tables 6 and 7, and described in detail in the following sections. 7.1 Molecularly Imprinting Membranes A molecularly imprinted polymeric membrane is designed to mimic the recognition site of an enzyme with its shape, formed by interactions with a “template” target molecule. There are two basic methods of preparing molecularly imprinted membranes—covalent and noncovalent molecular imprinting methods. In both cases, the template molecules are chosen to allow interactions with the functional group of the imprinted polymeric membranes. In the covalent method of molecular imprinting, the imprint (template) molecule is covalently coupled to a monomer during polymerization.59,60 The imprint molecule is Table 6 Chiral separation through polymeric membranes having chiral free volume shape (molecularly imprinted membranes) Membrane materials DIDE resin Polyamide DIDE resin EEE derivatives Carboxylated polysulfone Carboxylated polysulfone Polysulfone derivatives E2 K resin (DE)2 K resin (IDE)2 K resin (DIDE)2 K resin Chitosan/GPTMS hybrid Chitosan/GPTMS hybrid a

Targeted molecules

Flux or permeability coefficient (methoda)

Tryptophan Serine Ac-tryptophan Ac-tryptophan Glutamic acid (D)

P = 0.74 × 10−4 cm2/h (D) (UF) P = 3.33 × 10−7 cm2/sec (ED) P = 6.11 × 10−5 cm2/sec (ED) P = 5.86 × 10−5 cm/sec (D)

Separation factor Ref. 1.4 9.0 4.6 5.0 1.1

61 63 84 85 86

Glutamic acid (L) P = 5.94 × 10−5 cm/sec (D)

1.2

86

Ac-glutamic acid

(ED)

1.6

87

Ac-tryptophan Ac-tryptophan Ac-tryptophan Ac-tryptophan Phenylalanine

(D) (D) (D) (D) (D)

1.5 1.7 2.1 2.5 4.5

88 88 88 88 64

Phenylalanine

(D)

2.1

64

Method; D indicates dialysis,. ED indicates electro dialysis, UF indicates ultrafiltration.

Membranes for Chiral Separation

131

Table 7 Chiral separation through polymeric membranes having chiral free volume shape (chiral helical memory membranes) Membrane materialsa Poly(diphenylacetylene) Poly(diphenylacetylene) Poly(diphenylacetylene) Copoly(HPA) Copoly(HPA) Poly(HPA) Poly(HPA) Poly(phenylacetylene) a

Targeted molecules

Flux or permeability coefficient (methodb)

Tryptophan Tryptophan Tryptophan Phenylalanine Phenylalanine Phenylalanine Phenylalanine Phenylalanine

P = 22.5 × 10−12 m2/h (N) P = 19.7 × 10−12 m2/h (N) P = 13.4 × 10−12 m2/h (N) P = 12.6 × 10−14 m2/h (D) P = 13.1 × 10−14 m2/h (D) P = 11.0 × 10−14 m2/h (D) P = 10.7 × 10−14 m2/h (D) P = 9.6 × 10−14 m2/h (D)

Separation factor Ref. 1.2 1.4 2.9 1.3 1.4 1.2 1.1 1.5

66 66 66 67 67 67 67 67

Copoly(HPA); copoly(hydroxylphenylacetylene), Poly(HPA); Poly(hydroxylphenylacetylene). Method; N indicates nanofiltration, and D indicates dialysis.

b

chemically cleaved from the highly crosslinked polymer after copolymerization with a crosslinker. In the non-covalent method of molecular imprinting, the imprint (template) molecules are mixed with functional monomers capable of interacting non-covalently with the imprint molecules. The functional monomers are copolymerized with crosslinkers to generate a highly crosslinked and rigid polymer. The imprint molecules are subsequently removed from the polymer, leaving recognition sites complementary to the imprint species in the shape and positioning of functional groups.59 Preparation of the polymer in this way induces molecular memory, as the recognition sites are capable of selectively recognizing the imprint species. During both the imprinting procedure and the rebinding, the imprint molecules interact with the polymer via non-covalent interactions, e.g., ionic, hydrophobic, and hydrogen bonding interactions. Yoshikawa et al. applied this molecular imprinting method to develop chiral separation membranes by the non-covalent bonding method in 1995.61 Their work was the first report on chiral separation membranes prepared by a molecular imprinting method. They prepared polystyrene copolymer grafted tetrapeptide derivatives, and the molecular imprinting membranes were prepared from the polystyrene copolymer with template molecules of Boc-L-tryptophan or Boc-D-tryptophan. The imprinting molecules were extracted from the imprinting membranes by methanol. L-tryptophan permeated through the membranes imprinted with Boc-L-tryptophan. However, chiral separation using molecularly imprinted membranes has so far suffered from a relatively low selectivity. Itou and Yoshikawa et al. prepared molecularly imprinted polymeric membranes from polystyrene resin bearing a tetrapeptide of glycine (G-membranes).62 N-αtert-Butoxycarbonyl-D-tryptophan (Boc-D-Trp) or N-α-tert-butoxycarbonyl-L-tryptophan (Boc-L-Trp) were used as the template molecules to convert G-membranes into chiral separation membranes. Because the constitutional residue of the tetrapeptide, glycine, had no asymmetric carbon, both template molecules, Boc-D-Trp and Boc-L-Trp, worked as template molecules to construct chiral recognition sites within the membranes. The membrane imprinted by the D-isomer of the template molecule recognized the D-isomer of N-α-acethyltryptophan (Ac-D-Trp), while recognition sites for N-α-acethyl-L-tryptophan (Ac-L-Trp) were constructed by imprinting with Boc-L-Trp.62 Membrane separation of the racemic mixtures was investigated by adopting an applied electrical potential difference as

132

A. Higuchi et al.

a driving force for membrane transport. The two G-membranes imprinted as above showed chiral separation ability in this format as well. Permeation selectivity was reported to be ca. 2.3 at the optimum applied electrical potential difference of 2.0 V, and was based on adsorption selectivity.62 Son and Jegal prepared molecularly imprinted membranes by interfacial polymerization with a template (D-serine) of piperazine and trimesoyl chloride on polysulfone microfiltration support membranes.63 The D-serine template was removed by keeping the membrane in 50◦ C water for 50 h after interfacial polymerization to generate an active crosslinked polyamide layer. The three -COCl groups of trimesoyl chloride reacted with the two amine groups of the piperazine. Hydrogen bonding is expected to occur between the amine of piperazine and the -COOH of D-serine in the molecular cavity of the molecularly imprinted membranes. When a racemic serine solution was used for chiral resolution by pressure-driven permeation, the permeation rate of D-serine appeared to be much faster than that of L-serine. When the operating time reached 60 h, the enantiomeric excess (%ee) of the serine mixture in permeates was reported to be about 80%.63 Jiang and Wu reported on enantioselective chitosan (CS)/γ -glycidoxypropyltrime thoxysilane (GPTMS) hybrid membranes prepared in an aqueous phase by a sol–gel process using chitosan as the bulk polymer, L-phenylalanine as the imprinting molecule, and GPTMS as the crosslinking agent.64 The separation factor α D/L was reported to be as high as 4.5. These organic–inorganic hybrid CS/GPTMS materials represent promising materials for the development of sol–gel imprinting with effective crosslinking and a lower degree of swelling, and they exhibit good separation properties. The introduction of silica into the chitosan created a dense and uniform hybrid network and reduced the degree of swelling of the materials in an aqueous system, which ensured the formation and maintenance of imprinting sites. The specific interactions of imprinting cavities in the hybrid membrane with templates resulted in significantly improved chiral resolution of the imprinted membranes by improving the binding ability of the imprinting molecules, hindering their diffusion and facilitating transport of the other isomers.64 7.2 Membranes Composed of an Achiral Polymer with a One-Handed Helical Conformation The construction of controlled secondary structures in polymers has attracted a great deal of attention in recent years, with one-handed helical polymers as a typical example. The preparation of one-handed helical polymers has usually been achieved by one of the following methods: (a) polymerization of optically active monomers, (b) polymerization of prochiral monomers using optically active catalysts or initiators, (c) induction of chirality by interaction between achiral polymers and chiral additives, and (d) induction of chirality in achiral polymers with a chiral functional group followed by removal of the chiral functional group.65 Method (d) enables the construction of chiral membranes from achiral polymers. Such a “chiral memory” of the polymer helicity was first found by Yashima’s group, but they did not apply this technique for chiral separation technology.66 Teraguchi and Masuda verified enantioselective permeation through helical polymeric membranes composed of achiral poly(diphenylacetylene).65 The chiral memory of the membranes was created using poly(diphenylacetylenes) with a high content of the pinanylsilyl group, generating a chiral polymer, followed by depinanylsilylation after preparation of self-stable membranes. Depinanylsilylation of the membranes was carried out by exposure of the membranes to a mixture of hexane/trifluoroacetic acid at room temperature.65 The chiral selectivity of racemic tryptophan solutions through the membranes was found to

Membranes for Chiral Separation

133

be as high as 58.6%(ee), while that of the chiral poly(diphenylacetylenes) retaining the pinanylsilyl group was 80.5%(ee). These studies verify that membranes composed of achiral polymers with chiral helical memory can contribute to the chiral separation of chemicals and drugs.65 Teraguchi and Aoki et al. also prepared chiral helical poly(hydroxyl-containing phenylacetylene) and poly(phenylacetylene) with and without the chiral pinanylsiloxy group. First, they prepared poly(pinanylsiloxyl-containing) phenylacetyle membranes.67 Poly(hydroxyl-containing phenylacethylene) or poly(phenylacetylene) membranes made from an achiral polymer were then prepared by depinanylsilylation of the membranes in situ.67 The resulting membranes exhibited circular dichroism despite the absence of the chiral substituents (pinanylsiloxy group), indicating that the main polymer chains retained their chiral helicity (chiral memory in the membranes). The membrane prepared from a polymer with chiral side chains (i.e., chiral helical poly(phenylacetylene) containing the chiral pinanylsiloxy group) showed excellent chiral separation of phenylalanine in concentration-driven permeation, exhibiting 34.8–77.4 ee% depending on the chemical structure of the polymeric membranes.67 However, depinanylsiloxyed membranes (poly[hydroxyl-containing phenylacetylene] and poly[phenylacetylene]) also showed chiral separation of phenylalanine in concentration-driven permeation, with 6.05–21.1 ee% depending on the chemical structure of the polymeric membranes. Although the polymer used to prepare these membranes has no chirality, it retains its helical conformation due to the preparation method.67 Their study demonstrated that the chiral main chain is also important for chiral separation using membranes.

8. Comparison of Data Reported by Different Researchers There is an inherent trade-off, as the separation factor generally decreases with increasing permeability of solutes or gases through the membranes.68–70 This trade-off should be considered for the chiral separation of drugs and pharmaceuticals through polymeric membranes. The available data on permeability and separation factors of tryptophan and phenylalanine through chiral separation membranes was collected from the literature.2,23,25–31,34–42,48,49,51,55,56,61,63,64,66,67,71–88 The empirical upper bound relationship for chiral separation of racemic tryptophan and phenylalanine is shown in Figs. 5 and 6. The upper bound correlation is that the log of the permeability of amino acids versus the log of the higher separation factor yielded a limit for achieving the desired result of a high separation factor combined with high permeability. The upper bound correlations for the chiral separation of racemic phenyalanine and tryptophan through polymeric membranes under concentration-driven permeation (i.e., the dialysis method) were described as α = −0.333 − 0.333 ∗ log(P /cm2 sec−1 ) −1

α = −1.08 − 0.4167 ∗ log(P /cm sec ) 2

for racemic phenylalanine;

(13)

for racemic tryptophan.

(14)

This research did not clearly indicate an upper bound relationship in chiral separation through membranes, especially for the chiral separation of phenylalanine, while the upper bound relationship clearly follows P = kα n in gas separation membranes and processes,68 where P is the permeability of the fast gas, α is the separation factor, k is referred to as the “front factor,” and n is the slope of the log–log plot of the P vs kα n. Figures 5 and 6 will contribute to the evaluation of chiral separation properties for the development of improved chiral separation membranes.

134

A. Higuchi et al.

Figure 5. Relationship between permeability coefficients and separation factor of phenylalanine through several chiral polymeric membranes concentration-driven permeation. Data were cited from references (see Table 1S in Appendix A).

10. Conclusions and Future Perspectives More than 100 articles have been published on the development of polymeric membranes for chiral separation. Most chiral separation membranes have relatively low separation factors, except for affinity membranes. One of the solutions to this problem is to use a

Figure 6. Relationship between permeability coefficients and separation factor of tryptophan through several chiral polymeric membranes concentration-driven permeation (◦) and potential-driven permeation (•). Data were cited from references (see Table 2S in Appendix A).

Membranes for Chiral Separation

135

Figure 7. Flow diagram of a multi-stage ultrafiltration process using two types of membranes for chiral separation.27

multistage chiral separation process. Higuchi et al. proposed a multi-stage cascade in the ultrafiltration process through channel-type membranes for chiral separation in 2003.55 When two different types of membranes are used in a cascade filtration model (Fig. 7), there is no change in the concentration ratio between the R-enantiomer and S-enantiomer in the feed solution in each stage because two membranes with opposite enantiomeric selectivity are used. When there are n stages of the multi-stage process using two types of chiral membranes, the purity of the enantiomer (ee) from the racemic amino acid mixture can be estimated.55 Figure 8 shows some examples calculated with the following membrane conditions: (a) α = 1.2 and 1/1.2, (b) α = 1.6 and 1/1.6, (c) α = 2 and 0.5, and (d) α = 4 and 0.25.55 Only four stages are necessary to obtain 99% purity when membranes with α = 4 and 0.25 are used in the multi-stage process. The biggest advantage of the multi-stage process using two types of chiral membranes is the theoretical 100% product recovery.55 Several researchers have investigated chiral separation by affinity ultrafiltration using albumin as a large stereo-specific binding agent.46,48,89–92 Albumin has several chiral recognition sites for amino acids and small drugs.47 The stability and high cost of these proteins make it difficult to develop a large-scale commercial process for the chiral separation of pharmaceuticals by affinity ultrafiltration using albumin. DNA was recently discovered to have several chiral recognition sites for specific enantiomers.52–55 DNA is much more stable than proteins and is less expensive than albumin when using DNA isolated from salmon testis. The separation factors of immobilized DNA membranes and immobilized albumin membranes were both acceptable, although DNA seems to be a more promising stereo-specific binding agent than BSA or HSA. Several polymeric membranes were developed from natural chiral polymers (i.e., polyamino acids and polysaccharides) and synthetic polymers with a chiral main backbone

136

A. Higuchi et al.

Figure 8. Dependence of chiral purity (ee%) of pharmaceutical enantiomers on the number of filtrations in the multi-stage ultrafiltration process using chiral separation membranes of α = 4 and 0.25 (a), α = 2 and 0.5 (b), α = 1.6 and 1/1.6 (c), and α = 1.2 and 1/1.2 (d) from top to bottom lines.55

or chiral side chains. Molecularly imprinted membranes were also prepared from achiral monomers and/or polymers. Currently, there is no optimized membrane for the chiral separation of pharmaceuticals and chemicals because of the low separation factors and/or low permeabilities of these membranes (flux). New, intelligent designs of chiral polymeric membranes are necessary to enable the development of chiral separation membranes that can be applied to industrial pharmaceutical production. In conclusion, advanced polymeric materials are playing an important role in the development of chiral separation membranes for pharmaceutical applications.

Acknowledgments This research was supported by grants from the National Science Council of Taiwan under Grants No. 97-2221-E-008-011-MY3 and NSC97-2120-M-008-002, the VGHUST Joint Research Program, Tsou’s Foundation (98DFA0700006), and the Cathay General Hospital Project (98CGH-NCU-B1). Grants-in-Aid for Scientific Research (No. 21500436) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan are also acknowledged.

References 1. Rikken, G. L. A.; Raupach, E. “Enantioselective magnetochiral photochemistry,” Nature, 2000, 405, 932–935. 2. Higuchi, A.; Hayashi, A.; Kanda, N.; Sanui, K.; Kitamura, H. “Chiral separation of amino acids in ultrafiltration through DNA-immobilized cellulose membranes,” J. Molecular Structure, 2005, 739(1–3), 145–152. 3. Romero, J.; Zydney, A. L. “Chiral separations using ultrafiltration with a stereoselective binding agent,” Sep. Sci. Tech., 2001, 36(7), 1575–1594. 4. Stinson, S. “Counting on chiral drugs,” Chem. Eng. News, 2001, 79(40), 79–97. 5. Romero, J.; Zydney, A. L. “Staging of affinity ultrafiltration processes for chiral separations,” J Membrane Sci., 2002, 209(1), 107–119. 6. Hutt, A. J. “Drug chirality: Impact on pharmaceutical regulation,” Chirality, 1991, 3, 161–164.

Membranes for Chiral Separation

137

7. Cayen, M. N. “Racemic mixtures and single stereoisomers: Industrial concerns and issues in drug development,” Chirality, 1991, 3, 94–98. 8. Shafaati, A. “Chiral drugs: Current status of the industry and the market,” Iranian Journal of Pharmaceutical Research, 2007, 6(2), 73–74. 9. Nohira, H.; Watanabe, K.; Kurokawa, M. “Optical resolution of N-benzylcis-2-aminocyclohexanecaroxylic acid by preferential crystallization,” Chem. Lett., 1976, 5(4), 299–300. 10. Jacques, J.; Collet, A.; Wilen, S. Enantiomers, Racemates and Resolutions; Wiley-Interscience: New York, 1981. 11. Mallik, R.; Xuan, H.; Hage, D. S. “Development of an affinity silica monolith containing alpha(1)-acid glycoprotein for chiral separations,” J. Chromatography A, 2007, 1149(2), 294–304. 12. Ljungberg, H.; Nilsson, S. “Protein-based capillary affinity gel-electrophosresis for chiral separation of beta-adrenergic blockers,” J. Liquid Chromatography, 1995, 18(18–19), 3685–3698. 13. Nakamura, M.; Kiyohara, S.; Saito, K.; Sugita, K.; Sugo, T. “Chiral separation of DL-tryptophan using porous membranes containing multilayered bovine serum albumin crosslinked with glutaraldehyde,” J. Chromatography A, 1998, 822, 53–58. 14. Millot, M. C. “Separation of drug enantiomers by liquid chromatography and capillary electrophoresis, using immobilized proteins as chiral selectors,” J. Chromatography B, 2003, 797, 131–159. 15. Maier, N. M.; Franco, P.; Lindner, W. “Separation of enantiomers: Needs, challenges, persectives” J. Chromatogr. A, 2001, 906, 3–33. 16. Jagt, J. C.; van Leusen, A. M. “Diels-Alder Cycloadditions of Sulfonyl Cyanides with Cyclopentadiene. Synthesis of 2-azabicyclo[2.2.1]hepta-2,5-dienes,” J. Org. Chem., 1974, 39, 564. 17. Griffiths, G. J.; Previdoli, F. E. “Diels-Alder Reactions of Methanesulfonyl Cyanide with Cyclopentadiene. Industrial Synthesis of 2-azabicyclo[2.2.1]hepta-5-en-3-one,” J. Org. Chem., 1993, 58, 6129. 18. Huang, D. S.; Huang, K. L.; Chen, S. P.; Liu, S.; Yu, J. “Rapid reaction-diffusion model for the enantioseparation of phenylalanine across hollow fiber supported liquid membrane,” Sep. Sci. Tech., 2008, 43(2), 259–272. 19. Miyako, E.; Maruyama, T.; Kamiya, N.; Goto, M. “Highly enantioselective separation using a supported liquid membrane encapsulating surfactant-enzyme complex,” J. Am. Chem. Soc., 2004, 126(28), 8622–8623. 20. Cbia, P. S. K.; Lindoy, L. F.; Walker, G. W.; Everett, G. W. “Supramolecular transport of metal ammine and amine complexes through chloroform membranes by the natural ionophore lasalocid A. The selective enantiomeric transport of chiral metal complexes,” J. Am. Chem. Soc., 1991, 113, 2533–2537. 21. Xie, R.; Chu, L. Y.; Deng, J. G. “Membranes and membrane processes for chiral resolution,” Chem. Soc. Rev., 2008, 37(6), 1243–1263. 22. Lopez, J. L.; Matson, S. L. “A multiphase/extractive enzyme membrane reactor for production of diltiazem chiral intermediate,” J. Membr. Sci., 1997, 125(1), 189–211. 23. Xiao, Y.; Lim, H. M.; Chung, T. S.; Rajagopalan, R. “Acetylation of β-cyclodextrin surfacefunctionalized cellulose dialysis membranes with enhanced chiral separation,” Langmuir, 2007, 23(26), 12990–12996. 24. Kakuchi, T.; Takaoka, T.; Yokota, K. “Polymeric Chiral Crown Ethers VI. Optical resolution of α-amino acid by polymers incorporating 1,3;4,6-di-o-benzylidene-D-mannitol residues,” Polym. J., 1990, 22, 199–205. 25. Shinohara, K.; Aoki, T.; Oikawa, E. “Optical resolution by vapor permeation of 1,3-butanediol and 2-butanol through (+)-poly(1-[dimethyl(10-pinanyl)silyl]-1-propyne) membrane. Polymer, 1995, 36(12), 2403–2405. 26. Thoelen, C.; De bruyn, M.; Theunissen, E.; Kondo, Y.; Vankelecom, I. F. J.; Grobet, P.; Yoshikawa, M.; Jacobs, P. A. “Membranes based on poly(gamma-methyl-L-glutamate): Synthesis, characterization and use in chiral separations,” J. Membrane Sci., 2001, 186(2), 153–163.

138

A. Higuchi et al.

27. Aoki, T.; Tomizawa, S.; Oikawa, E. “Enantioselective permeation through poly(gamma[3[(pentamethyldisiloxanyl)propyl]-L-glutamate] membranes. J. Membrane Sci., 1995, 99(2), 117–125. 28. Maruyama, A.; Adachi, N.; Takatsuki, T. “Enantioselective permeation of alpha-aminoacid isomers through poly(amino acid)-derived membranes,” Macromolecules, 1990, 23(10), 2748–2752. 29. Kim, J. H.; Kim, J. H.; Jegal, J.; Lee, K.-H. “Optical resolution of alpha-amino acids through enantioselective polymeric membranes based on polysaccharides,” J. Membrane Sci., 2003, 213(1–2), 273–283. 30. Kim, J. H.; Jegal, J.; Kim, J. H.; Lee, K.-H.; Lee, Y. “Enantioselective permeation of alphaamino acid optical isomers through crosslinked sodium alginate membranes,” J. Appl. Polym. Sci., 2003, 89(11), 3046–3051. 31. Xie, S. M.; Wang, W. F.; Ai, P.; Yang, M.; Yuan, L.-M. “Chiral separation of (R,S)-2-phenyl1-propanol through cellulose acetate butyrate membranes,” J. Membrane Sci., 2008, 321(2), 293–298. 32. Yashima, E.; Noguchi, J.; Yoshio Okamoto, Y. “Photocontrolled chiral recognition by [4(Phenylazo)phenyl]carbamoylated cellulose and amylose membranes,” Macromolecules, 1995, 28(24), 8368–8374. 33. Vogt, U.; Zugenmaier, P. “Structural models of some liquid crystalline cellulose derivatives.” Ber. Bunsen-Ges. Phys. Chem., 1985, 89, 1217–1224. 34. Masawaki, T.; Sasai, M.; Tone, S. “Optical resolution of an amino-acid by an enantioselective ultrafiltration membrane,” J. Chem. Eng. Jpn., 1992, 25(1), 33–39. 35. Lee, N. H.; Frank, C. W. “Separation of chiral molecules using polypeptide-modified poly(vinylidene fluoride) membranes,” Polymer, 2002, 43, 6255–6262. 36. Gumi, T.; Minguillon, C.; Palet, C. “Separation of propranolol enantiomers through membranes based on chiral derivatized polysulfone,” 2005, 46(26), 12306–12312. 37. Aoki, T.; Fukuda, T.; Shinohara, K. I.; Kaneko, T.; Teraguchi, M.; Yagi, M. “Synthesis of chiral helical poly[p-(oligopinanylsiloxanyl)phenylacetylene]s and enantioselective permeability of their membranes,” J. Polym. Sci., Polym. Chem., 2004, 42(18), 4502–4517. 38. Satoh, T.; Tanaka, Y.; Yokota, K.; Kakuchi, T. “Enantioselective permeability of membranes prepared from polyacrylonitrile-graft-(1 -> 6)-2,5-anhydro-D-glucitol,” Reactive & Functional Polymers, 1998, 37(1–3), 293–298. 39. Krieg, H. M.; Breytenbach, J. C.; Keizer, K. “Chiral resolution by beta-cyclodextrin polymerimpregnated ceramic membranes,” J. Membrane Sci., 2000, 164(1–2), 177–185. 40. Wang, H. D.; Chu, L. Y.; Song, H.; Yang, J.-P.; Xie, R.; Yang, M. “Preparation and enantiomer separation characteristics of chitosan/beta-cyclodextrin composite membranes,” J. Membrane Sci., 2007, 297(1–2), 262–270. 41. Ishihara, K.; Suzuki, N.; Matsui, K. “Optical resolution of amino acids by a polymer membrane having cyclodextrin moieties,” Nikkashi, 1987, 3, 446–451. 42. Xiao, Y. C.; Chung, T. S. “Functionalization of cellulose dialysis membranes for chiral separation using beta-cyclodextrin immobilization” J. Membrane Sci., 2007, 290(1–2), 78–85. 43. Schneiderman, E.; Stalcup, A. M. “Cyclodextrins: A versatile tool in separation science,” J. Chromatogr. B, 2000, 745, 83–102. 44. Kim, B. Y.; Yang, J.; Gong, M. J.; Flachsbart, B. R.; Shannon, M. A.; Bohn, P. W.; Sweedler, J. V. “Multidimensional separation of chiral amino acid mixtures in a multilayered three-dimensional hybrid microfluidic/nanofluidic device,” Anal. Chem., 2009, 81, 2715–2722. 45. Yang, M.; Chu, L. Y.; Wang, H. D.; Xie, R.; Song, H.; Niu, C. H. “A thermoresponsive membrane for chiral resolution,” Adv. Mater., 2008, 18, 652–663. 46. Higuchi, A.; Ishida, Y.; Nakagawa, T. “Surface modified polysulfone membranes: Separation of mixed proteins and optical resolution of tryptophan,” Desalination, 1993, 90, 127– 136. 47. Kragh-Hansen, U. “Octanoate binding to the indole- and benzodiazepine-binding region of human serum albumin,” Biochem. J., 1991, 273, 641–644.

Membranes for Chiral Separation

139

48. Higuchi, A.; Hara, M.; Horiuchi, T.; Nakagawa, T. “Optical resolution of amino acids by ultrafiltration membranes containing serum albumin,” J. Membrane Sci., 1994, 93(2), 157– 164. 49. Lakshmi, B. B.; Martin, C. R. “Enantioseparation using apoenzymes immobilized in a porous polymeric membrane,” Nature, 1997, 388(6644), 758–760. 50. Lee, S. B.; Mitchell, D. T.; Trofin, L.; Nevanen, T. K.; Soderlund, H.; Martin, C. R. “Antibody-based bio-nanotube membranes for enantiomeric drug separation,” Science, 2002, 296, 2198–2200. 51. Wagenknecht, H.-A.; Stemp, E. D. A.; Barton, J. K. “Evidence of electron transfer from peptides to DNA: Oxidation of DNA-bound tryptophan using flash-quench technique,” J. Am. Chem. Soc., 2000, 122, 1–7. 52. Higuchi, A.; Yomogita, H.; Yoon, B. O.; Kojima, T.; Hara, M.; Maniwa, S.; Saitoh, M. “Optical resolution of amino acid by ultrafiltration using recognition sites of DNA,” J. Membrane Sci., 2002, 205(1–2), 203–212. 53. Higuchi, A.; Komuro, A.; Hirano, K.; Yoon, B. O.; Hara, M.; Hirasaki, T.; Nishimoto, Y.; Yokogi, M.; Manabe, S. “Permeation of γ -globulin through Microporous Membranes under Existence of Trace DNA,” J. Membrane Sci., 2001, 186(1), 9–18. 54. Higuchi, A.; Furuta, K.; Yomogita, H.; Yoon, B. O.; Hara, M.; Maniwa, S.; Saitoh, M. “Optical resolution of amino acid by ultrafiltration through immobilized DNA membranes,” Desalination, 2002, 148, 155–157. 55. Higuchi, A.; Higuchi, Y.; Furuta, K.; Yoon, B. O.; Hara, M.; Maniwa, S.; Saitoh, M.; Sanui, K. “Chiral separation of phenylalanine by ultrafiltration through immobilized DNA membranes,” J. Membrane Sci., 2003, 221(1–2), 207–218. 56. Matsuoka, Y.; Kanda, N.; Lee, Y. M.; Higuchi, A. “Chiral separation of phenylalanine in ultrafiltration through DNA-immobilized chitosan membranes,” J. Membrane Sci., 2006, 280(1–2), 116–123. 57. Zhang, L. J.; Song, M. F.; Tian, Q.; Min, S. “Chiral separation of L,D-tyrosine and L,D-tryptophan by ct DNA,” Separation and Purofication Technology, 2007, 55(3), 388– 391. 58. Higuchi, A.; Wang, P.; Tak, T.-M.; Nohmi, T.; Hashimoto, T. “Preparation and properties of surface-modified polyacrylonitrile hollow fibers,” in: I. Pinnau, B. D. Freeman (Eds.), Membrane Formation and Modification, ACS Symposium Series, vol. 744, 1999, p. 228– 237. 59. Kempe, M.; Mosbach, K. “Molecular imprinting used for chiral separations,” J. Chromatgr. A, 1995, 694, 3–13. 60. Kondo, Y.; Yoshikawa, M. “Effect of solvent composition on chiral recognition ability of molecularly imprinted DIDE derivatives,” Analyst, 2001, 126, 781–783. 61. Yoshikawa, M.; Izumi, J.; Kitano, T. “Molecularly imprinted polymeric membranes for optical resolution,” J. Membrane Sci., 1995, 108(1–2), 171–175. 62. Itou, Y.; Nakano, M.; Yoshikawa, M. “Optical resolution of racemic amino acid derivatives with molecularly imprinted membranes from tetrapetide consisting of glycinyl residues,” J. Membrane Sci., 2008, 325, 371–375. 63. Son, S. H.; Jegal, J. “Chiral separation of D,L-serine racemate using a molecularly imprinted polymer composite membrane,” J. Appl. Polym. Sci., 2007, 104(3), 1866–1872. 64. Jiang, Z. Y.; Yu, Y. X.; Wu, H. “Preparation of CS/GPTMS hybrid molecularly imprinted membrane for efficient chiral resolution of phenylalanine isomers,” J. Membr. Sci., 2006, 280(1–2), 876–882. 65. Teraguchi, M.; Masuda, T. “Poly(diphenylacetylene) membranes with high gas permeability and remarkable chiral memory,” Macromolecules, 2002, 35(4), 1149–1151. 66. Teraguchi, M.; Suzuki, J.; Kaneko, T.; Aoki, T.; Masuda, T. “Enantioselective permeation through membranes of chiral helical polymers prepared by depinanylsilylation of poly(diphenylacetylene) with a high content of the pinanylsilyl group,” Macromolecules, 2003, 36(26), 9694–9697.

140

A. Higuchi et al.

67. Teraguchi, M.; Mottate, K.; Kim, S.-Y.; Aoki, T.; Kaneko, T.; Hadano, S.; Masuda, T. “Synthesis of Chiral Helical Poly(hydroxyl-containing phenylacetylene) Membranes by in-Situ Depinanylsilylation and Their Enantioselective Permeabilities,” Macromolecules, 2005, 38(15), 6367–6373. 68. Robeson, L. M.; Freeman, B. D.; Paul, D. R.; Rowe, B. W. “An empirical correlation of gas permeability and permselectivity in polymers and its theoretical basis,” J. Membrane Sci., 2009, 341, 178–185. 69. Robeson, L. M. “The upper bound revisited,” J. Membrane Sci., 2008, 320, 390–400. 70. Freeman, B. D. “Basis of permeability/selectivity tradeoff relations in polymeric gas separation membranes,” Macromolecules, 1999, 32, 375. 71. Taki, K.; Arita, I.; Satoh, M.; Komiyama, J. “Selective transport of D,L-tryptophan through poly(L-glutamic acid) membranes,” J. Polym. Sci. Part B.-Polym. Phys., 1999, 37(10), 1035–1041. 72. Kim, J. H.; Kim, J. H.; Jegal, J.; Lee, K.-H. “Optical resolution of alpha-amino acids through enantioselective polymeric membranes based on polysaccharides,” J. Membrane Sci., 2003, 213(1–2), 273–283. 73. Tone, S.; Masawaki, T.; Hamada, T. “Optical resolution of amino-acids by ultrafiltration membranes fixed with plasma-polymerized L-menthol,” J. Membrane Sci., 1995, 103(1–2), 57–63. 74. Osada, Y.; Ohta, F.; Mizumoto, A.; Takase, M.; Kurimura, Y. “Specific Permeation and Adsorption of Aqueous Amino Acids by Plasma-Polymerized Membranes of d-Camphor and 1-Menthol,” Nippon Kagaku Kaishi, 1986, 7, 866–872. 75. Tone, S.; Masawaki, T.; Eguchi, K. “The optical resolution of amino acids by plasma polymerized terpene membranes,” J. Membrane Sci., 1996, 118(1), 31–40. 76. Nakagawa, M.; Ikeuchi, Y.; Yoshikawa, M. “Chiral separation of racemic amino acid with novel polyamides having N-alpha-acetyl-L-glutamyl residue as a diacid component,” Polymer, 2008, 49, 4612–4619. 77. Hazarika, S. “Enantioselective permeation of racemic alcohol through polymeric membrane,” J. Membrane Sci., 2008, 310(1–2), 174–183. 78. Aoki, T.; Shinohara, K.; Kaneko, T.; Oikawa, E. “Enantioselective permeation of various racemates through an optically active poly{1-[dimethyl(10-pinanyl)silyl]-1-propyne} Membrane,” Macromolecules, 1996, 29(12), 4192–4198. 79. Ogata, N. “Optical resolution through membranes having supramolecular self-organized structures,” Reactive & Functional Polymers, 1995, 26(1–3), 201–208. 80. Wang, J.; Fu, C. J.; Lin, T.; Yu, L.; Zhu, S. “Preparation of chiral selective membranes for electrodialysis separation of racernic mixture,” J. Membrane Sci., 2006, 276(1–2), 193–198. 81. Hadik, P.; Kotsis, L.; Eniszne-Bodogh, M.; Szab’o, L.-P.; Nagya, E. “Lactic acid enantio separation by means of porous ceramic disc and hollow fiber organic membrane,” Sep. Purif. Tech., 2005, 41(3), 299–304. 82. Gumi, T.; Palet, C.; Ferreira, Q.; Viegas, R. M. C.; Crespo, J. G.; Coelhoso, I. M. “Enantioselective separation of propranolol by chiral activated membranes,” Sep. Sci. Tech., 2005, 40(4), 773–789. 83. Higuchi, A.; Ishida, Y.; Nakagawa, T. “Surface Modified Membranes: Separation of Mixed Proteins and Optical Resolution of Tryptophan,” Desalination, 1993, 90(1–3), 127–136. 84. Yoshikawa, M.; Yonetani, K. “Molecularly imprinted polymeric membranes with oligopeptide tweezers for optical resolution,” Desalination, 2002, 149(1–3), 287–292. 85. Yoshikawa, M.; Ooi, T.; Izumi, J. “Novel membrane materials having EEE derivatives as a chiral recognition site,” European Polymer J., 2001, 37, 335–342. 86. Yoshikawa, M.; Nakai, K.; Matsumoto, H.; Tanioka, A.; Guiver, M. D.; Robertson, G. P. “Molecularly imprinted nanofiber membranes from carboxylated polysulfone by electrospray deposition,” Macromolecuar Rapid Communications, 2007, 28(21), 2100–2105. 87. Yoshikawa, M.; Murakoshi, K.; Kogita, T.; Hanaoka, K.; Guiver, M. D.; Robertson, G. P. “Chiral separation membranes from modified polysulfone having myrtenal-derived terpenoid side groups,” European Polymer J., 2006, 42(10), 2532–2539.

Membranes for Chiral Separation

141

88. Yoshikawa, M.; Koso, K.; Yonetani, K.; Kitamura, S.; Kimura, S. “Optical resolution of racemic amino acid derivatives with molecularly imprinted membranes bearing oligopeptide tweezers,” J. Polym. Sci. Part A, Polym. Chem., 2005, 43(2), 385–396. 89. Higuchi, A.; Hashimoto, T.; Uemiya, N.; Kojima, T.; Hara, M. “Effect of surfactive agents and lipids on optical resolution of amino acid by ultrafiltration membranes containing serum albumin,” J. Membrane Sci., 1997, 130, 31–39. 90. Romero, J.; Zydney, A. L. “Staging of affinity ultrafiltration processes for chiral separations,” J. Membrane Sci., 2002, 209(1), 107–119. 91. Romero, J.; Zydney, A. L. “pH and salt effects on chiral separations using affinity ultrafiltration,” Desalination, 2002, 148(1–3), 159–164. 92. Romero, J.; Zydney, A. L. “Chiral separations using ultrafiltration with a stereoselective binding agent,” Sep. Sci. Tech., 2001, 36(7), 1575–1594.

APPENDIX A

Table 1S Relationship between permeability coefficients and separation factor of phenylalanine through several chiral polymeric membranes concentration-driven permeation. Data were cited from references No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Membranes poly(1) original Poly(2) original copoly(1/3) original copoly(2/3) original poly(3) original poly(1) depinalsilylated Poly(2) depinalsilylated copoly(1/3) depinalsilylated copoly(2/3) depinalsilylated poly(3) depinalsilylated PDSP in methanol Poly(p-PSPA) Poly(1,3-BPDSPA) Poly(3-PDSPA) Poly(1-PDSPA) Poly(1,3,5-TPTSPA) Poly(1,3-BPTSPA) PAN-bCD PAN-Glucose PAN-bCD PLGA β-cyclodextrin immobilized cellulose D-amino acid oxidase apoenzyme

Preferential permeation

alpha

Ref. No.

8.06E-14 8.39E-14 1.09E-13 1.08E-13 1.21E-13 3.06E-13 2.97E-13 3.64E-13 3.50E-13 2.67E-13 4.58E-07 3.08E-12 7.33E-13 8.78E-12 3.08E-13 3.94E-14 1.29E-13 2.94E-07 3.69E-07 2.31E-07 1.71E-07 1.53E-07

R R R R R R R R R R R R R R R R R S S S S R

2.60 2.07 4.05 3.47 7.85 1.19 1.13 1.40 1.32 1.53 3.35 640.00 20.50 1.12 6.10 1.74 1.39 1.40 1.02 1.61 1.26 1.30

67 67 67 67 67 67 67 67 67 67 78 37 37 37 37 37 37 41 41 41 35 21

2.20E-05

R

3.30

49

P (cm2/sec)

142

A. Higuchi et al.

Table 2S Relationship between permeability coefficients and separation factor of tryptophan through several chiral polymeric membranes concentration-driven permeation and potential-driven permeation. Data were cited from references No. 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

Membranes PPDPA-PDPA-1 PPDPA-PDPA-2 PPDPA-PDPA-3 de-PPDPA-PDPA-1 de-PPDPA-PDPA-2 de-PPDPA-PDPA-3 (+)-poly(DPSP) (+)-poly(DPSP) PMLGa supported PMLGa (+)-poly(p-PMPA) (+)-poly(p-PSPA) PAN-bCD PAN-Glucose PAN-bCD Physisorbed PLGA Chemisorbed PLGA Chemisorbed PLGA Chemisorbed PLTEG Chemisorbed PBLG β-cyclodextrin immobilized cellulose Copoly(PSDPA) Copoly(PSDPA) Copoly(PSDPA) Poly(PPDPA/PDPA) PEGDI37 PEGDI37 PEGDI37 PEGDI37Li PEGDI37Li PEGDI46 PEGDI55 PEGDI55 PEGDI55 PEGDI55 PEGDI55 PEGDI37 PEGDI37 PEGDI37

Preferential permeation

Alpha

Ref. No.

2.49E-08 4.30E-08 2.24E-08 1.78E-08 2.08E-07 2.85E-07 2.04E-07 1.42E-07 1.66E-07

R R R R R R R R R R R R S S S S S S R S R

1.39 1.60 3.38 1.22 1.36 2.88 2.64 1.41 3.04a 1.57a 1.38 2.90 1.28 1.02 1.45 1.29 1.47 1.46 1.29 1.00 1.10

66 66 66 66 66 66 78 78 26 26 37 37 41 41 41 35 35 35 35 35 42

1.53E-11 1.43E-11 1.00E-11 2.06E-08 4.10E-08 2.46E-07 5.35E-07 2.63E-07 6.32E-07 2.32E-07 1.36E-07 1.82E-07 1.84E-07 4.36E-07 1.54E-07 8.10E-09 1.20E-08 9.90E-08

R R R R R R R R R R R R R R R R R R

1.40 1.60 3.40 1.40 1.03 1.21 1.13 1.13 1.10 1.20 1.02 1.13 1.33 1.17 1.27 1.50 1.90 1.70

66 66 66 66 71 71 71 71 71 71 71 71 71 71 71 71 71 71

P (cm2/sec) 1.53E-11 1.43E-11 1.00E-11 6.25E-11 5.47E-11 3.72E-11 3.78E-07 2.72E-07 2.33E-06a 3.61E-07a

Membranes for Chiral Separation

143

Table 2S Relationship between permeability coefficients and separation factor of tryptophan through several chiral polymeric membranes concentration-driven permeation and potential-driven permeation. Data were cited from references (Continued) No. 40 41 42 43 44 45 46 47

Membranes HDI55 HDI55 HDI55 HDI73 Plasma polymerized terpene Plasma polymerized terpene Plasma polymerized terpene Plasma polymerized terpene

a

Potential-driven permeation

P (cm2/sec) 1.50E-08 2.20E-08 4.10E-08 2.10E-08 5.02E-09 1.87E-09 6.04E-09 4.71E-09

Preferential permeation

Alpha

Ref. No.

R R R R R R R R

1.50 2.60 1.20 1.40 1.40 1.90 1.10 1.20

71 71 71 71 75 75 75 75

Polymer Reviews, 50:144–177, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583721003698846

Dense Gas Processing of Polymers R. B. YOGANATHAN§, R. MAMMUCARI, AND N. R. FOSTER School of Chemical Sciences and Engineering, Chemical Sciences Building, University of New South Wales, Sydney NSW 2052, Australia There is a growing global awareness about environmental pollution, and many sanctions and sustainable practices have been implemented. In particular, the use of volatile organic compounds (VOCs) is a practice that is being limited and minimized world-wide. These VOCs are not only damaging to the environment, but are also an occupational hazard. The polymer processing industry is known to use VOCs extensively for polymerization, fractionation, plasticization, degradation, extraction and purification. More environmentally-friendly methods to circumvent the use of these toxic and hazardous compounds are being explored. The use of dense gases in polymer processing can respond to the need for more environmentally-friendly industrial processes. Products with high-purity, sterility, and porosity can be achieved using dense gas technology (DGT). Currently, DGT has been used for different aspects of polymer processing including polymerization, micronization, and impregnation. Due to its high solubility in polymers and diffusivity, dense CO2 can penetrate and plasticize polymers, whilst impregnating them with low-molecular weight CO2 -soluble compounds. The dense CO2 properties of inertness, non-toxicity, and affinity for various therapeutic compounds are specifically advantageous to the medical and biomedical industries. Biodegradable polymers and other medical-grade polymers have benefited from the application of DGT. The aim of this review was to show the versatility of dense CO2 for polymer processing applications, specifically polymerization, polymer blend preparation, drug loading and sterilization. Keywords polymer processing, supercritical fluids, dense gas, supercritical CO2 , dense CO2 , polymer impregnation, polymer blends, dense gas technology

1. Introduction The polymer industry expels over 20 million tons of volatile organic compounds (VOCs) each year.1 Organic solvents are used at various stages of polymer production; as reaction media and as post-polymerization processing media (extraction, purification). To avoid damaging the environment and creating unsafe working environments, alternative and sustainable media and methods to reduce organic solvent use have been sought. There are two routes to reducing the usage of organic solvents and emission of VOCs2; • develop a solvent-free method • use environmentally-friendly solvents Received July 20, 2009; accepted January 6, 2010. § Current address: Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, 144 College Street, Toronto, Ontario, Canada, M5S 3M2. Address correspondence to Neil Foster, The University of New South Wales, Sydney, NSW 2052. E-mail: [email protected]

144

Dense Gas Processing of Polymers

145

Figure 1. Progression of dense CO2 -related polymer publications.9

Using a solvent-free method such as melt phase polymerization has the problems of high viscosity and limited mass transfer. Also, the selection of an environmentally-friendly liquid solvent may still require steps to remove the solvent after the polymerization process or post-processing stages. The use of a dense gas (DG) such as CO2 has been researched and studied for polymerization and polymer processing. A DG is a fluid with a temperature and pressure close to the critical point. The applicability of dense CO2 as a sustainable alternative to conventionally used organic solvents has been reported in the literature.2–8 Research on dense CO2 -related polymer processing is a growing field. Over the past two decades, the number of publications per year related to dense CO2 polymer processing (Fig. 1) has been steadily increasing.9 The focus of this review is the application of dense gas technology to polymer processing, with attention on the usage of dense CO2 as polymerization, blending, impregnation and sterilization media.10–13 Unlike earlier reviews on DG processing of polymers, this paper focuses on the feasibility of concurrent processes (blending, impregnation and sterilization), and on the implications of DG polymer processing for biomedical applications. Relevant published work on various dense CO2 -related polymer processing topics can be found in Table 1. 2. Dense Gas Technology Dense gas technology (DGT) optimizes the properties of fluids close to and above their critical point for applications such as purification, extraction, polymer processing, impregnation of chemical compounds, pharmaceutical processing, and sterilization of medical devices.12,111,114 A dense gas is a substance which exists at or near its critical point. The critical point of a substance is located at the termination of the phase boundary between the gas and liquid phases of a substance (Fig. 2). As pressure and temperature increase beyond the critical point the phases become indistinguishable as either a gas or a liquid, and become a supercritical fluid (SCF). Above the critical temperature and critical pressure, the substance exists as a homogeneous medium which exhibits liquid-like densities and gas-like mass transfer properties.

146

R. Yoganathan et al. Table 1 Dense CO2 -related polymer processing reported in the literature

Topics

References

Polymer Phase Behavior Polymer Plasticization Polymer Foaming Polymerization Polymer Blending Polymer Impregnation Polymer Cross-linking Polymer Sterilization Polymer Micronization

12, 14–41 11, 42–51 26, 27, 52–63 10, 64–84 24, 54, 80–94 2, 26, 27, 32, 43, 65, 66, 95–104 12, 26, 27, 101 13, 105–17 27, 65, 70, 108–113

3. Physical Properties of DGs The advantages of DGs over conventional solvents include low surface tension, low viscosity, high diffusivity, and density-dependent solvent power.116 It is possible to make the transition smoothly from a gas (point A) to a liquid (point B) without exhibiting distinct phase transitions by following the A-B path outlined in Fig. 2. The path connecting point A and point B does not cross the phase boundary, but passes through the supercritical region. All substances in the supercritical region are referred to as supercritical fluids (SCFs). All SCFs, and compounds at or near the critical point, fall under the broader definition of DGs. The different physical properties of SCFs, gases and liquids are listed in Table 2. The density of a SCF is similar to a liquid; however, its viscosity is much closer to a gas. The solvent strength of a DG defines its ability to dissolve solutes. The greater the density the greater the ability of the DG to dissolve a solute. In the late 19th century, Hannay and

Figure 2. Phase diagram of a pure compound.115

Dense Gas Processing of Polymers

147

Table 2 Physical properties of a gas, a liquid and a supercritical fluid120 Property Density (gc/m−3) Diffusion Coefficient (cm2/s−1) Viscosity (cP)

Gas

SCF

Liquid

0.0006–0.002 0.1–0.4 0.01–0.03

0.2–0.9 0.0002–0.0007 0.01–0.09

0.6–1.6 0.000002–0.00002 0.2–3.0

Hogarth117, 118 reported the increased solubility of the inorganic salts, cobalt (II) chloride and iron (III) chloride, in supercritical ethanol with increments in pressure. The solvent properties of liquid CO2 for various organic and inorganic compounds have been reported by Francis.119 The solvent strength of a DG is dependent on its density, which is related to pressure and temperature. The reduced temperature (TR = T/TC ) and reduced pressure (PR = P/PC ) of a DG is between 0.9 and 1.2.120 Within the aforementioned range, the density of a DG is highly sensitive to changes in pressure (Fig. 3). At a constant temperature, a small increase in pressure causes a noticeable change in the density. A high sensitivity translates to an

Figure 3. Density vs pressure of CO2 .121

148

R. Yoganathan et al. Table 3 Critical temperatures and critical pressures of dense gases120 Substance Ethylene Xenon Carbon Dioxide Ethane Nitrous Oxide Propane N-Pentane Trichlorofluoromethane Isopropanol Methanol Cyclohexane Benzene Toluene p-Xylene Water

TC (◦ C)

PC (bar)

9.3 16.6 31.1 32.2 36.5 96.7 196.5 198.1 235.2 239.5 280.3 289 318.6 343.1 374.2

50.4 58.4 73.8 48.8 71.7 42.5 33.7 44.1 47.6 81 40.7 48.9 41.1 35.2 220.5

ability to control the density and solvent strength of a DG. By tuning the solvent strength of the DG, its selectivity to dissolve various compounds can also be controlled. The properties of DGs such as tunable solvent strength and selectivity have enabled their use in various food-related industrial applications such as decaffeination,120 and the extraction of tea, hops, spices, and other flavors.122 The critical temperatures and critical pressures of selected DGs can be found in Table 3. According to Pereda et al.,116 DGs can be classified into two groups based on their TC . There are low TC gases which are condensable and high TC fluids which have a much greater solvent power, and are better for higher molecular weight compounds. The low TC gases can be applied to processes at moderate conditions, and also have a better selectivity for low molecular weight compounds. Carbon dioxide is the most used DG because it is widely available, environmentally benign, and has an easily accessible critical temperature (31.1◦ C) and critical pressure (73.8 bar). At the critical point, CO2 has a liquid-like density of 0.47 gcm−3, whereas conventional organic solvents have densities in the range of 0.8–1.0 gcm−3. In the DG region, the density of CO2 can be tuned by varying the pressure and temperature of the system. At a temperature of 35◦ C, which is above the CO2 critical temperature, it is possible to obtain a CO2 density in the range of 0.3–0.6 gcm−3 by varying the pressure between 65 bar and 80 bar. Such liquid-like densities promote the use of dense CO2 as an alternative to organic solvents.71,111,112 Dense gases also have gas-like diffusivity properties, which allow them to penetrate different materials and lower the viscosity of mixtures. These intermediary physicochemical properties allow for the use of dense CO2 for extraction, fractionation, chromatography, polymerization, micronization, and degradation.2,31,64,114,120,123–131 The physical properties of DGs greatly affect their interactions with other compounds and solvents. Processing polymers with DGs is a growing field and has been reviewed in the past.12,64,67,96,111,129,130–132 The focus of this review will be on the scope and use of DG for polymer processing.

Dense Gas Processing of Polymers

149

Figure 4. Schematic of CO2 -polymer interactions and potential applications.2

4. CO2 -Polymer Interaction The type of CO2 -polymer interaction ultimately decides its potential application, as depicted in Fig. 4. The application of dense CO2 to polymer processing depends on the solubilities of dense CO2 and the polymer in each other. A polymer swelling in the presence of dense CO2 is the first indication of an interaction. The swelling can be caused by either the polymer being CO2 -soluble, or CO2 being soluble in the polymer. Dense gas processing of polymers encompasses both types of interactions which cause swelling. Carbon dioxide on its own has a strong quadrupole moment (Fig. 5) and low polarizability, similar to methane. The aforementioned properties make dense CO2 a good solvent for both small polar and non-polar compounds,111,130,133 such as monomers, initiators, catalysts, cross-linkers, and oligomers. The majority of larger compounds such as polymers have limited or selective solubility in dense CO2 . The polymers which are CO2 -soluble possess some level of polarity and have some electronegative groups.120 Examples of CO2 -soluble polymers include polydimethylsiloxane (PDMS), polyalkene oxide (PAO), perfluorinated polypropylene oxide (PPO), polyvinyl acetate (PVAc), and polymethyl acrylate (PMA).2 The class of high molecular weight polymers most soluble in CO2 is fluorinated acrylates.38 The geometrical interaction of CO2 with various fluorinated polymers was studied.40 The

150

R. Yoganathan et al.

Figure 5. Dipole and quadrupole charge distribution.2

results indicate that high solubility results from close interaction between the center of masses and an increased free volume in the polymer matrix. Within high molecular weight polymers, PVAc has been identified as the most CO2 soluble.39,41 The differences in solubility of various hydrocarbon-based CO2 -philic polymers (e.g. PVAc and PMA) can be linked to crystallinity, the presence of side chains, and accessibility of carbonyl groups.39 Spacing of the ether oxygen and carbonyl oxygen along the backbone also may affect solubility in CO2 .39,41 In general, the presence of ether oxygens in the polymer structure results in higher solubility in CO2 .41 Polymer solubility is dependent on polymer molecular weight and polydispersity.21 The solubility of a polymer in dense CO2 can be increased by the addition of co-solvents. The behavior of polymers in the presence of dense CO2 , with and without co-solvents has been reported in the literature.21 The other type of CO2 -polymer interaction occurs when CO2 is soluble in a polymer. For this type of interaction, the physical and chemical structure of the polymer affects the solubility of CO2 into its matrix. Extra polar groups and increased free volume in the polymer matrix increases the CO2 solubility in the polymer.43,104,132,134 The spectroscopic evidence of the chemistry behind the CO2 -polymer interaction was first obtained by Kazarian and co-workers43,104,132,134 while observing polymer plasticization with IR spectroscopy.

5. Chemical Processing 5.1 Polymerization Dense CO2 is an attractive alternative to conventional polymerization media because it is widely available, and most importantly, it is environmentally-friendly. The use of dense CO2 as a medium can help overcome some of the problems faced by the conventional polymerization media, such as high viscosity, low solubility, and rigorous purification processes. The DG polymerization field is gaining more interest because dense CO2 is a proven alternative to organic solvents and can readily solubilize various organic compounds. The interactions between certain polymers and dense CO2 are favorable to polymerizations in dense CO2 . Conventional polymerization methods commonly employ high temperatures to overcome the activation energy, break bonds, and make the monomer reactive.135 A suppressed glass transition temperature (Tg ) translates to a lower activation energy required to initiate the polymerization reaction. With its growing popularity over the past two decades, books and many reviews have been written on dense CO2 polymerization, where its use as a reaction medium and reactant have been discussed.2,67,126

Dense Gas Processing of Polymers

151

Carbon dioxide in the DG region provides better mass transfer properties than liquid CO2 , but maintains similar solvent behavior to liquid CO2 .124 Dense CO2 behaves like hydrocarbon solvents because it has an affinity for low molecular weight non-polar molecules. The interactions between CO2 and regularly insoluble high molecular weight polymers have been researched over the past two decades.62,64,102,108,114,124,130, 131 Many researchers have reported that in the presence of dense CO2 the Tg of polymers become suppressed.62,67,68,81,111,131,135–139 Polymerizations conducted at high temperatures in the melt phase are highly viscous processes. The high viscosity of the melt reduces the diffusive ability of compounds through the melt, thus restricting polymerization rates, thereby only allowing for the production of low molecular weight polymers. One way to alleviate the viscosity of melt polymerizations is to conduct it in dense CO2 . A reduced viscosity means a lower mass transfer resistance, which allows compounds to move more freely within the medium. Dense CO2 has good mass transfer and diffusivity properties. Many researchers have undertaken studies to develop new polymerization methods involving dense CO2 to help the polymer industry circumvent the use of environmentally hazardous and biologically toxic organic solvents.140, 141 Dense CO2 polymerizations are either homogeneous or heterogeneous. Homogeneous polymerizations in dense CO2 are polymerizations where all reactants/monomers are completely CO2 -soluble, and thereby form a homogeneous phase at the beginning of the polymerization. A heterogeneous polymerization in dense CO2 polymerization involves the use of surfactants and other additives to form a multi-phase reaction medium. The surfactants and other additives are used to create colloids and/or micelles. The classification scheme in Fig. 6 is different from the classification of CO2 -polymer interactions found in Fig. 4. The classification entitled “Polymer soluble in DG” in Fig. 4 encompasses the same class of polymers found under the classification “CO2 -philic” (Fig. 6). The CO2 -phobic classification is given to polymers that undergo swelling when

Figure 6. Classification of polymer behavior in CO2 in relation to dense CO2 polymerization.71

152

R. Yoganathan et al.

dense CO2 is used as the polymerization medium. Polymers that are CO2 -phobic will exhibit hydrophilic or lipophilic properties, a characteristic which is used to find a suitable stabilizer/surfactant for heterogeneous polymerizations. 5.1.1 Homogeneous Polymerization. The use of dense CO2 as a homogeneous reaction medium has been discussed in the literature and the solubilities of high molecular weight polymers in dense CO2 have been reported.67,111,142 Studies show that fluorinated polymers are soluble in dense CO2 whereas polymers with only hydrocarbon backbones have limited solubility.2,111,124,126,130 The dense CO2 polymerization of fluorine containing co-polymers has been documented by Desimone et al.102 and Wood et al.10,129,142 Fluoropolymers are synthesized using free-radical chain growth and cationic chain growth polymerization. Conventional synthesis of fluoropolymers involves the use of chlorofluorocarbons (CFCs), a solvent subject to strict environmental regulations. Free-radical polymerization in dense CO2 is applicable because the initiator, azobisisobutyronitrile (AIBN) is CO2 -soluble. Also, AIBN creates free radicals efficiently and exhibits limited decomposition in dense CO2 compared to its behavior in other conventional solvents.111 The inertness of CO2 makes it unreactive to free radicals, making it an ideal reaction medium for free radical polymerization.67 Homogeneous polymerizations in dense CO2 have been limited to fluoropolymers because no other polymers have shown strong affinity for CO2 . Heterogeneous polymerizations in dense CO2 are applicable to a wider range of polymers and do not require both the monomer and polymer to be CO2 -soluble. 5.1.2 Heterogeneous Polymerization. Heterogeneous polymerizations in DG employ a continuous dense CO2 phase and one or more other phases where the polymerization reaction takes place. Dispersion, emulsion, precipitation, and suspension polymerizations are the main forms of dense CO2 heterogeneous polymerizations. These four types of heterogeneous polymerizations are differentiated by;77 1. initial state of the reactants 2. particle formation mechanism 3. size and shape of the formed particles Precipitation and dispersion polymerization are the most frequently used.71,77,111 At the beginning of the precipitation polymerization process, the monomer and the initiator are soluble in the continuous phase (dense CO2 ), and as the polymer forms and exceeds a certain Mw , it precipitates out of the continuous phase. Precipitation polymerization in dense CO2 uses free-radical or cationic chain growth as the reaction mechanism, and is popular because the product can be extracted immediately after the reaction in a dry state. Upon depressurization and added flushing with dense CO2 , a dry solvent-free product can easily be collected. Studies have shown that the morphology of the product from a precipitation polymerization was not as consistent as polymer particles produced using dispersion polymerization.77,101 In one case, Cooper had synthesized relatively uniform microspheres using precipitation polymerization (void of stabilizers); however, it was under very specific reaction conditions.101 According to Cooper, the uniform microspheres were produced because of the nature of the polymer (divinylbenzene (DVB)) rather than the properties of the process. Dispersion polymerizations have similar starting conditions to precipitation polymerizations because the initiator and monomer are also CO2 -soluble. The resulting product is coated by amphipathic molecules (surfactants, stabilizers) as it precipitates out of the CO2

Dense Gas Processing of Polymers

153

phase.71 The amphipathic coating is a colloid stabilizer which is used to help produce uniform particles (<100 µm) and allow the oligomer to progress to higher molecular weight. The use of colloid stabilizers or surfactants enables the resulting polymer to be collected as a fine powder. Dense CO2 dispersion polymerizations have been used to polymerize methyl methacrylate (MMA), vinyl monomers, and styrene.71,77,102 Both precipitation and dispersion polymerizations in dense CO2 have been modeled mathematically.77 Surfactants and amphiphilic compounds alike play a key role in dispersion polymerizations. Their application to the field of dense CO2 polymerizations has been reported by Woods et al.130 Surfactants are stabilizers used to enhance the solubility of CO2 -insoluble polymers/monomers. Surfactants are amphiphilic molecules, having CO2 -philic and CO2 phobic segments. The CO2 -philic segment of the molecule is typically a fluorine-based or silicone-based compound. Fluorine and silicone-based polymers have the highest known solubility in CO2 .101,130 Research is ongoing for hydrocarbon-based compounds which exhibit superior or similar CO2 -solubility properties to fluoropolymers and silicone-based polymers. To further increase solubility, the pressure and temperature of the system can be raised to extreme levels within the DG region. In a dense CO2 continuous medium, the segments of fluorine in the polymer chain help create a micelle structure, encapsulating the CO2 -phobic (CO2 -insoluble) portion of the polymer chain inside. A micelle is essentially an aggregate with CO2 -phobic and CO2 philic sections. The CO2 -philic section makes up the outside of the micelle in contact with the continuous CO2 phase and the CO2 -phobic section is anchored to the polymer, encasing it within and shielding it away from the outside CO2 environment. The CO2 -phobic part is usually a hydrocarbon polymer, which allows it to easily attach onto the surface of the growing polymer chains. The hydrocarbon has a dual role; not only is it CO2 -phobic, but it also exhibits affinity for the polymer. Micelles have been successfully employed by many researchers to conduct dispersion polymerizations in dense CO2 .71,102,108 Surfactants used in dense CO2 polymerizations are classified based on their intended application and specific method of heterogeneous polymerization (dispersion, emulsion or suspension). Precipitation polymerization does not employ surfactants or stabilizers; therefore they do not tend to produce particles as uniformly as dispersion polymerizations. The majority of the heterogeneous methods of polymerization in the presence of dense CO2 discussed thus far have utilized surfactants with a Mw greater than 600 Da. Lower Mw surfactants are mainly anionic and have been discussed elsewhere.130 One class of high Mw surfactants are Pluronics, which possess desirable behavior in dense CO2 and have been employed for dispersion polymerizations. Pluronics are triblock co-polymers made up of different combinations of polypropylene oxide (PPO) and polyethylene oxide (PEO), and are normally found as PEO-PPO-PEO. By controlling the surfactant concentration, many researchers have been able to produce uniform micro- and nanoparticles in dense CO2 .62,100,108,143,144 Pluronics have been used to produce high Mw PC nanoparticles.108 The inertness and low critical point of CO2 allow for the enzyme-catalyzed polymerizations to be conducted in dense CO2 . Enzymatic reactions in dense gases have been reviewed elsewhere.145 Polymers such as divinyl adipate and polycaprolactone (PCL) have been synthesized in dense CO2 using enzymes.146 Dense CO2 has been successfully used as a reaction media to synthesize polymers. High Mw polymers have been produced by both homogeneous and heterogeneous DG methods. Post and pre-synthesis DG processing is also possible for solute impregnation, blending, purification, and sterilization.

154

R. Yoganathan et al.

6. Physical Processing Dense CO2 solubility in polymers has lead to polymer plasticization, polymer foaming, the formation of porous polymers and polymer impregnation. The ability of CO2 to diffuse into selected polymers provides several benefits to polymer processing. Polymer plasticization is possible because of the solubility of CO2 in the polymer, mainly the ability of CO2 to diffuse into the polymer and the ability of the polymer to absorb the CO2 . One of the earliest incidences of plasticization effects of CO2 on a polymer was discovered as CO2 gas was being carried through polycarbonate (PC) tubing.49 It was later discovered that the PC tubing had a Tg 8–9◦ C lower in the presence of 6.8atm of CO2 .49 Plasticization of polymers via a DG is an underlying principle of various DG polymer processes such as polymer impregnation, extraction, and heterogeneous polymerizations.111 Polymer foaming occurs during the depressurization as the dense CO2 reverts back to the gaseous state and expanding as it exits the polymer matrix. Tomasko et al.,12 report on polymers which have shown good adsorption and solubility of CO2 . Polymers such as PMMA, PS, HIPS, PC, PET, PVC, LDPE, HDPE, and PP have been able to solubilize CO2 . While plasticized, amorphous polymers become less viscous and provide a pathway for the impregnation of CO2 -soluble additives and/or other compounds. 6.1 Polymer Plasticization The plasticization effects of dense CO2 on polymers for DG polymer processing have been studied closely by several researchers.44,139,147–149 While dense CO2 plasticizes polymers it also lowers the Tg causing the compound to become more amorphous. Lowering the Tg makes the polymer susceptible to the penetration of other compounds. In Table 4 various polymers that are susceptible to dense CO2 plasticization are listed.111 The physical properties of a polymer that are most affected by dense CO2 plasticization are Tg and the elastic modulus. Shieh et al. conducted a study on both crystalline and amorphous polymers, with their findings being that amorphous polymers were more susceptible to dense gas plasticization.149 Qualitative analysis of dense CO2 -plasticized polymers involves physical observation of the polymer post-processing for change in color and surface morphology. Quantitative analysis of dense CO2 plasticized polymers normally involves mechanical testing, but Table 4 Plasticization of polymers in DG104 Polymer Polystyrene (PS)31,35 Polyethylene (PE)31 Polymethyl Methacrylate (PMMA)71 Polycarbonate (PC)44 Polyvinyl Chloride (PVC)44 Polyethyl Teraphthalate (PET)44 Cellulose Acetate Butyrate149 Polyurethane (PU)149 Acrylonitrile Butadiene Styrene (ABS)149

DG CO2 , Alkanes CO2 CO2 CO2 CO2 CO2 CO2 CO2 CO2

Dense Gas Processing of Polymers

155

recently spectroscopic analysis was used to define degrees of plasticization. Most recently, Kazarian et al. studied the spectroscopic effect of high pressure CO2 on polymers.104,132 The polymer sorption of high pressure CO2 , polymer swelling in high pressure CO2 and polymer melting in high pressure CO2 were observed and it was found that polymers with a low Tm had the tendency to melt in dense CO2 , whereas polymers with a higher Tm had the tendency to plasticize in dense CO2 . Kazarian et al.43,150 and Shieh et al.,149 have both reported on how CO2 -polymer interactions are molecularly defined by the interactions of the carbonyl groups of the polymer with CO2 .

6.2 Polymer Foaming Amorphous polymers can be plasticized and foamed by dense CO2 . One special class of amorphous polymers which could benefit from the use of dense CO2 foaming are biomedical polymers. The majority of degradable biomedical polymers (PLA, PCL, PLGA, etc.) have a low Tg which would be suppressed in the presence of dense CO2 . As stated above, mechanical properties can be affected and altered by polymer plasticization. For biodegradable polymers, altering the mechanical strength will in turn increase or decrease their degradation rate. 52,103,151–153 Recently, a foaming study on PS and PDL LA using dense CO2 showed that variables such as sorption, degree of plasticization and surface tension of a polymer-dense CO2 system were quantitatively correlated.59 Plasticization leads to foaming and foaming can lead to pore formation. Studies of dense CO2 with biomedical polymers have shown the occurrence of foaming. Dense CO2 processing, fabrication, and relevance of porous biomedical polymer systems has been reported.112 The foaming of biomedical polymers provides an alternative to using toxic porogens, leaching, and other conventional foaming methods.58, 59,61,154 In the biomedical field, the formation of pores in polymers is highly desired for scaffold and tissue engineering applications.59,155–159 Conventional pore formation involves the use of organic solvents, which lead to tedious stages of purification to remove the residual solvents. When dense CO2 is used as the pore-forming agent, it does not leave residual solvents because it reverts to gas as it is depressurized. The rate of depressurization is also used to control the pore sizes.

6.3 Polymer Impregnation Dense gases can be used for the impregnation of polymers with low molecular weight compounds. Dense gases have high diffusivity, low surface tension and also provide an easy method of solvent recovery. With high diffusivity and low surface tension, dense CO2 can penetrate easily into polymers, thereby dispersing solutes throughout the polymer matrix. When returning back to atmospheric temperature and pressure, dense CO2 will revert back to its gaseous state without leaving any residual solvent, thereby leaving behind a highly pure product. Shen et al.95 developed a mathematical model for the impregnation of a hypothetical additive into a polymer substrate. The model indicated that the additive uptake declined at higher CO2 pressures and regardless of the pressure, increased with impregnation time. Experimental observations of vanillin and L-menthol impregnation into cellulose acetate (CA) fibers showed that the additive uptake was affected by CO2 pressure, the Tg of the polymer, and the solubility of the additive in the pure polymer and in the polymer CO2 mixture. Shen et al. conclude that the thermodynamic principles governing the impregnation

156

R. Yoganathan et al.

of a compound into a polymer can be described as inverse to the extraction of compounds from polymers. A study by Berens et al.160 has shown that in the presence of dense CO2 the adsorption of additives, specifically low molecular compounds into PVC, PC, PMMA, and PVA, was accelerated. Berens et al. have reported on the adsorption of naphthalene and aspirin into the aforementioned glassy polymers. The uptake of the additives into the polymer is more dependent on the solubility of the additive in the polymer, and not as much on the solubility of the additive in the CO2 . The additive must obviously be CO2 -soluble, but the ability for it to be impregnated effectively into the bulk of the polymer depends more on the affinity of the compound for the polymer. Kikic and Vecchione96 have reported on the application and advantages of dense CO2 impregnation; Applications: • preparation of drug delivery systems • impregnation of dyes for textile applications • deposition of organic metallic complexes Advantages: • • • • •

high diffusivity into polymers ability to control solvent power and impregnation rate no need for a final drying stage low surface tension ability to act as a porogenic agent

Dense CO2 has been used to create drug delivery systems with both biodegradable and non-biodegradable polymers. Both types of biomedical polymers are susceptible to melting point depression and plasticization in the presence of dense CO2 . The plasticization effect causes a decrease in the Tg of the polymer. Therapeutic compounds which are CO2 -soluble are often used for impregnation using dense CO2 .99 The efficacy of a dense CO2 process to create a controlled drug delivery system is dependent on the interactions between the polymer, drug, and dense CO2 . The dispersion of the drug within the polymer matrix is dependent on the drug being CO2 -soluble, and the drug having affinity for the polymer. Before commencing impregnation experiments, the solubility of the drug in dense CO2 should be studied.96 The impregnation of PVA, PVC, PMMA, and PC using binary (dense CO2 + drug) and ternary (dense CO2 + co-solvent + drug) systems has been studied.160 For the binary system, it was found that PVA was the most plasticized by dense CO2 , then PMMA, PC, and lastly PVC was least affected by dense CO2 . Lopez-Periago et al.161, 162 showed that a therapeutic agent can be impregnated using dense CO2 and still maintain its activity. Lopez-Periago et al. developed a controlled drug delivery system by impregnating triflusal (2-acetyloxy-4-(trifluorometyl)benzoic acid) into PMMA. Upon depressurization, triflusal was molecularly dispersed in the PMMA matrix and pores were formed because dense CO2 acted as a porogen agent. Also, Natu et al.163 impregnated trimolol maleate (anti-glaucoma therapeutic agent) into a PCL blend to create a drug delivery device to battle glaucoma. Polycaprolactone has good biocompatibility and swells well in the presence of dense CO2 . Water and ethanol were also used as co-solvents for the impregnation of trimolol maleate into pure PCL and it was found that the best results were achieved with no co-solvent. On the other hand, for the PCL blends, water increased the uptake of the anti-glaucoma drug.

Dense Gas Processing of Polymers

157

The impregnation of polymers such as PMMA with highly CO2 -soluble compounds such as naphthalene has been reported by Uzer et al.99 The ability of dense CO2 to impregnate PMMA with naphthalene at pressures between 80–150 bar and temperature range of 35◦ C45◦ C was studied and it was shown that PMMA with naphthalene swelled more than non-impregnated PMMA.99 Kazarian was able to monitor the amount of impregnated ibuprofen in a PVP matrix using in situ FTIR spectroscopy.150 The method employed by Kazarian provides a quantitative approach to measure the partition coefficients and study the reactions in polymer matrices at high pressures and low temperatures. The use of FTIR spectroscopy is unique because it can provide a means of controlling the amount of drug and dyes being impregnated by dense CO2 into a polymer matrix. Kazarian found that the amount of drug that can be impregnated into a polymer matrix increases with increasing pressures and temperatures for CO2 -soluble drugs. For the textile industry, the inert nature of dense CO2 makes it a suitable carrier for the dye. The use of DGT can help reduce the amount of waste water normally produced in the dyeing process (i.e. textiles, polymers) by using CO2 as the carrier fluid rather than aqueous water or organic solvents. The success of polymer impregnation processes conducted in DGs is not dependent on the high solubility of the low molecular weight compounds in the DG phase. Favorable partitioning between the DG phase and polymer phase can determine the successful impregnations of compounds with low solubility in the DG solvent.164 Also, Kikic reports on the impregnation of dyes into polymers.96 The formation of metalpolymer complexes involves the chemical altering of the organo-metallic complex after impregnation. Again the inert nature of dense CO2 allows it to undertake the impregnation without altering or compromising the organo-metallic compound. The impregnation of organo-metallic complexes into and onto polymer substrates using dense CO2 has been reviewed elsewhere.165, 166 The properties of CO2 were tuned for the impregnation of bioactive heat-labile compounds such as enzymes. Recently, dense CO2 was used at low temperatures to immobilize beta-galactosidase onto PS to create a lab-on-a-chip at low temperatures.167 A novel use of dense CO2 has been for the formation of polymer blends. They can be created using dense CO2 to infuse CO2 -soluble monomers/initiators into a CO2 swollen polymer matrix, followed by subsequent polymerization of the impregnated monomers/initiators. 6.4 Polymer Blends Polymer blends are created with the purpose of improving physical properties such as yield strength, tensile strength, and impact strength. Polymer blends can also provide easier manufacturing (lowered viscosity), optimize economical factors, and adherence to environmental sanctions. Utraki168 breaks down the reasons for creating blends into two distinct groups of motivating factors: 1. Material-related reasons: • creating materials with a broad range of properties • improving physical properties such as ductility, flammability, and impact strength • cost-effectively customizing the material properties to suit an intended application • reusing plastic waste

158

R. Yoganathan et al. 2. Production-related reasons: • improved processability • ability to create customizable products, thus deliver better customer service • plant flexibility brought on by quick formulation changes • recyclability of blends

Not all polymers can blend easily, therefore many other factors must be considered before trying to create a polymer blend. A desired quality in polymer blends is the ability of the constituents to form a stable mixture and maintain its desired properties and therefore this is the distinguishing factor by which a blend is classified as either miscible or immiscible. Polymer blends are physical mixtures, and are further classified as miscible or immiscible blends. It is possible to have more than two polymers involved in a physical mixture; however, the focus of this report will be binary mixtures only. According to the International Union of Pure and Applied Chemistry (IUPAC) a polymer blend is defined as a macroscopically homogeneous mixture of two or more different species of polymer.169 Miscibility of a binary mixture refers to the ability of the two constituent polymers to form one homogeneous phase in all compositions. The IUPAC definition describes the miscible (or homogeneous) polymer blend, rather than the immiscible blend. Miscible blends are also referred to as compatible blends and homogeneous blends. In several texts the term polymer alloy has been used interchangeably with polymer blend; however in this report polymer alloy will rarely be used. More specifically, miscibility is the capability of a mixture to form a single phase over certain ranges of temperature, pressure, and composition.169 Miscibility is not only the ability of a blend to exhibit one single glass transition temperature (Tg ), but also the ability of the polymer blend to exhibit other single phase characteristics as measured by various analytical techniques such as light scattering, X-ray scattering, and neutron scattering. A polymer blend which exhibits immiscibility is referred to as an immiscible polymer blend, or heterogeneous polymer blend. The immiscibility may only be limited to a certain temperature, pressure or composition range, yet the polymer blend will still be defined as being immiscible. A document released by IUPAC in 2004 provides more comprehensive definitions of polymer blend terminology, and related terms.169 For the purposes of this report, the term polymer blend will define a mixture of two polymers which exhibits superior mechanical and/or chemical properties at atmospheric conditions and the conditions of the intended application. Thereby, the term polymer blend will be used to describe both miscible and immiscible blends unless otherwise specified. Mixtures of two polymers may be considered immiscible even if it is incompatible only over a small temperature or pressure range, therefore the mixture still may exhibit superior mechanical and/or chemical properties to the individual homopolymers. The composition of the polymer blend affects the polymer properties, for example a blend of PC and LLDPE having PC as the major component is tougher than stand-alone PC, whereas a PC blend with LLDPE as the major component is stiffer than stand-alone LLDPE.168 The majority of polymer blends are used for structural support applications which makes them dependent on the mechanical, thermal, and chemical properties. Chemical properties refer to the resistance of the polymer blend to various organic compounds. The modifiers listed in Table 5 are normally added as the minor component of the polymer blend to help compensate for the disadvantageous properties of the major polymer component. A new and growing field of applications for polymer blends is the biomedical field. Current FDA-approved polymers for implantation into the human body include PLA,

Dense Gas Processing of Polymers

159

Table 5 Polymers used to enhance specific polymer blend properties168,170 Property

Polymer modifier

Impact strength Stiffness Flame retardancy Chemical resistance Processability

ABS, ASA, SBS, SAN, PSO, HIPS PPE, PC, PSO PVC, siloxanes Thermoplastic polyesters, siloxanes, phosphates ABS, PSO, siloxanes

PLGA, PCL, PGA, PEEK, PMMA, PC, PTFE, and PU. These special biomedical polymers are distinct because of their behavior and properties within the human body. Polymer composites have also been widely used for biomedical applications (Table 6). 6.4.1 Bioblends. Within the wide scope of biomedical polymer blends, there is a special class of polymer blends termed bioblends. Bioblends are polymer blends where one or both components are biodegradable and/or biocompatible. They have the dual advantage of being biocompatible or biodegradable, and possessing suitable mechanical properties which may not be provided by the individual biocompatible/biodegradable components.152, 153,172 Bioblends are suitable for drug delivery, implants, cell cultures, tissue engineering, and non-biomedical related applications such as packaging and lubrication in the agricultural industry.172 As with normal blends, bioblend properties and successful application depend on the compatibility of the component polymers. The majority of bioblend work thus far has focused on polymer blends where the biodegradable polymer component was one of the polyesters PLA or PCL. Mohamed et al. have reported on the characterization of PLA/PS and PCL/PS bioblends.152, 153 Both PLA/PS and PCL/PS bioblends were observed as having Tm and crystallization temperature (Tc ) values which were composition-dependent. Composition-dependent transition temperatures are an indication of miscibility. Bioblend and polymer blend formation conventionally involve using high temperatures and labor-intensive melt extrusion and/or heavy amounts of organic solvents. The different types of polymer blend processing methods available will be discussed briefly in the following sections.

Table 6 Biomedical applications of polymer mixtures141,171 Application Bone Cement Bone Plate Screws Bone Replacement Cartilage Replacement Dental Implants Hip Replacement Intramedullary Nails Knee Replacement

Polymer Mixture CF/PMMA, Titanium/PMMA, UHMWPE/PMMA PLLA/PLDLA, CF/PS, CF/PEEK CF/PTFE, PET/PU, HA/HDPE PET/PU, PTFE/PU, CF/PTFE CF/C CF/PEEK, CF/PTFE, CF/PMMA CF/PEEK CF/UHMWPE

160

R. Yoganathan et al.

6.4.2 Conventional Blending Techniques. Physically mixing polymer components to create a polymer blend with desirable physical and chemical properties has required a special mixer. The parameters and conditions of the mixer affect and control the miscibility and morphology of the polymer blend. The most commonly used instruments are the internal mixer, single-screw extruder (SSE), twin-screw extruder, and injection molding device.168,170,172 There are also variations in mixers and mixing screw designs. Each of these mixers has been used widely for processing immiscible blends. The aim of processing immiscible blends is to increase the dispersion of one polymer in the other. In order to achieve a uniform dispersion, processing is conducted at elevated temperatures which allows for easier physical mixing in the melt. The internal mixer, SSE, twin-screw extruder, and injection molder all apply varying shear forces to the molten polymers. The conventional mixing techniques use elevated temperatures, organic solvents, and require effort to design the mixer/extruder. Well-designed mixers/extruders are sought to help minimize the temperature and use of solvents. Also, these conventional techniques have not been able to control the morphology of the dispersed phase in the immiscible polymer blends; therefore there is still much work to be done in this field. The field of polymer blend preparation is growing and research is ongoing to find new techniques to process both miscible and immiscible polymer blends. Many polymers are highly sensitive to changes in temperature and can easily degrade at the elevated temperatures currently employed by the conventional techniques. Also, the processing of polymer blends for the medical and biomedical industry has strict requirements regarding the residual solvents in the final product. Upon plasticization the free volume of the polymer matrix increases, the overall entropy increases, and the viscosity is reduced. A plasticized amorphous polymer (as a major component) is increasingly susceptible to mixing and entangling its chains with the chains of the other component. An increase in free volume and entropy makes entanglement of the component polymer chains easier. Plasticizing the entire polymer blend causes better dispersion of the minor polymer component and possibly increases its miscibility. Normally to increase miscibility, compatibilizers are required. 6.4.3 Compatibilizers. Compatibilizers such as mutually miscible co-solvents, copolymers, or chemical reactants can be added as a third component to increase the miscibility of immiscible polymer blends.168 Since the majority of polymer blends are immiscible, the use of dense CO2 rather than conventional compatibilizers such as toxic solvents and additives, would be an ideal and environmentally-friendly alternative. Tomasko et al.12 discuss how CO2 is an effective compatibilizer because of its ability to reduce the viscosity of one polymer component more than another, thus providing a way to produce finely dispersed domains of one component polymer in another. Finely dispersed polymer domains of the minor component strengthen the mechanical properties of the immiscible polymer blend. Moreover, phase separation tends to occur in immiscible blends when the domain phases are large. Various polymer blends processed and prepared in the presence of dense CO2 are listed in Tables 7, 8, and 9. In Table 9, polymer blends which were not classified as either miscible or immiscible can be found. The operating conditions and analytical methods for the various studies are also listed. Depending on the type of polymer blend, and the polymer components, different methods of analysis were used. Conventional methods of measuring miscibility include DSC and TGA; however, sometimes these methods do not provide definitive answers.

161

0–350 bar, 0–200◦ C 0.5 hr, 60 bar, 80◦ C

PS/PVME (poly(vinyl methyl ether)) PE/PVAc

PVP/PEG

—

40 bar(0–5 hr), 80 bar(0–0.5 hr), 35◦ C

160◦ C

24 hr, 90–160 bar, 35◦ C

LDPE/PS

PS/PET PS/PC

2–15 hr, 120 bar, 35◦ C

PMMA/iPP

12 hr, 120 bar, 35◦ C 16 hr, 214 bar, 120◦ C 14 hr, 100–200 bar, 120◦ C

—

4 hr, 244 bar, 40◦ C

PCTFE/PS

iPP (isotactic polypropylene)/PMMA, sPP (syndiotactic polypropylene)/PMMA PVC/PS PMMA/UHMWPE PP/PS

4 hr, 104.3 bar, 40◦ C

Processing Conditions

HDPE/PS

Polymer Blend

SEM, DSC, DMA DSC, TMAFM TEM, DSC, DMA, Universal Tensile Tester FTIR, SEM, DSC, WAXD SEM,Universal Tensile Tester — DSC, CP/MAS NMR, WAXD, Gravimetric sorption apparatus ATR-FTIR

DSC, Mass difference, FTIR, TEM DSC, FTIR, TEM, Mass Difference, energy dispersive X-ray SANS ATR-FTIR, XPS, DMA, Protein Adsorption test —

Analytical Techniques

Table 7 Miscible polymer systems prepared using dense CO2

183

75 15

97

182

179 180 181

42, 177, 179

176 73

75, 83, 173–175

83, 173–175

Reference

162

PCL/UHMWPE

1 hr, 357 bar, 32◦ C 1 hr, 357 bar, 32◦ C 1 hr, 307.95 bar, 90◦ C and 24 hr, 307.95 bar, 30◦ C 57–176 bar, 115◦ C

137.8 bar, 200◦ C

PS/rubber

PEO/PMMA PEO/PVAc PCL/PVC

1 hr, 137.8 bar, 200◦ C

—

Processing Conditions

PIB (poly(isobutylene))/PDMS (poly(dimethylsiloxane)), PnBMA(poly(n-butyl-methacrylate))/PDMS PS/PMMA

Polymer Blend

DSC, TGA, TMAFM

TEM, Rheology (single screw extruder), solubility measurements, TEM, Rheology (single screw extruder) DSC, SAXS DSC, SAXS DSC, DMA, SAXS

Model interfacial properties, solubility measurements

Analytical Techniques

Table 8 Immiscible polymer systems prepared using dense CO2

62

25, 89 24 88

187

185, 186

184

Reference

Dense Gas Processing of Polymers

163

Table 9 Polymer systems prepared using dense CO2 Polymer Blend

Processing Conditions

PS/PC

160◦ C

PS/PET

—

Analytical Techniques DSC,Gravimetric sorption apparatus, CP/MAS NMR, WAXD —

Reference 15 75

6.4.4 Immisicible Polymer Blends. Immiscible polymers normally exhibit poor adhesion towards one another; however, with dense CO2 acting as a common solvent, there is an increase in adhesion at the interface. Both Wang et al.184 and Elkovitch et al.94,185–188 use this increase in adhesion at the interface to blend two immiscible polymers and study the effect of dense CO2 more closely. Wang et al. studied the welding of immiscible polymers, which occurs primarily at the interface, with and without dense CO2 . A reduction in viscosity of both or one of the immiscible polymers increases the ability of both polymer components to blend. A reduction in melt viscosity and reduced glass transition temperature depends on the ability of the polymer component to adsorb CO2 and the ability of CO2 to diffuse into the bulk of the polymer. Elkovitch et al. report the use of dense CO2 to blend PS and PMMA by altering the melt viscosity ratio of the two polymer components. By trying to match the viscosities of the two polymers, they created a finer dispersion of the minor component in the major component. Polymer blend processing by dense CO2 can also apply to both miscible and immiscible polymer blends. Blending two miscible polymers in the presence of dense CO2 also has advantages as it is a medium with a reduced viscosity and liquid-like density. Many thermoplastics and engineering polymers have shown favorable plasticization effects in the presence of dense CO2 , and it is these polymers which are used in industrial grade polymer blends.168 Amorphous polymers are most susceptible to dense CO2 processing, as discussed in section 6.1. Plasticization of the polymer blend is required to help shape and mold the blend. Dense CO2 processing can increase the miscibility of the entire polymer blend. Ultimately, the multiple polymer processing functions of dense CO2 such as plasticization, foaming and impregnation, can provide an environmentally-friendly method of polymer blend preparation. Carbon dioxide can be used for both polymer blend processing and preparation because CO2 can plasticize and suppress the melting point of amorphous polymers. The preparation of polymer blends where the minor component is polymerized in the major component is easiest in dense CO2 when the major component is amorphous and plasticized. New methods of polymer blend preparation and processing have been developed by combining the different dense CO2 polymer processing and polymerization techniques. The new dense CO2 method of polymer blend preparation involves impregnating monomers/initiators/catalysts into a dense CO2 -swollen polymer using dense CO2 as the carrier, and then subsequently activating the monomers/initiators/catalysts. Monomers/initiators/catalysts which are CO2 -soluble can be impregnated into various amorphous polymers and then polymerized in situ. The depth of penetration into the bulk polymer depends on the operating conditions and affinity between the compounds. One of the earliest works on polymerization in dense CO2 -swollen polymers was the preparation

164

R. Yoganathan et al.

of HDPE/PS and CTFE/PS blends.83 Watkins et al.83 showed that PS, normally a CO2 insoluble polymer with low affinity for PCTFE, can be blended in PCTFE with the aid of dense CO2 . The PS monomer is CO2 -soluble, therefore dense CO2 was used to impregnate the PS monomer into a dense CO2 -swollen PCTFE substrate. Similarly, a blend of the immiscible polymers PS and PVC were prepared by polymerizing styrene in a CO2 -swollen PVC substrate.179 The use of dense CO2 to produce PS/PVC blends is most useful because PS and PVC are immiscible polymers. Polymer blends involving biomedically-relevant polymers have also been prepared using dense CO2 .62,180 Again, dense CO2 was able to blend PE and PVAc, two incompatible and biomedically-relevant polymers which with conventional techniques were difficult to blend.73 Hoshi et al.73 were able to make the surface of the PE/PVAc blends biocompatible by attaching CO2 -soluble bioactive compounds. Various biodegradable polymers (PCL, PLA, PGA, and PLGA) are also suitable for processing in dense CO2 . The mechanical properties of many biocompatible/biodegradable polymers are unsatisfactory and it is possible to reinforce them with a biocompatible minor component such as a thermoplastic and/or engineering polymer using dense CO2 technology. Many thermoplastics and engineering polymers such as PC, PE, PS, PU, and UHMWPE, are biocompatible and have been used in the biomedical field. The use of dense CO2 for processing biocompatible/biodegradable polymers and polymer blends is favorable because it leaves minimal to zero residual solvent. In fact, if a co-solvent was required, CO2 can extract it if it was a CO2 -soluble co-solvent. Many biocompatible polymers foam in the presence of dense CO2 and therefore porous bioblends can be prepared. Depending upon how deep the dense CO2 penetrates the polymer matrix and the depressurization rate, the pore size, and interconnectivity may vary. Micropores and nanopores with and without interconnectivity have been reported in biodegradable polymers.189, 190 Tomasko et al.12 discussed the demixing of polymer components; however, this only occurs for CO2 -assisted extrusion. The outcome of demixing are unfoamed polymer blends, and therefore are not a concern unless a foamed product was originally sought. By controlling the depressurization rate, it is possible to limit the demixing. Foamed polymer blends are highly desired for biomedical applications because of their potential use as scaffolds. 6.5 Sterilization Sterilization of heat-sensitive polymeric biomaterials is problematic. The current sterilization technologies are steam, ethylene oxide (EtO), γ -radiation, and hydrogen peroxide (H2 O2 ) sterilization. The cheapest and easiest is steam sterilization, which is conducted at 121◦ C. The drawbacks of steam sterilization are the high temperatures and compromising effects of heat and vapor on the manufactured surfaces. Ethylene oxide and γ -radiation work at lower temperatures, and so are more suitable for heat-sensitive polymers. On the other hand, EtO is toxic and γ -radiation can affect the mechanical properties of polymers. The use of dense CO2 as a sterilization media is being explored because it is nonflammable, non-toxic, and has low operating conditions (critical point). Alternative gases (Ar, N2 ,N2 O) have been used to sterilize various micro-bacteria. For example, Fraser sterilized E.coli at 37◦ C and pressures between 1.7–6.2 MPa.191 Fraser found that CO2 was the most effective of the three gases. Novasterilis has recently marketed a dense CO2 sterilization instrument, Nova2200, and won the Small Business Award as part of the 2007 Presidential Green Chemistry Challenge Awards.192 The sterilization conditions of CO2 on gram positive, gram negative bacteria and spores have been reported by Zhang et al.193 Examples of gram positive bacteria include Bacillus

Dense Gas Processing of Polymers

165

subtilis and Staphylococcus aureus, and examples of gram negative bacteria include E.coli and Pseudomonas aeruginosa. Generally, dense CO2 is more potent on vegetative cells (gram positive and gram negative) than spores. Spores are more resistant to sterilization than vegetative cells because they can withstand heat, UV, free radicals and chemicals.193 The ability of a sterilant to deactivate spores is the industrial test for equipment sterilization. The exact mechanism which influences the ability of CO2 to terminally sterilize bacteria is still being studied. The following factors are known to affect CO2 sterilization efficiency:193 • • • • • • • •

pressure temperature depressurization rate pressure cycling growth medium additives treatment time deactivation kinetics

The sterilizing temperature and pressure of dense CO2 are species specific. The two main CO2 cell deactivation mechanisms are classified as follows:107,193 1. Mechanical cell rupture • during pressurization • during depressurization 2. Physiological deactivation • lowers intracellular pH • enzyme denaturation • extraction of cell contents Spores may be deactivated by alternative mechanisms. As a spore is dehydrated (60–70% less water than vegetative cells), it is resistant to the mass transfer abilities of dense CO2 . Not many studies have been conducted on the sterilization of spores by dense CO2 .107,193–201 The sterilizing effect of dense CO2 on polymers was reported by Dillow et al.105 and Jimenez et al.102 Dillow et al. prepared PLGA and PLA microspheres, and exposed them to vegetative cells only. The dense CO2 sterilization of the microspheres was conducted with and without water as a co-solvent at 34◦ C and 140 bar. Water helped reduce the sterilization time. The PLGA and PLA microspheres underwent FTIR analysis before and after dense CO2 , and showed no changes in the spectra. Jimenez et al. sterilized a poly(acrylic acidco-acrylamide) hydrogel at 40◦ C and 276 bar for 4 hrs. They also introduced hydrogen peroxide into the process, and found that similar bacterial log reduction was achieved with and without hydrogen peroxide. Dense CO2 is seen as a viable alternative to current sterilization technologies because it does not diminish the mechanical properties of the polymers.13 The tensile strength and modulus of elasticity of medical-grade polymers (UHMWPE, PS, LDPE, HDPE, PMMA, PU, ABS, PPO, PVC, and PC) upon dense CO2 treatment, were measured. Of the tested polymers, the tensile strength of PC, PVC, and ABS had decreased after dense CO2 exposure while the PVC modulus of elasticity statistically increased. The field of dense CO2 sterilization of polymers is still growing. Dense CO2 is a proven sterilization technology for vegetative cells and companies such as NovaSterilis have brought it to market. While the effects of dense CO2 on bacteria and polymers separately have been observed, there is limited literature on the sterilizing effect of CO2

166

R. Yoganathan et al.

on contaminated polymers. Dense CO2 treatment can preserve the physical properties of medical-grade polymers which opens the way to testing polymer sterilization via dense CO2 .

7. Conclusions and Recommendations Dense CO2 can plasticize various thermoplastics and engineering polymers, and can act as a viable reaction media for various polymerization reactions. During plasticization, dense CO2 is able to diffuse into the polymer matrix and reduce the viscosity of the polymer melt, making it easier to process. Upon depressurization the CO2 -soluble compounds become entrapped in the polymer substrate, and can also cause the polymer substrate to foam. The advantageous mass transfer properties of dense CO2 allows it to impregnate amorphous polymers with various CO2 -soluble compounds, including monomers, catalysts, initiators, and various bio-actives that have shown affinity for CO2 . Biocompatible/biodegradable polymers are susceptible to dense CO2 plasticization and thereby impregnation. The resulting products can be used for tissue engineering applications such as scaffolds and controlled drug delivery devices. Dense CO2 processing of miscible and immiscible polymer blends is feasible. For immiscible polymer blends, dense CO2 has provided a better dispersion of the minor polymer in the major polymer component, thus increasing mechanical properties and miscibility than conventional methods. Research into the preparation of polymer blends via dense CO2 has yielded a method of impregnating monomers into dense CO2 -swollen polymers which act as the substrate. The entrapped monomers/initiators/catalysts can be activated by changes in temperature and pressure. Dense CO2 has also shown to be successfully used to create miscible blends of two incompatible polymers. More recently, bio-actives have been immobilized onto polymer surfaces using dense CO2 . Enzymes too have been used to catalyze polymerization reactions in the presence of dense CO2 . Dense gas processing of polymers is moving toward biomedical applications. In-vivo applications of biomedical polymers are very sensitive to residual solvents, thus making the use of an environmentally-friendly and non-toxic medium such as CO2 very favorable. The extraction of residual solvents and sterilization of bacteria make dense CO2 applicable for medical-grade polymer processing. Bacteria, both gram positive and gram negative, are easily sterilized by dense CO2 . Also, the mechanical properties of medical-grade polymers have not been affected by dense CO2 exposure, therefore sterilization of inoculated medicalgrade polymers is feasible using dense CO2 . The field of medical polymers, biomedical polymers, and bioblends have benefited from the use of a non-toxic, inert, and environmentally-friendly processing media such as dense CO2 for a variety of polymer processing applications such as polymerization, blending, foaming, impregnation, and sterilization.

Acknowledgments The authors would like to thank the Australian Research Council for financial support from the Discovery Grant (Reference DP0665514).

General Abbreviations DG DGT FDA

Dense gas Dense gas technology Food and drug administration

Dense Gas Processing of Polymers IUPAC SCF

International union of pure and applied chemistry Supercritical fluid

Polymer Abbreviations ABS AIBN ASA C CA CF CFC DVB HDPE iPP LDPE LLDPE PC PCL PDMS PE PEEK PEG PEO PET PGA PHB PIB PLA PLGA PMMA PPO PS PSO PTFE PU PVA PVAc PVC PVME PVP SBS sPP UHMWPE

Acrylonitrile butadiene styrene Azobisisobutyronitrile Acrylonitrile styrene acrylate Carbon Cellulose acetate Carbon fibers Chlorofluorocarbons Divinyl benzene High density polyethylene Isotactic polypropylene Low density polyethylene Linear low density polyethylene Polycarbonate Polycaprolactone Poly(dimethylsiloxane) Polyethylene Polyaryletheretherketone Poly(ethylene glycol) Poly(ethylene oxide) Poly(ethyl teraphthalate) Poly(glycolic acid) Poly(hydroxyl butyrate) Poly(isobutylene) Poly(lactic acid) Poly(lactic-co-glycolic acid) Poly(methyl methacrylate) Poly(propylene oxide) Polystyrene Poly(styrene oxide) Polytetrafluoroethylene Polyurethane Poly(vinyl alcohol) Poly(vinyl acetate) Poly(vinyl chloride) Poly(vinyl methyl ether) Poly(vinyl)propylene Styrene-butadiene-styrene Syndiotactic polypropylene Ultra high molecular weight polyethylene

Analytical Technique Abbreviations CPMAS NMR DMA

Cross-polarization magic angle spinning NMR Dynamic mechanical analysis

167

168 DSC NMR SANS SAXS SEM TEM TGA TMAFM XPS

R. Yoganathan et al. Differential scanning calorimetry Nuclear magnetic resonance Small-angle neutron scattering Small-angle X-ray scattering Scanning electron microscopy Transmission electron microscopy Thermogravimetric analysis Tapping mode atomic force microscopy X-ray photoelectron spectroscopy

References 1. Allen, D. T.; Shonnard, D. R. “Green engineering: Environmentally conscious design of chemical processes and products,” AIChE Journal, 2001, 47, 9. 2. Kemmere, M. F.; Meyer, T. Supercritical Carbon Dioxide in Polymer Reaction Engineering; Wiley-VCH. 2005. 3. Jachuck, R. “Process intensification for green chemistry,” In Handbook of Green Chemistry and Technology; 2002; pp 366–371. 4. Kemmere, M. F. “Supercritical carbon dioxide for sustainable polymer processes,” In Supercritical Carbon Dioxide in Polymer Reaction Engineering; Kemmere, M. F.; Meyer, T., Eds.; Wiley-VCH. 2005; pp 1–14. 5. Ramshaw, C. “Process intensification: A game for n players,” Chem. Engineer (London), 1985, 416, 30–33. 6. Stankiewicz, A. I.; Moulijn, J. A. “Process intensification: Transforming chemical engineering,” CEP, 2000, 96(1), 22–33. 7. Kazarian, S. G. “Polymer processing with supercritical fluids,” Polym. Sci. Ser. C Sel. Top., 2000, 42, 78–101. 8. Eckert, C. A.; Knutson, B. L.; Debenedetti, P. G. “Supercritical fluids as solvents for chemical and materials processing,” Nature, 1996, 383(6598), 313–318. 9. Jessop, P.G.; Leitner, W. Chemical Synthesis using Supercritical Fluids; Wiley-VCH: New York, 1999. 10. Cooper, A. I.; Wood, C. D.; Holmes, A. B. “Synthesis of well-defined macroporous polymer monoliths by sol-gel polymerization in supercritical CO2 ,” Ind. Eng. Chem. Res., 2000, 39, 4741–4744. 11. Kikic, I.; Vecchione, F.; Alessi, P.; Cortesi, A.; Eva, F.; Elvassore, N. “Polymer plasticization using supercritical carbon dioxide: Experiment and modeling,” Ind. Eng. Chem. Res., 2003, 42(13), 3022–3029. 12. Tomasko, D. L.; Li, H.; Liu, D.; Han, X.; Wingert, M. J.; Lee, L. J.; Koelling, K. W. “A review of CO2 applications in the processing of polymers,” Ind. Eng. Chem. Res., 2003, 42, 6431–6456. 13. Jimenez, A.; Thompson, G. L.; Matthews, M. A.; Davis, T. A.; Crocker, K.; Lyons, J. S.; Trapotsis, A. “Compatibility of medical-grade polymers with dense CO2 ,” J. Supercrit. Fluids, 2007, 42, 366–372. 14. Kiran, E. “Polymer miscibility, phase separation, morphological modifications and polymorphic transformations in dense fluids,” J. Supercrit. Fluids, 2009, 47, 466–483. 15. Kato, S.; Tsujita, Y.; Yoshimizu, H.; Kinoshita, T.; Higgins, J. S. “Characterization and CO2 sorption behavior of polystyrene/polycarbonate blend system,” Polym., 1997, 38, 2807–2811. 16. Kerschner, J. L.; Jureller, S. H.; Harris, R. “Polymer modification in supercritical carbon dioxide,” Book of Abstracts, 211th ACS National Meeting, New Orleans, LA, March 24–28, 1996; pp 151. 17. Lepilleur, C.; Beckman, E. J.; Schonemann, H.; Krukonis, V. J. “Effect of molecular architecture on the phase behavior of fluoroether-functional graft copolymers in supercritical CO2 ,” Fluid Phase Equilib., 1997, 134(1–2), 285–305.

Dense Gas Processing of Polymers

169

18. O’Neill, M. L.; Cao, Q.; Fang, M.; Johnston, K. P.; Wilkinson, S. P.; Smith, C. D.; Kerschner, J. L.; Jureller, S. H. “Solubility of homopolymers and copolymers in carbon dioxide,” Ind. Eng. Chem. Res., 1998, 37(8), 3067–3079. 19. Ohshima, M. “Effect of supercritical CO2 on polymer properties,” Nippon Gomu Kyokaishi, 2004, 77, 266–271. 20. Pasquali, I.; Andanson, J. M.; Kazarian, S. G.; Bettini, R. “Measurement of CO2 sorption and PEG 1500 swelling by ATR-IR spectroscopy,” J. Supercrit. Fluids, 2008, 45, 384–390. 21. Sadowski, G. Supercritical Carbon Dioxide in Polymer Reaction Engineering: Phase Behavior of Polymer Systems in High-Pressure Carbon Dioxide; Wiley-VCH Verlag 2005. 22. Saeki, S. “Calculation of phase diagram of polymer solution over supercritical fluid region based on the Patterson-Delmas theory of corresponding state,” Fluid Phase Equilib., 1997, 136(1–2), 87–92. 23. Saraf, V. P.; Kiran, E. “Supercritical fluid-polymer interactions: phase equilibrium data for solutions of polystyrenes in n-butane and n-pentane,” Polym., 1988, 29(11), 2061–2065. 24. Shieh, Y. T.; Lin, Y. G.; Chen, H. L. “Effect of supercritical CO2 on phase structure of PEO/PVAc blends evaluated from SAXS absolute intensity measurement,” Polym., 2002, 43, 3691– 3698. 25. Shieh, Y. T.; Liu, K. H.; Lin, T. L. “Effect of supercritical CO2 on morphology of compatible crystalline/amorphous PEO/PMMA blends,” J. Supercrit. Fluids, 2004, 28, 101–112. 26. Aklonis, J. J.; Amis, E. J. “Processing of polymers using supercritical fluids: Technical report” Dep. Chem., Univ. South. California, Los Angeles, CA, USA., 1990. CAN 115:73113 38–2 Plastics Fabrication and Uses. 27. Alsoy, S.; Duda, J. L. “Processing of polymers with supercritical fluids,” Chem. Eng.Technol., 1999, 22, 971–973. 28. Arora, K. A.; Lesser, A. J.; McCarthy, T. J. “Synthesis, characterization, and expansion of poly(tetrafluoroethylene-co-hexafluoropropylene)/polystyrene blends processed in supercritical carbon dioxide,” Macromol., 1999, 32, 2562–2568. 29. Balashova, I. M.; Russell, M. C.; Danner, R. P.; Duda, J. L. “Solubility of carbon dioxide and cosolvents in a photoresist polymer,” J. Appl. Polym. Sci., 2008, 108, 1865–1870. 30. Briscoe, B. J.; Lorge, O.; Wajs, A.; Dang, P. “Carbon dioxide poly(vinylidene fluoride) interactions at high pressure,” J. Polym. Sci., Part B: Polym. Phys., 1998, 36(13), 2435–2447. 31. Cansell, F.; Rey, S.; Beslin, P. “Thermodynamic aspects of supercritical fluids processing: applications to polymers and wastes treatment,” Revuede l’Institut Francais du Petrole, 1998, 53, 71–98. 32. Fleming, O. S.; Kazarian, S. G. “Polymer processing with supercritical fluids.” Supercritical Carbon Dioxide In Polymer Reaction Engineering; Kemmere, M. F.; Meyer, T., Eds.; WileyVCH. 2005; pp 205–238. 33. Gregorowicz, J. “Phase behavior of l-lactide in supercritical carbon dioxide at high pressures,” J. Supercrit. Fluids, 2008. 46(2), 105–111. 34. Guadagno, T.; Kazarian, S. G. “High-pressure CO2 -expanded solvents: Simultaneous measurement of CO2 sorption and swelling of liquid polymers with in-situ near-ir spectroscopy,” J. Phys. Chem. B, 2004, 108, 13995–13999. 35. Gupta, R. R.; Lavery, K. A.; Francis, T. J.; Webster, J. R. P.; Smith, G. S.; Russell, T. P.; Watkins, J. J. “Self-diffusion of polystyrene in a CO2 -swollen polystyrene matrix: A real time study using neutron reflectivity,” Macromol., 2003, 36, fa346–352. 36. Jiang, C. Y.; Shen, Y. F.; Pan, Q. M. “Review of enhancement effect of supercritical carbon dioxide on melt phase polycondensation,” Gaofenzi Cailiao Kexue Yu Gongcheng/Polym. Mat. Sci. Eng., 2007, 23(5), 28–32. 37. Kasturirangan, A.; Teja, A. S. “Phase behavior of CO2 + biopolymer and CO2 +fluoropolymer systems,” Fluid Phase Equilib., 2007, 261, 64–68. 38. Kilic, S.; Michalik, S.; Wang, Y.; Johnson, J. K.; Enick, R. M.; Beckman, E. J. “Phase behavior of oxygen-containing polymers in CO2 ,” Macromol., 2007, 40, 1332–1341.

170

R. Yoganathan et al.

39. Shen, Z.; McHugh, M. A.; Xu, J.; Belardi, J.; Kilic, S.; Mesiano, A.; Bane, S.; Karnikas, C.; Beckman, E.; Enick, R. “CO2 -solubility of oligomers and polymers that contain the carbonyl group,” Polym., 2003, 44, 1491–1498. 40. Fried, J. R.; Hu, N. “The molecular basis of CO2 interaction with polymers containing fluorinated groups: Computational chemistry of model compounds and molecular simulation of poly [bis(2,2,2-trifluoroethoxy)phosphazene],” Polym., 2003, 44, 4363–4372. 41. Tapriyal, D.; Wang, Y.; Enick, R. M.; Johnson, J. K.; Crosthwaite, J.; Thies, M. C.; Paik, I. H.; Hamilton, A. D. “Poly(vinyl acetate), poly((1-o-(vinyloxy) ethyl-2,3,4,6-tetra-o-acetyl–dglucopyranoside) and amorphous poly(lactic acid) are the most CO2 -soluble oxygenated hydrocarbon-based polymers,” J. Supercrit. Fluids, 2008, 46, 252–257. 42. Ishikawa, H.; Sawaguchi, T.; Muroga, Y.; Hoshi, T.; Hagiwara, T.; Yano, S. “Plasticization and miscibilization of syndiotactic polystyrene/poly(methyl methacrylate) blend by supercritical carbon dioxide fluid,” In Polym. Prepr. (54th SPSJ Symp. Macromol., Japan), 2005, 54, 3476–3477. 43. Kazarian, S. G.; Brantley, N. H.; West, B. L.; Vincent, M. F.; Eckert, C. A. “In situ spectroscopy of polymers subjected to supercritical CO2 : Plasticization and dye impregnation,” Appl. Spectrosc., 1997, 51(4), 491–494. 44. Chiou, J. S.; Barlow, J. W.; Paul, D. R. “Plasticization of glassy polymers by carbon dioxide,” J. Appl. Polym. Sci., 1985, 30, 2633–42. 45. Kusmanto, F.; Billham, M.; Hornsby, P. “Polymer plasticization using supercritical carbon dioxide,” J. Vinyl Add. Tech., 2008, 14(4), 163–166. 46. Liu, D.; Li, H.; Tomasko, D.; Verreck, G.; Arien, A.; Jef, P.; Marcus, B. “CO2 -induced plasticization and viscosity reduction in pharmaceutical polymers,” 2005 AIChE Annual Meeting and Fall Showcase (Cincinnati, OH), 1411. 47. Madsen, L. A. “Plasticization of poly(ethylene oxide) in fluid CO2 measured by in-situ NMR,” Macromol., 2006, 39(4), 1483–1487. 48. Moriyoshi, T. “ Plasticization and processing polymeric materials with high-pressure fluids,” Rev. High Press. Sci. Tech./Koatsuryoku No Kagaku To Gijutsu, 2003, 13(2), 111–118. 49. Hojo, H.; Findley, W. “Effect of gas diffusion on creep behavior of polycarbonate,” Polym. Eng. Sci., 1973, 13, 255–26. 50. Shende, R. V.; Kline, M.; Lombardo, S. J. “Effects of supercritical extraction on the plasticization of poly(vinyl butyral) and dioctyl phthalates,” J. Supercrit. Fluids, 2004, 28(1), 113–120. 51. Shenoy, S. L.; Fujiwara, T.; Wynne, K. J. “Quantifying plasticization and melting behavior of poly(vinylidine fluoride) in supercritical CO2 utilizing a linear variable differential transformer,” Macromol., 2003, 36(9), 3380–3385. 52. Barry, J. J. A.; Nazhat, S. N.; Rose, F. R. A. J.; Hainsworth, A. H.; Scotchford, C. A.; Howdle, S. M. “Supercritical carbon dioxide foaming of elastomer/heterocyclic methacrylate blends as scaffolds for tissue engineering,” J. Mater. Chem., 2005, 15, 4881–4888. 53. Guo, Z.; Wingert, M. J.; Shen, J.; Lee, L. J.; Tomasko, D. L. “Foaming dynamics of immiscible polymer blends,” In Annual Technical Conference—ANTEC, Conference Proceedings, Society of Plastics Engineers Annual Technical Conference: Plastics Encounter at ANTEC 2007, 5, 2970–2975. 54. Han, X.; Lee, L. J.; Tomasko, D. L. “CO2 foaming of polymer nanocomposite blends,” Aust. J. Chem., 2005, 58, 492–503. 55. Han, X.; Shen, J.; Huang, H.; Tomasko, D. L.; Lee, L. J. “CO2 foaming based on polystyrene/poly(methyl methacrylate) blend and nanoclay,” Polym. Eng. Sci., 2007, 47, 103–111. 56. Huang, H. X.; Wang, J. K.; Sun, X. H. “Microcellular foaming of PP/HDPE blend and its nanocomposites using supercritical fluid,” Gongneng Gaofenzi Xuebao, 2008, 21, 1–5. 57. Mascia, L.; Re, G. D.; Ponti, P. P.; Bologna, S.; Giacomo, G. D.; Haworth, B. “Crystallization effects on autoclave foaming of polycarbonate using supercritical carbon dioxide,” Adv. Polym. Technol., 2006, 25, 225–235.

Dense Gas Processing of Polymers

171

58. Nie, H.; Lee, L. Y.; Tong, H.; Wang, C. H. “PLGA/chitosan composites from a combination of spray drying and supercritical fluid foaming techniques: New carriers for DNA delivery,” J. Controlled Release, 2008, 129, 207–214. 59. Tsivintzelis, I.; Angelopoulou, A. G.; Panayiotou, C. “Foaming of polymers with supercritical CO2 : An experimental and theoretical study,” Polym., 2007, 48, 5928–5939. 60. Wang, J. K.; Huang, H. X.; Lin, D. H. “Effect of viscoelasticity on the cellular structure during PP/PE blends microcellular foaming using supercritical fluid,” Suliao, 2007, 36, 58– 62. 61. Zhu, X. H.; Lee, L. Y.; Jackson, J. S. H.; Tong, Y. W.; Wang, C. H. “Characterization of porous poly(d,l-lactic-co-glycolic acid) sponges fabricated by supercritical CO2 gas-foaming method as a scaffold for three-dimensional growth of hep3b cells,” Biotechnol. Bioeng., 2008, 100, 998–1009. 62. Busby, A. J.; Zhang, J.; Roberts, C. J.; Lester, E.; Howdle, S. M. “Novel nanostructured polymeric composites of polycaprolactone and ultra-high molecular weight polyethylene via a supercritical-fluid route,” Adv. Mater., 2005, 17, 364–367. 63. Chen, C. F.; Chang, C. S.; Chen, Y. P.; Lin, T. S.; Su, C. Y.; Lee, S. Y. “Applications of supercritical fluid in alloplastic bone graft: A novel method and in vitro tests,” Ind. Eng. Chem. Res., 2006, 45, 3400–3405. 64. Ajzenberg, N.; Trabelsi, F.; Recasens, F. “What’s new in industrial polymerization with supercritical solvents? A short review,” Chem. Eng. Technol., 2000, 23, 829–839. 65. Bergeot, V.; Tassaing, T.; Besnard, M.; Cansell, F.; Mingotaud, A. F. “Anionic ring-opening polymerization of e-caprolactone in supercritical carbon dioxide: parameters influencing the reactivity,” J. Supercrit. Fluids, 2004, 28, 249–261. 66. Bratton, D.; Brown, M.; Howdle, S. M. “Ti(II) ethyl hexanoate catalyzed precipitation polymerization of e-caprolactone in supercritical carbon dioxide,” Macromol., 2005, 38, 1190– 1195. 67. Canelas, D. A.; Desimone, J. M. “Polymerizations in liquid and supercritical carbon dioxide,” Adv. Polym. Sci., 1997, 133, 103–140. 68. Chatzidoukas, C.; Pladis, P.; Kiparissides, C. “Mathematical modeling of dispersion polymerization of methyl methacrylate in supercritical carbon dioxide,” Ind. Eng. Chem. Res., 2003, 42, 743–751. 69. Christian, P.; Howdle, S. M.; Irvine, D. J. “Dispersion polymerization of methyl methacrylate in supercritical carbon dioxide with a monofunctional pseudo-graft stabilizer,” Macromol., 2000, 33, 237–239. 70. Darensbourg, D. J.; Stafford, N. W.; Katsurao, T. “Supercritical carbon dioxide as solvent for the copolymerization of carbon dioxide and propylene oxide using a heterogeneous zinc carboxylate catalyst,” J. Mol. Catal. A: Chem., 1995, 104, L1–L4. 71. DeSimone, J. M.; Maury, E. E.; Menceloglu, Y. Z.; McClain, J. B.; Romack, T. J.; Combes, J. R. “Dispersion polymerizations in supercritical carbon dioxide,” Sci., 1994, 265, 356. 72. Gross, S. M.; Roberts, G. W.; Kiserow, D. J.; DeSimone, J. M. “Synthesis of high molecular weight polycarbonate by solid-state polymerization,” Macromol. 34, 3916–3920. 73. Hoshi, T.; Sawaguchi, T.; Konno, T.; Takai, M.; Ishihara, K. “Preparation of molecular dispersed polymer blend composed of polyethylene and poly(vinyl acetate) by in situ polymerization of vinyl acetate using supercritical carbon dioxide,” Polym., 2007, 48, 1573–1580. 74. Kim, J.; Roberts, G. W.; Kiserow, D. J. “Effect of prepolymer molecular weight on solid state polymerization of poly(bisphenol a carbonate) with nitrogen as a sweep fluid,” J. Polym. Sci. Part A: Polym. Chem., 2008, 46, 4959–4969. 75. Li, D.; Han, B. X.; Yan, H. K. “Polymerization of styrene in supercritical CO2 -swollen poly(ethylene terephthlate),” Chin. Chem. Lett., 1999, 10, 871–872. 76. Lu, S.; Zhang, Z.; Nawaby, V.; Day, M. “A kinetic investigation of the in situ polymerization of methyl methacrylate under supercritical fluid CO2 conditions using high-pressure DSC,” J. Appl. Polym. Sci., 2004, 93:1236–1239.

172

R. Yoganathan et al.

77. Mueller, P. A.; Bonavoglia, B.; Storti, G.; Morbidelli, M. “Heterogeneous Polymerization in Supercritical Carbon Dioxide,” Supercritical Carbon Dioxide In Polymer Reaction Engineering; Kemmere, M. F.; Meyer, T., Eds.; Wiley-VCH. 2005, pp 105–138. 78. Scholsky, K. M. “Polymerization reactions at high pressure and supercritical conditions,” J. Supercrit. Fluids, 1993, 6, 103–27. 79. Shieh, Y. T.; Lin, Y. T. “Transesterification and polymerization reactions of aliphatic polyesters in supercritical CO2 fluids without the presence of a catalyst,” Eur. Polym. J., 2007, 43, 1847– 1856. 80. Stassin, F.; Halleux, O.; Jerome, R. “Ring-opening polymerization of e-caprolactone in supercritical carbon dioxide,” Macromol., 2001, 34, 775–781. 81. Tan, B.; Woods, H. M.; Licence, P.; Howdle, S. M.; Cooper, A. I. “Synthesis and CO2 solubility studies of poly(ether carbonate)s and poly(ether ester)s produced by step growth polymerization,” Macromol., 2005, 38, 1691–1698. 82. Thurecht, K. J.; Heise, A.; Geus, M. D.; Villarroya, S.; Zhou, J.; Wyatt, M. F.; Howdle, S. M. “Kinetics of enzymatic ring-opening polymerization of e-caprolactone in supercritical carbon dioxide,” Macromol., 2006, 39, 7967–7972. 83. Watkins, J. J.; McCarthy, T. J. “Polymerization in supercritical fluid-swollen polymers: A new route to polymer blends,” Macromol., 1994, 27, 4845–7. 84. Wood, C. D.; Cooper, A. I. “Synthesis of macroporous polymer beads by suspension polymerization using supercritical carbon dioxide as a pressure-adjustable porogen,” Macromol., 2001, 34, 5–8. 85. Kung, E.; Arora, K.; Hayes, H.; Rajagopalan, P.; McCarthy, T. J. “Reactions in supercritical fluid—swollen polymers: A new route to polymer blends, modified polymers and composite foams,” Book of Abstracts, 216th ACS National Meeting 1998, PMSE-063. 86. Peng, Q.; Xu, Q.; Wang, Y. D.; Liu, M. Y.; Zhao, Q. X. “Preparation of poly(acrylic acid)/nylon 1313 and polystyrene/nylon 3,3 blends through polymerization in supercritical CO2 -swelling nylon 3,3,” Yingyong Huaxue, 2005, 22:337–339. 87. Procter and Gamble. Manufacture of homogeneous polymeric blends for super critical fluid solution, 19920326. 1994. 88. Shieh, Y. T.; Yang, H. S.; Chen, H. L.; Lin, T. L. “Nonisothermal crystallization of compatible PCL/PVC blends under supercritical CO2 ,” Polym. J., 2005, 37, 932–938. 89. Shieh, Y. T.; Liu, K. H. “Conformational changes of poly(ethylene oxide) in poly(ethylene oxide)/poly(methyl methacrylate) blends by supercritical carbon dioxide,” J. Polym. Sci. Part B: Polym. Phys., 2004, 42, 2479–2489. 90. Singh, J.; Somavarapu, S.; Bramwell, V.; McHugh, C.; Shah, K.; Alpar, O. “Comparative study of nanospheres composed of various blends of poly-(dl-lactic-acid), poly(lactic-coglycolic acid) and poly-e-caprolactone polymers as vaccine adjuvants for diphtheria toxoid,” Proceedings—28th International Symposium on Controlled Release of Bioactive Materials and 4th Consumer & Diversified Products Conference, San Diego, CA, United States, June 23–27, 2001, 2:1081–1082. 91. Watkins, J. J.; Brown, G. D.; Pollard, M. A. “Transitions in polymer blends and block copolymers induced by phase-selective dilation with supercritical CO2 ,” Polym. Mater. Sci. Eng., 1998, 79, 386–387. 92. Watkins, J. J.; Brown, G. D.; Pollard, M. A.; Ramachandrarao, V. S.; Russell, T. P. “Phase transitions in polymer blends and block copolymers induced by selective dilation with supercritical CO2 ,” NATO ASI Ser., Ser. E, 2000, 366, 277–289. 93. Elkovitch, M. D. “Supercritical fluid assisted polymer blending,” PhD thesis: Ohio State University, 2001. 94. Elkovitch, M. D.; Lee, L. J.; Tomasko, D. L. “Supercritical fluid assisted polymer blending,” In Anon, editor, Annual Technical Conference—ANTEC, Conference Proceedings 1998 56th Annual Technical Conference, 2, 2538–2541. 95. Shen, Z.; Huvard, G. S.; Warriner, C. S.; Mc Hugh, M.; Banyasz, J. L.; Mishra, M. K. “Carbon dioxide-assisted fiber impregnation,” Polym., 2008, 49, 1579–1586.

Dense Gas Processing of Polymers

173

96. Kikic, I.; Vecchione, F. “Supercritical impregnation of polymers,” Curr. Opin. Solid State Mater. Sci., 2003, 7, 399–405. 97. Li, D.; Han, B. “Impregnation of polyethylene (PE) with styrene using supercritical CO2 as the swelling agent and preparation of PE/polystyrene composites,” Ind. Eng. Chem. Res., 2000, 39, 4506–4509. 98. Muth, O.; Hirth, T.; Vogel, H. “Polymer modification by supercritical impregnation,” J. Supercrit. Fluids, 2000, 17, 65–72. 99. Uzer, S.; Akman, U.; Hortacsu, O. “Polymer swelling and impregnation using supercritical CO2 : A model-component study towards producing controlled-release drugs,” J. Supercrit. Fluids, 2006, 38, 119–128. 100. Cooper, A. I. “Synthesis of advanced materials using supercritical fluids,” Supercritical Carbon Dioxide In Polymer Reaction Engineering; Kemmere, M. F.; Meyer, T., Eds.; Wiley-VCH. 2005, pp 239–254. 101. Cooper, A. I.; Hems, W. P.; Holmes, A. B. “Synthesis of cross-linked polymer microspheres in supercritical carbon dioxide,” Macromole. Rapid Commun., 1998, 19, 353–357. 102. DeSimone, J. M.; Guan, Z.; Elsbernd, C. S. “Synthesis of fluoropolymers in supercritical carbon dioxide,” Sci., 1992, 257, 945–7. 103. Ginty, P. J.; Howard, D.; Upton, C. E.; Barry, J. J. A.; Rose, F. R. A. J.; Shakesheff, K. M.; Howdle, S. M. “A supercritical CO2 injection system for the production of polymer/mammalian cell composites,” J. Supercrit. Fluids, 2008, 43, 535–541. 104. Kazarian, S. G.; Flichy, N. M. B.; Coombs, D.; Poulter, G. “Potential of ATR-IR spectroscopy in applications to supercritical fluids and liquefied gases,” Am. Lab., 2001, 33. 105. Dillow, A. K.; Dehghani, F.; Hrkach, J. S.; Foster, N. R.; Langer, R. “Bacterial inactivation by using near-and supercritical carbon dioxide,” PNAS, 1999, 96, 10344– 10348. 106. Sproule, T. L.; Lee, J. A.; Li, H.; Lannutti, J. J.; Tomasko, D. L. “Bioactive polymer surfaces via supercritical fluids,” J. Supercrit. Fluids, 2004, 28, 241–248. 107. Zhang, J.; Matthews, M. A.; Dalal, N.; Fox, A.; Fox, K.; Hemmer, J.; Laberge, M.; Drews, M.; Stump, M. “Mechanisms of supercritical carbon dioxide sterilization of bacterial spores,” AlChE Symp. Ser., 2005, 9052. 108. Lee, J. Y.; Song, C. H.; Kim, J. I.; Kim, J. K. “Preparation of aromatic polycarbonate nanoparticles using supercritical carbon dioxide,” J. Nanopart. Res., 2002, 4, 53–59. 109. Lee, S.; Lanterman, H. B.; Pettit Jr., P.; Fullerton, K. “Polymerization, compatibilized blending, and particle size control of powder coatings in a supercritical fluid,” US PATENT 6,849,678; 2000. 110. Wu, H. T.; Lee, M. J.; Lin, H. M. “Supercritical fluid-assisted dispersion of ultra-fine pigment red 177 particles with blended dispersants,” J. Supercrit. Fluids, 2006, 39, 127–134. 111. Cooper, A. I. “Polymer synthesis and processing using supercritical carbon dioxide,” J. Mater. Chem. (UK), 2000, 10, 207–234. 112. Davies, O. R.; Lewis, A. L.; Whitaker, M. J.; Tai, H.; Shakesheff, K. M.; Howdle, S. M. “Applications of supercritical CO2 in the fabrication of polymer systems for drug delivery and tissue engineering,” Adv. Drug Delivery Rev., 2008, 60, 373–387. 113. Duarte, A. R. C.; Gordillo, M. D.; Cardoso, M. M.; Simplicio, A. L.; Duarte, C. M. M. “Preparation of ethyl cellulose/methyl cellulose blends by supercritical antisolvent precipitation,” Int. J. Pharm., 2006, 311, 50–54. 114. Foster, N.; Mammucari, R.; Dehghani, F.; Barrett, A.; Bezanehtak, K.; Coen, E.; Combes, G.; Meure, L.; Ng, A.; Regtop, H. L. “Processing pharmaceutical compounds using dense gas technology,” Ind. Eng. Chem. Res., 2003, 42, 6476–6493. 115. Smith, J. M.; Van Ness, H. C.; Abbott, M. M. Introduction to Chemical Engineering Thermodynamics 5th and international edition; McGraw-Hill: Singapore, 1996. 116. Pereda, S.; Bottini, S. B.; Brignole, E. A. Supercritical Fluid Extraction of Nutraceuticals and Bioactive Compounds: Fundamentals of supercritical fluid technology; CRC Press, 2008. 117. Hannay, J.B.; Hogarth, J. “On the solubility of gases,” R. Soc. London, 1879, 29, 324–326.

174

R. Yoganathan et al.

118. Hannay, J.B.; Hogarth, J. “On the solubility of gases II,” R. Soc. London, 1880, 30, 484. 119. Francis, A. W. “Ternary systems of liquid carbon dioxide,” J. Phys. Chem., 1954, 58(12), 1099–1114. 120. McHugh, M.A.; Krukonis, V. Supercritical fluid extraction: principles and practice second edition, Butterworth-heinemann series in chemical engineering; Butterworth-Heinemann, 1994. 121. Zosel, K. Process for the decaffeination of coffee. US PATENT 4,260,639; 1981. 122. Kiran E.; Sengers, J. Supercritical Fluids: Fundamentals for Application; Kluwer Academic Pub, 1994. 123. Sako, T.; Sugeta, T.; Sato, M. “Application of supercritical fluids to clean-up technology and environment-friendly reaction process,” Ryusan to Kogyo, 1998, 51, 27–33. 124. Hyatt, J. A. “Liquid and supercritical carbon dioxide as organic solvents,” J. Org. Chem., 1984, 49, 5097–5101. 125. Leitner, W. “Supercritical carbon dioxide as a green reaction medium for catalysis,” Acc. Chem. Res., 2002, 35, 746–756. 126. Nalawade, S. P.; Picchioni, F.; Janssen, L. P. B. M. “Supercritical carbon dioxide as a green solvent for processing polymer melts: Processing aspects and applications,” Prog. Polym. Sci., 2006, 31, 19–43. 127. Perrut, M. “Supercritical fluid applications: Industrial developments and economic issues,” Ind. Eng. Chem. Res., 2000, 39, 4531–4535. 128. Savage, P. E.; Gopalan, S.; Mizan, T. I.; Martino, C. J.; Brock, E. E. “Reactions at supercritical conditions: Applications and fundamentals,” AIChE Journal, 1995, 41, 1723–1778. 129. Wood, C. D.; Cooper, A. I.; DeSimone, J. M. “Green synthesis of polymers using supercritical carbon dioxide,” Curr. Opin. Solid State Mater. Sci., 2004, 8, 325–331. 130. Woods, H. M.; Silva, M. M. C. G.; Nouvel, C.; Shakesheff, K. M.; Howdle, S. M. “Materials processing in supercritical carbon dioxide: Surfactants, polymers and biomaterials,” J. Mater. Chem., 2004, 14, 1663–1678. 131. Young, J. L.; DeSimone, J. M. “Frontiers in green chemistry utilizing carbon dioxide for polymer synthesis and applications,” Pure Appl. Chem., 2000, 72, 1. 132. Kazarian, S. G. “Polymers and supercritical fluids: Opportunities for vibrational spectroscopy,” Macromol. Symp., 2002, 184, 215–228. 133. Kirby, C. F.; McHugh, M. A. “Phase behavior of polymers in supercritical fluid solvents,” Chem. Rev., 1999, 99(2), 565–602. 134. Kazarian, S. G.; Chan, K. L. A. “Applications of atr-ftir spectroscopic imaging to biomedical samples,” Biochim. Biophys. Acta, Biomembr., 2006, 1758, 858–867. 135. Stille, J. K. Introduction to Polymer Chemistry; Wiley 1962. 136. Gross, S. M.; Givens, R. D.; Jikei, M.; Royer, J. R.; Khan, S.; DeSimone, J. M.; Odell, P. G.; Hamer, G. K. “Synthesis and swelling of poly (bisphenol-A carbonate) using supercritical CO2 ,” Macromol., 1998, 31, 9090–9092. 137. Kim, W. B.; Lee, J. S. “A one-step synthesis of reactive polycarbonate precursors by the coupled oxidative carbonylation of bisphenol-A and phenol with carbon monoxide,” Chem. Lett., 2001, 1044–1045. 138. Wang, L. S.; Cheng, S. X.; Zhuo, R. X. “Novel biodegradable aliphatic polycarbonate based on ketal protected dihydroxyacetone,” Macromol. Rapid Commun., 2004, 25, 959–963. 139. Wang, W. C. V.; Kramer, E. J.; Sachse, W. H. “Effect of high-pressure carbon dioxide on the glass transition temperature and mechanical properties of polystyrene,” J. Polym. Sci., Polym. Phys. Ed., 1982, 20, 1371–84. 140. Agrawal, C. M.; Ray, R. B. “Biodegradable polymeric scaffolds for musculoskeletal tissue engineering,” J. Biomed. Mater. Res., 2001, 55, 141–150. 141. Greco, R. S. Implantation Biology: The Host Response and Biomedical Devices; CRC Press, 1994. 142. Du, L.; Kelly, J. Y.; Roberts, G.W.; DeSimone, J. M. “Fluoropolymer synthesis in supercritical carbon dioxide,” J. Supercrit. Fluids, 2009, 47, 447–457.

Dense Gas Processing of Polymers

175

143. Lee, K. N.; Lee, H. J.; Kim, J. H. “Synthesis of phenolic/furfural gel microspheres in supercritical CO2 ,” J. Supercrit. Fluids, 2000, 17, 73–80. 144. Matsuyama, K.; Mishima, K.; Umemoto, H.; Yamaguchi, S. “Environmentally benign formation of polymeric microspheres by rapid expansion of supercritical carbon dioxide solution with a nonsolvent,” Envir. Sci. Technol., 2001, 35, 4149–4155. 145. Knez, Z. “Enzymatic reactions in dense gases,” J. Supercrit. Fluids, 2009, 47, 357– 3729. 146. Hobbs, H. R.; Thomas, N. R. “Biocatalysis in supercritical fluids, in fluorous solvents, and under solvent-free conditions,” Chem. Rev., 2007, 107(6), 2786–2820. 147. Roby, M. S.; Totakura, N.; Christoforou, C. T. “Method of increasing the plasticity and/or elasticity of polymers via supercritical fluid extraction and medical devices fabricated therefrom,” US Patent 5,670,614; 1997. 148. Sawan, S. P.; Shieh, Y. T.; Su, J. H.; Manivannan, G.; Spall, W. D. “Evaluation of supercritical fluid interactions with polymeric materials,” Supercritical Fluid Cleaning; McHardy, J.; Sawan, S.P., Eds.; Noyes Publications 1998; pp 121–161. 149. Shieh, Y. T.; Su, J. H.; Manivannan, G.; Lee, P. H. C.; Sawan, S. P.; Spall, W. D. “Interaction of supercritical carbon dioxide with polymers. ii. amorphous polymers,” J. Appl. Polym. Sci., 1996, 59, 707–17. 150. Kazarian, S. G. “Applications of FTIR spectroscopy to supercritical fluid drying, extraction and impregnation,” Appl. Spectrosc. Rev., 1997, 32, 301–348. 151. Ginty, P. J.; Barry, J. J. A.; White, L. J.; Howdle, S. M.; Shakesheff, K. M. “Controlling protein release from scaffolds using polymer blends and composites,” Eur. J. Pharm. Biopharm., 2008, 68, 82–89. 152. Mohamed, A.; Gordon, S. H.; Biresaw, G. “Polycaprolactone/polystyrene bioblends characterized by thermogravimetry, modulated differential scanning calorimetry and infrared photoacoustic spectroscopy,” Polym. Degrad. Stab., 2007, 92, 1177–1185. 153. Mohamed, A.; Gordon, S. H.; Biresaw, G. “Poly(lactic acid)/polystyrene bioblends characterized by thermogravimetric analysis, differential scanning calorimetry, and photoacoustic infrared spectroscopy,” J. Appl. Polym. Sci., 2007, 106, 1689–1696. 154. Tsioptsias, C.; Stefopoulos, A.; Kokkinomalis, I.; Papadopoulou, L.; Panayiotou, C. “Development of micro- and nano-porous composite materials by processing cellulose with ionic liquids and supercritical CO2 ,” Green Chem., 2008, 10, 965–971. 155. Cai, Q.; Yang, J.; Bei, J.; Wang, S. “A novel porous cells scaffold made of polylactide-dextran blend by combining phase separation and particle-leaching techniques,” Biomater., 2002, 23, 4483–4492. 156. Jang, B. Z. Direct injection of nanofibers and nanofiber composites for biomedical applications, US Patent App. 10/442,561; 2004. 157. Yamazaki, M.; Teduka, M.; Ikeda, K.; Ichihara, S. “Preparation of carbon materials with corallike continuous pores using miscible polymer blends,” J. Mater. Chem., 2003, 13, 975–977. R olds: 158. Barry, J. J. A.; Gidda, H. S.; Scotchford, C. A.; Howdle, S. M. “Porous methacrylate sca Supercritical fluid fabrication and in vitro chondrocyte responses,” Biomater., 2004, 25(17), 3559–3568. 159. Barry, J. J. A.; Silva, M. M. C. G.; Popov, V. K.; Shakesheff, K. M.; Howdle, S. M. “Supercritical carbon dioxide: Putting the fizz into biomaterials,” Philos. Trans. R. Soc. London, Ser. A, 2006, 364(1838), 249–261. 160. Berens, A. R.; Huvard, G. S.; Korsmeyer, R. W.; Kunig, F. W. “Application of compressed carbon dioxide in the incorporation of additives into polymers,” J. Appl. Polym. Sci., 1992, 46, 231–242. 161. Lopez-Periago, A.; Argemi, A.; Andanson, J. M.; Fernndez, V.; Garcia-Gonzlez, C. A.; Kazarian, S. G.; Saurina, J.; Domingo, C. “Impregnation of a biocompatible polymer aided by supercritical CO2 : evaluation of drug stability and drug-matrix interactions,” J. Supercrit. Fluids, 2009, 48, 56–63.

176

R. Yoganathan et al.

162. Lopez-Periago, A. M.; Vega, A.; Subra, P.; Argemi, A.; Saurina, J.; Garcia-Gonzalez, C. A.; Domingo, C. “Supercritical CO2 processing of polymers for the production of materials with applications in tissue engineering and drug delivery,” J. Mater. Sci., 2008, 43, 1939–1947. 163. Natu, M. V.; Gil, M. H.; de Sousa, H. C. “Supercritical solvent impregnation of poly(caprolactone)/poly(oxyethylene-b-oxypropylene-b-oxyethylene) and poly(ecaprolactone)/poly(ethylene-vinyl acetate) blends for controlled release applications,” J. Supercrit. Fluids, 2008, 47, 93–102. 164. Kazarian, S. G.; Brantley, N. H.; Eckert, C. A. “Dyeing to be clean: Use supercritical carbon dioxide,” Chem. Tech., 1999, 29, 36–41. 165. Erkey, C. “Preparation of metallic supported nanoparticles and films using supercritical fluid deposition,” J. Supercrit. Fluids, 2009, 47, 517–522. 166. Zhang Y.; Erkey, C. “Preparation of supported metallic nanoparticles using supercritical fluids: A review,” J. Supercrit. Fluids, 2006, 38, 252–267. 167. Ellis, J. L.; Tomasko, D. L.; Dehghani, F. “Novel dense CO2 technique for beta-galactosidase immobilization in polystyrene microchannels,” Biomacromol., 2008, 9, 1027–1034. 168. Utracki, L. A. Commercial Polymer Blends; Chapman and Hall: London, 1998. 169. Work, W. J.; Horie, K.; Hess, M.; Stepto, R. F. T. “Definitions of terms related to polymer blends, composites, and multiphase polymeric materials (iupac recommendations 2004),” Pure Appl. Chem., 2004, 76, 1985–2007. 170. Paul, D. R.; Newman, S. Polymer Blends Vol. 2; Academic Press: New York, 1978. 171. Ramakrishna, S.; Mayer, J.; Wintermantel, E.; Leong, K. W. “Biomedical applications of polymer-composite materials: a review,” Compos. Sci. Technol., 2001, 61, 1189–1224. 172. Biresaw, G.; Carriere, C. J.; Willett, J. L. “Interfacial adhesion in model bioblends,” J. Appl. Polym. Sci., 2004, 94, 65–73. 173. Kung, E.; Lesser, A. J.; McCarthy, T. J. “Morphology and mechanical performance of polystyrene/polyethylene composites prepared in supercritical carbon dioxide,” Macromol., 1998, 31, 4160–4169. 174. Kung, E.; Lesser, A. J.; McCarthy, T. J. “Mechanical properties of polystyrene/polyethylene blends prepared using supercritical CO2 processing,” Polym. Prepr. (Am. Chem. Soc., Div. Polym. Chem.), 1997, 38, 462–463. 175. Watkins, J. J.; McCarthy, T. J. “Polymerizations in supercritical fluid-swollen polymers: A new route to polymer blends and composite foams,” Polym. Prepr. (Am. Chem. Soc., Div. Polym. Chem.), 1994, 35, 699–700. 176. Watkins, J. J.; Brown, G. D.; RamachandraRao, V. S.; Pollard, M. A.; Russell, T. P. “Phase separation in polymer blends and diblock copolymers induced by compressible solvents,” Macromol., 1999, 32, 7737–7740. 177. Hataya, T.; Sawaguchi, T.; Muroga, Y.; Ishikawa, H.; Hoshi, T.; Hagiwara, T.; Yano, S. “Nano structure and physical properties of sPS/PMMA blend prepared in supercritical carbon dioxide fluid,” In Polym. Prepr. (55th SPSJ Annual Meeting, Nagoya, Japan), 2006, 55, 1014. 178. Sawaguchi, T.; Muroga, Y.; Ishikawa, H.; Hoshi, T.; Hagiwara, T.; Yano, S. “Preparation of syndiotactic polystyrene/poly(methyl methacrylate) blend using supercritical carbon dioxide fluid,” Kobunshi Ronbunshu, 2005, 62, 251–260. 179. Dai, X.; Liu, Z.; Han, B.; Yang, G.; Zhang, X.; He, J.; Xu, J.; Yao, M. “Preparation of poly(vinyl chloride)/polystyrene miscible blends using supercritical CO2 as a swelling agent,” Macromol. Rapid Commun., 2002, 23, 626–629. 180. Zhang, J.; Busby, A. J.; Roberts, C. J.; Chen, X.; Davies, M. C.; Tendler, S. J. B.; Howdle, S. M. “Preparation of a poly(methyl methacrylate)/ultrahigh molecular weight polyethylene blend using supercritical carbon dioxide and the identification of a three-phase structure: An atomic force microscopy study,” Macromol., 2002, 35, 8869–8877. 181. Li, D.; Liu, Z.; Han, B.; Song, L.; Yang, G.; Jiang, T. “Preparation of nanometer dispersed polypropylene/polystyrene interpenetrating network using supercritical CO2 as a swelling agent,” Polym., 2002, 43, 5363–5367.

Dense Gas Processing of Polymers

177

182. Liu, Z.; Song, L.; Dai, X.; Yang, G.; Han, B.; Xu, J. “Grafting of methyl methylacrylate onto isotactic polypropylene film using supercritical CO2 as a swelling agent,” Polym., 2002, 43, 1183–1188. 183. Fleming, O. S.; Chan, K. L. A.; Kazarian, S. G. “High-pressure CO2 -enhanced polymer interdiffusion and dissolution studied with in situ ATR-FTIR spectroscopic imaging,” Polym., 2006, 47, 4649–4658. 184. Wang, X.; Sanchez, I. C. “Welding immiscible polymers with a supercritical fluid,” Langmuir, 2007, 23, 12192–12195. 185. Elkovitch, M. D.; Lee, L. J.; Tomasko, D. L. “Effect of supercritical carbon dioxide on morphology development during polymer blending,” Polym. Eng. Sci., 2000, 40:1850–1861. 186. Elkovitch, M. D.; Tomasko, D. L.; Lee, L. J. “Supercritical carbon dioxide assisted blending of polystyrene and poly(methyl methyacrylate),” Polym. Eng. Sci., 1999, 39, 2075–2084. 187. Elkovitch, M. D.; Lee, L. J.; Tomasko, D. L. “Effect of supercritical carbon dioxide on pmma/rubber and polystyrene/rubber blending: Viscosity ratio and phase inversion,” Polym. Eng. Sci., 2001, 41, 2108–2125. 188. Elkovitch, M. D.; Lee, L. J.; Tomasko, D. L. “Supercritical fluid assisted polymer blending,” Annu. Techn. Conf.—Soc. of Plastics Engineers, 1998, 56, 2538–2541. 189. Debenedetti, P. G.; Tom, J. W.; Yeo, S. “Supercritical fluids: A new medium for the formation of particles of biomedical interest,” Proc. Controlled Release Soc., 1993, pp 141–142. 190. Webb, P. B.; Marr, P. C.; Parsons, A. J.; Gidda, H. S.; Howdle, S. M. “Dissolving biomolecules and modifying biomedical implants with supercritical carbon dioxide,” Pure Appl. Chem., 2000, 72, 1347–1355. 191. Fraser, D. “Bursting bacteria by release of gas pressure,” Nature, 1951, 167, 33–34. 192. Burns, D. C.; Brown, R. J.; Eisenhut, A. R. “Sterilization of drugs using supercritical carbon dioxide sterilant,” US Patent App. 11/889,345; 2007. 193. Zhang, J.; Burrows, S.; Gleason, C.; Matthews, M. A.; Drews, M. J.; LaBerge, M.; An, Y. H. “Sterilizing bacillus pumilus spores using supercritical carbon dioxide,” J. Microbiol Methods, 2006, 66:479–485. 194. Ballestra, P.; Cuq, J. L. “Influence of pressurized carbon dioxide on the thermal inactivation of bacterial and fungal spores,” Lebensm. Wiss. Technol., 1998, 31, 84–88. 195. Hemmer, J. D.; Drews, M. J.; LaBerge, M.; Matthews, M. A. “Sterilization of bacterial spores by using supercritical carbon dioxide and hydrogen peroxide,” J. Biomed. Mater. Res. Part B, Appl. Biomater., 2007, 80:511–518. 196. Ishikawa, H.; Shimoda, M.; Tamaya, K.; Yonekura, A.; Kawano, T.; Osajima, Y. “Inactivation of bacillus spores by the supercritical carbon dioxide micro-bubble method” Biosci. Biotechnol. Biochem., 1997, 61, 1022–1023. 197. Shieh, E.; Paszczynski, A.; Wai, C. M.; Lang, Q.; Crawford, R. L. “Sterilization of bacillus pumilus spores using supercritical fluid carbon dioxide containing various modifier solutions,” J. Microbiol. Methods, 2009, 76, 247–252. 198. Spilimbergo, S.; Bertucco, A.; Lauro, F. M.; Bertoloni, G. “Inactivation of bacillus subtilis spores by supercritical CO2 treatment,” Innovative Food Sci. Emerging Technol., 2003, 4, 161–165. 199. Spilimbergo, S.; Dehghani, F.; Bertucco, A.; Foster, N. R. “Inactivation of bacteria and spores by pulse electric field and high pressure CO2 at low temperature,” Biotechnol. Bioeng., 2003, 82, 118–125. 200. Watanabe, T.; Furukawa, S.; Hirata, J.; Koyama, T.; Ogihara, H.; Yamasaki, M. “Inactivation of geobacillus stearothermophilus spores by high-pressure carbon dioxide treatment,” Appl. Environ. Microbiol., 2003, 69, 7124–7129. 201. Zhang, J.; Dalal, N.; Gleason, C.; Matthews, M. A.; Waller, L. N.; Fox, K. F.; Fox, A.; Drews, M. J.; LaBerge, M.; An, Y. H. “On the mechanisms of deactivation of bacillus atrophaeus spores using supercritical carbon dioxide,” J. Supercrit. Fluids, 2006, 38, 268–273. 202. Jimenez, A.; Zhang, J.; Matthews, M. A. “Evaluation of CO2 -based cold sterilization of a model hydrogel,” Biotechnol. Bioeng., 2008, 101, 1344–1352.

Polymer Reviews, 50:178–230, 2010 Copyright © 2010, King Fahd University of Petroleum & Minerals ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583721003704289

Synthesis of Functional Polyolefins using Metallocenes: A Comprehensive Review M. ATIQULLAH,1 M. TINKL,2 R. PFAENDNER,3 M. N. AKHTAR,1 AND I. HUSSAIN1 1

Center for Refining & Petrochemicals, Research Institute; and Center of Research Excellence in Petroleum Refining & Petrochemicals, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia 2 Ciba, Plastic Additives Segment, Schwarzwaldallee 215 R-1059.4.02, 4002 Basel, Switzerland 3 Ciba Lampertheim GmbH, Chemiestrasse, 68623 Lanpertheim, Germany The copolymerization mainly of ethylene and propylene with various polar functional comonomers, using various metallocenes and the methylaluminoxane (MAO) cocatalyst, has been primarily reviewed from the perspective of the following two subjects. One is the influence of the various functional groups on the copolymerization reactions, and the properties of the resulting products; the other is the areas of future research. The functional groups have been classified into oxygen-, nitrogen-, and halogen-containing moieties; borane-containing α-olefins; silanes; dienes; cyclic olefins; as well as styrene and its various derivatives. The following areas—synthesis of easily soluble functional cooligomer/copolymer; products with uniform distribution of the comonomer; establishment of relation among the catalyst structure and the various steps of copolymerization (initiation, propagation, and chain termination); the degradation and stabilization study of the functional copolymers and the correlation of the same with the catalyst structure; the application of supported metallocenes to synthesize the resulting polymer; the minimization of multi-step synthesis; and the development MAO cocatalyst formulation (to improve the activity and comonomer incorporation)—have been identified to be the subjects of future research, which will be of special importance to the researchers working in functional polyolefins. Keywords metallocenes, functional polyolefins, protection-deprotection technique, research challenges

1. Introduction The absence of a functional moiety either in a polyolefin (polyethylene and polypropylene) main backbone or at the chain end limits its applications that require adhesion, dyability, pronounced surface effects, grafting, and compatibility with other materials. Therefore, the synthesis of functional polyolefins to a large extent focuses on the enhancement of their overall performances to expand the application profile. For example, functional polyolefins hold the potential for supporting the catalyst, medication, photoelectron materials, Received July 12, 2009; accepted November 22, 2009. Address correspondence to Dr Muhammad Atiqullah, Research Inst, Center for Refining & Petrochemicals, Dhahran 31261, Saudi Arabia. E-mail: [email protected]

178

Synthesis of Functional Polyolefins using Metallocenes

179

biomaterials, photo-materials, additive development, and environmental protection, which the commodity polyolefins cannot access.1 The functional polyolefins with block and graft structures constitute the other examples of applications. They can make the ideal compatibilizers to increase the adhesion of polyolefin through interfacial interactions.2–4 However, the lack of chemical functionality and reactivity poses a major stumbling bottleneck. The selective and catalytic incorporation of polar groups into the chemically unreactive polyolefin backbone can overcome the above situation; but this remains a significant challenge. The superior properties of polyolefins must be retained during functionalization. This requires that the functionalization be conducted in a controlled manner that renders all the polyolefin structural and compositional parameters, including the concentration of functional groups and their distributions as well as polymer molecular weight, molecular weight distribution, and stereoregularity, highly controllable and retainable.5 Therefore, research continues to incorporate functionality using the following two major catalytic routes: i. Direct copolymerization of an olefin with a polar functional comonomer; and ii. Reactive polyolefin intermediate approach that synthesizes a precursor by incorporating a reactive functional comonomer which can be selectively and conveniently converted to the desired ultimate functional group(s). Another method for incorporating functional groups into polyolefins is reactive grafting, where functionality is introduced via radical abstraction/recombination steps, for example, the reaction of maleic anhydride with polypropylene or polyethylene in the presence of peroxides. However, this kind of functionalization is outside the scope of this review. Our literature search reveals that metallocenes, as well as post-metallocene late transition and rare earth metal complexes have been used for the above purpose. An excellent review on the application of late transition and rare earth metal complexes (that are less oxophilic and more stable to the heteroatoms) are already available in this area.6 The same observation applies to the metallocene-catalyzed commodity ethylenes and polyethylenes.7–15 On the other hand, so far as the use of metallocenes particularly to synthesizing functional polyolefins is concerned, the knowledge relating to the mechanism of olefin and comonomer insertion, propagation, and termination reactions, as well as the relationship between the catalyst symmetry and functional polyolefin microstructure remain scattered and wide-spread. Therefore, we design the current review that systematically consolidates the application of metallocenes cocatalyzed by methylaluminoxane (MAO) in this area based on the following major four classes of functional comonomers: i. Aromatic comonomers (styrene and its analogues); ii. Oxygen-containing functional groups (acrylates, alcohols, esters, anhydrides, ethers, oxazolines, etc.); iii. Nitrogen-containing functional groups (amines, acryl amides, imides, etc.); and iv. Halogen-containing functional groups (F, Cl, Br, and I). We hope that the present review will significantly contribute to using metallocenecatalyzed process to synthesize structurally well-defined functional polyolefins, including side group-functionalized polyolefins, chain end-functionalized polyolefins, and functional polyolefins with novel graft and block structures. Consequently, commodity polyolefins will greatly expand to value-added specialty products.

180

M. Atiqullah et al.

2. Co- and Terpolymerization of Olefins with Aromatic Comonomers An aromatic moiety can undergo nucleophilic substitutions. Consequently, it is amenable to subsequent functionalization such as chlorination, bromination, lithiation, carboxylation, silylation, hydroxylation, amination, etc. Therefore, the co- and terpolymerization of ethylene and propylene with an aromatic comonomer, such as styrene, p-methyl styrene, divinyl benzene, allylbenzene, and allylanisole, is of considerable research interest. What follows summarizes the pertinent work published in this area. 2.1 Co- and Terpolymerization of Olefins with Styrene and P-Methyl Styrene 2.1.1 Synthesis of Polyethylenes with Phenyl or Substituted Phenyl Pendant Groups. Pellechia et al.16,17 copolymerized ethylene with styrene using the Cp∗ TiMe3 -B(C6 F5 )3 catalyst system, which formed mixed products. A major component in this mixture consisted of a polyethylene backbone with 4-phenyl-1-butyl branches. However, unsaturated ethylenestyrene co-oligomers were also formed; styrene or phenyl end-capped the backbone. The cooligomerization reaction selectively formed various phenyl hexenes (linear and branched) and no phenylbutenes. Chung and co-workers18–23 copolymerized ethylene with p-methylstyrene (p-MeSt) using Cp2 ZrCl2 , Et(Ind)2 ZrCl2 , and Me2 Si(Cp)(NtBu)TiCl2 . The results can be summarized as follows: i. The copolymerization efficiency, expressed in terms of conversion of p-MeSt, decreased as Me2 Si(Cp)(NtBu)TiCl2 > Et(Ind)2 ZrCl2 > Cp2 ZrCl2 . ii. Me2 Si(Cp)(NtBu)TiCl2 showed the maximum activity and p-MeSt incorporation. The average p-MeSt content, as a function of the structural variation of the experimental metallocenes, ranged from 1.3 to 40 mol%. iii. In the copolymer backbone, p-MeSt, instead of end-capping the chain, was randomly distributed. This result particularly differs from what Pellechia et al.16,17 noted for styrene using Cp∗ TiMe3 -B(C6 F5 )3 . iv. Et(Ind)2 ZrCl2 , unlike Cp∗ TiMe3 -B(C6 F5 )3 , produced high molecular weight ethylene/styrene and ethylene/p-MeSt copolymers. No cooligomers were obtained. Me2 Si(Cp)(NtBu)TiCl2 synthesized ethylene/propylene)/p-MeSt and ethylene/1octene/p-MeSt terpolymers. The molecular weight distributions were narrow, PDI = 2–3. Chung and Dong,24 and Dong and Chung25 also copolymerized ethylene with pMeSt; however, in the presence of hydrogen—the conventional chain-transfer agent. The following metallocenes were used: Cp∗ 2 ZrCl2 , (nBuCp)2 ZrCl2 , Cp2 ZrCl2 , Et(Ind)2 ZrCl2 , and Me2 Si(Cp)(NtBu)TiCl2 . The objectives were to study the influence of (a) opening at the active sites on the catalyst performance and product properties; and (b) hydrogen on the resulting copolymer backbone. The openings at the active sites increase in the order the above metallocenes have been reported. Cp∗ 2 ZrCl2 with the most closed active site did not incorporate p-MeSt. On the other hand, the constrained geometry Me2 Si(Cp)(NtBu)TiCl2 with the most open active site exhibited a very high level of copolymerization activity. The effect of hydrogen was found to depend on the structure of the experimental metallocenes. For Me2 Si(Cp)(NtBu)TiCl2 , hydrogen did not affect the resulting copolymer backbone, which means that the insertion of p-MeSt as the pendant side groups was retained. However, for Cp2 ZrCl2 and (nBuCp)2 ZrCl2 , corresponding to a given p-MeSt

Synthesis of Functional Polyolefins using Metallocenes H2

CH2 CH2 CH M

181

CH2 CH3

p-MeSt CH2 CH2 CH2 CH Ph

H2 CH2 CH2 CH2 CH2

CH3

CH3

Scheme 2.1. The chain transfer mechanism of copolymerization of ethylene with p-MeSt in the presence of hydrogen.25

feed concentration, the increase in hydrogen concentration increased the catalyst activity, and decreased the molecular weights. The resulting copolymer chain showed Me or p-MeSt end groups. This is opposite to what happened with Me2 Si(Cp)(NtBu)TiCl2 . See Scheme 2.1. The p-MeSt mol% in the copolymer, and the PDI did not significantly vary.24,25 2.2.2 Synthesis of Polypropylenes with Phenyl or Substituted Phenyl End Groups. From the chemical and structural viewpoint, propylene differs from ethylene. Propylene, unlike ethylene, is prochiral. Therefore, it copolymerized with styrene and p-MeSt in a different fashion.26–29 Et(Ind)2 ZrCl2 synthesized isotactic polypropylene backbone containing a terminal phenyl (styrene) or p-MeSt group. This is a spectacular difference from what happens when ethylene is copolymerized with styrene or p-MeSt. See Section 2.1.1. Hydrogen served as the external chain transfer agent while styrene or p-MeSt acted as an in-situ one. With the increase in styrene feed concentration, the catalytic activity, styrene conversion, and the number average molecular weight Mn decreased. Mn ranged from 23,500 to 10,000 g/mol. However, the average styrene content in the copolymer showed the opposite trend. Also, an increased amount of hydrogen was needed to maintain high catalyst activity and styrene conversion. Me2 Si[2-Me-4-Ph(Ind)]2 ZrCl2 , like Et(Ind)2 ZrCl2 , evidenced similar results but Mn ranged from 4,600 to 1,800 g/mol.27,28 It further demonstrated the following:24,25 i. The Mw of the resulting polymers was proportional to [propylene]:[p-MeSt]. ii. Corresponding to a given p-MeSt feed concentration, the increasing concentration of hydrogen increased the catalyst activity and the conversion of p-MeSt. The p-MeSt mol% in the polymer backbone remained fairly unaffected. However, it increased with the increase in p-MeSt feed concentration. iii. Hydrogen was necessary to complete the chain transfer reaction to p-MeSt. Scheme 2.2 illustrates the mechanism of copolymerization of propylene with p-MeSt using Me2 Si[2-Me-4-Ph(Ind)2 ]ZrCl2 in the presence of H2 . During the polymerization of propylene (with 1,2 insertion), the propagation Zr–C site (II) can also react with p-MeSt (with 2,1 insertion) to form p-MeSt-terminated PP (III). The catalytic Zr–C site in compound (III) becomes inactive to propylene and p-MeSt due to (a) the steric hindrance between the active site (Zr–C) and the incoming monomer (propylene with 1,2 insertion), and (b) the formation of a complex with the adjacent phenyl group. However, in the presence of H2 the dormant Zr–C (III) species can react with H2 to form PP–t–p-MeSt (V) and regenerate a Zr–H species (I) that can reinitiate the polymerization

182

M. Atiqullah et al.

Scheme 2.2. The mechanism of copolymerization of propylene with p-MeSt using Me2 Si[2-Me-4Ph(Ind)2 ]ZrCl2 in the presence of H2 chain transfer agent.30

of propylene to continue the cycle. Overall, this whole process resembles sequential chain transfer reactions first with p-MeSt, next with H2 . 2.2 Co- and Terpolymerization of Olefins with 1,4-Divinylbenzene Chung and Dong30–32 copolymerized ethylene with 1,4-divinylbenzene (DVB) using Cp2 ZrCl2 , Ind2 ZrCl2 , Et(Ind)2 ZrCl2 , Me2 Si(Ind)2 ZrCl2 , and (C5 Me4 )Me2 Si(Cp) (NtBu)TiCl2 . Et(Ind)2 ZrCl2 turned out to incorporate maximum 1,4-DVB (7.2 mol%). This product showed Tm = 88.2◦ C and Mw = 34,000 g/(g mol). They and also Dong et al.33 conducted the following terpolymerization reactions: Polymerization systems i. (Ethylene + 1-octene)/divinyl benzene [EO-DVB] ii. (Ethylene + propylene)/divinyl benzene [EP-DVB]

Zirconocene catalyst type Et(Ind)2 ZrCl2 , Me2 Si(Cp)(NtBu)TiCl2 , and Cp2 ZrCl2 Et(Ind)2 ZrCl2 and Me2 Si(Cp)(NtBu)TiCl2

They noted that the catalyst type influenced the incorporation of divinyl benzene. For the (ethylene + 1-octene)/divinyl benzene system, the divinyl benzene incorporation capability varied as follows: Me2 Si(Cp)(NtBu)TiCl2 >Et(Ind)2 ZrCl2 >Cp2 ZrCl2 (poor incorporation). However, for the (ethylene + propylene)/divinyl benzene system, Et(Ind)2 ZrCl2

Synthesis of Functional Polyolefins using Metallocenes

183

and Me2 Si(Cp)(NtBu)TiCl2 showed comparable incorporation capability. Each catalyst produced a high molecular weight product for either of the terpolymerization systems. However, Et(Ind)2 ZrCl2 showed balanced properties with an effective incorporation of comonomers and monoenchainment of DVB. The resulting terpolymers were completely soluble in common organic solvents such as toluene or hexane. The following selected properties were reported:

Terpolymer type

DVB (mol%)

Properties Mw (g/mol)

Tg (◦ C)

EP-DVB EO-DVB

1.1−21.1 2.0−8.0

86,000−138,000 69,000−136,000

–22 to –51 –50 to –60

2.3 Copolymerization of Olefins with Allylbenzene Byun et al.34 first copolymerized ethylene with allylbenzene using Et(Ind)2 ZrCl2 . A random copolymer containing the reactive pendant phenyl group was obtained. This result can be compared with what was obtained during the copolymerization of ethylene with pMeSt using the same metallocene.18–23 In a subsequent study, they investigated the effect of metallocene structural variation on this copolymerization system by using Cp2 ZrCl2 , (nBuCp)2 ZrCl2 , (2-MeInd)2 ZrCl2 , and Cp∗ 2 ZrCl2 .35 The results can be summarized as follows: i. The catalytic activity, compared to homopolymerization of ethylene, increased with the increasing allylbenzene feed concentration. This shows a positive comonomer effect. ii. The incorporation of allylbenzene in the copolymer decreased in the following order: Cp2 ZrCl2 > (nBuCp)2 ZrCl2 > (2-MeInd)2 ZrCl2 > Cp∗ 2 ZrCl2 , and Mw as: Cp2 ZrCl2 > (2-MeInd)2 ZrCl2 > (nBuCp)2 ZrCl2 > Cp∗ 2 ZrCl2 ; 2.2 ≤ PDI ≤ 2.8. iii. The metallocene structure influenced the chain transfer mechanism. The following two mechanisms—β-H elimination and chain transfer to Al (due to the presence of trimethyl aluminum in MAO)—mostly prevail in metallocene-catalyzed olefin polymerization. For Cp2 ZrCl2 and (nBuCp)2 ZrCl2 , the former was predominant while for (2-MeInd)2 ZrCl2 and Cp∗ 2 ZrCl2 , the latter showed to be the major one, which was attributed to the incorporation of the allylbenzene in the propagating chain end. However, exceptions to what has been stated above have also been reported in the literature. For example, in propylene polymerization using metallocenes such as Cp∗ 2 ZrCl2 , chain transfer takes place predominantly via β-Me rather than β-H elimination.36,37 Zheng et al.38,39 used the work of Byun et al.,34 and Byun and Kim35 to incorporate various anhydrides and chlorosulfonic acid group at the para-postion of the pendant phenyl ring of the ethylene-allylbenzene copolymer (that served as a reactive intermediate precursor). See Section 3.3.2 for further discussion. 2.4 Copolymerizationof Olefins with Allylanisole Byun et al.40 copolymerized 4-allylanisole with ethylene using the following zirconocenes—Et(Ind)2 ZrCl2 , (nBuCp)2 ZrCl2 , Cp2 ZrCl2 , Cp∗ 2 ZrCl2 , and (2-MeInd)2 ZrCl2 . The comonomer 4-allylanisole was randomly incorporated as a pendant group in

184

M. Atiqullah et al.

the backbone; however, it dropped the catalytic activity. The subsequent transformation of the anisole methoxy (−OCH3 ) group using AlCl3 into a hydroxyl group produced phenolcontaining polyethylenes. This means that ethylene-co-4-allylanisole acted as a reactive intermediate precursor for this final product. The functional group transformation was confirmed using the 1H and 13C NMR assay of the copolymers. The following results were obtained: i. All the metallocenes, except Cp∗ 2 ZrCl2 , despite structural variation formed cooligomers without using any conventional chain-transfer agent (2,900 ≤ Mw ≤ 7,700). ii. The ligand structure of the zirconocene significantly affected the copolymerization of ethylene with 4-allylanisole. The catalyst having bridged or less substituted ligands favored the incorporation of the anisole than the unbridged analogues. Et(Ind)2 ZrCl2 incorporated more allylanisole (6.4 mol%) than the remaining experimental metallocenes. Probably, the accessibility of the comonomer to the catalyst active center affected its insertion. iii. Cp∗ 2 ZrCl2 showed the highest catalytic activity, but the lowest incorporation of the anisole. This indicates that the polymerization activity is affected by the accessibility of the polar functional group (Lewis base) to the metal center. iv. The catalytic activity increased with the increase of the MAO concentration. However, this markedly reduced the incorporation of the anisole and the molecular weight of the resulting copolymer. v. Pretreating the anisole with MAO did not significantly change the activity or the comonomer incorporation, regardless of the pretreatment time. On the other hand, the activity of the catalyst and comonomer incorporation depended on the MAO concentration. vi. The molecular weight of the copolymers decreased with the increase of the MAO concentration. The end group analysis of the products revealed that the chain transfer to aluminum was the dominant chain transfer reaction. vii. The final copolymer was characterized with pendant phenol groups. viii. The activity and Mw varied as a function of the metallocene structures as follows: Activity: Et(Ind)2 ZrCl2 > (nBuCp)2 ZrCl2 , Cp2 ZrCl2 > (2-MeInd)2 ZrCl. Molecular weight: (2-MeInd)2 ZrCl > Et(Ind)2 ZrCl2 > (nBuCp)2 ZrCl2 > Cp2 ZrCl2 . The above order fairly established an inverse relation between activity and molecular weight. For Et(Ind)2 ZrCl2 , the activity of the catalyst and molecular weight of the copolymer decreased as the concentration of the anisole in the feed increased. This deactivating effect was attributed to the complexation between the polar anisole and the catalytic active site. Atiqullah et al.41 copolymerized propylene with allylanisole using Me2 Si(Ind)2 ZrCl2 and Et(Ind)2 ZrCl2 . The weight-average molecular weight Mw decreased linearly as the concentration of allylanisole (AA) in the feed increased. This happened with both metallocenes. Therefore, allylanisole acted as an in-situ chain transfer agent. The chain transfer constant ktr /kp was determined through kinetic modeling. This turned out to be 0.33 and 0.40 for Et(Ind)2 ZrCl2 and Me2 Si(Ind)2 ZrCl2 , respectively. The characterization of the resulting products by 1H NMR demonstrated that allylanisole end-capped the isotactic polypropylene chains which showed to be low molecular weight oligomers; 4.96 × 103 ≤

Synthesis of Functional Polyolefins using Metallocenes

185

Scheme 2.3. Propagation and chain transfer (termination) mechanisms of cooligomerization of allylanisole with propylene.41

Mw ≤ 9.80 × 103. This finding matches what was observed with styrene and p-MeS.26–29 Scheme 2.3 shows the proposed chain propagation and termination mechanisms. During the chain propagation step, propylene gets incorporated into the isotactic backbone through 1,2 regioselective insertion.42 On the other hand the chain termination step involves protonolytic chain transfer by allylanisole (AA). The latter aspect may be attributed to the 2,1 insertion by AA into the propagating Zr+−C site followed by proton transfer, or a four-centered σ –bond metathesis transition state.43,44 The regenerated Zr+−H species is capable of (a) reinitiating the polymerization of propylene, and (b) continuing the polymerization cycle. The 1H NMR spectra of the products did not evidence vinylic chain end resonance (at δ (ppm) = 4.65−4.73),36 suggesting that chain termination (transfer) by β-H elimination (that is, formation of a –CH=CH2 ) did not occur. In a typical 1H NMR spectrum of a typical AA-PP cooligomer, the end-capping by an AA unit is manifest by the resonances at: δ (ppm) = 3.5786 → –OCH3 ; δ (ppm) = 5.8032 → –CH=CHCH2 −Ph (allylic structure). The suppression of β-H elimination may be attributed to the unfavorable β-agostic interaction at the propagating active site with an allylbenzene end unit.24 The representative peak due to the allylic unsaturation supports Scheme 2.3. An eventual consequence of the above terminal chain transfer mechanism is the following. The inserted bulky phenyl group of AA (with the π –electrons therein) is likely to exert steric hindrance and interact with Zr+. See Scheme 2.3. Consequently, the catalytic activity

186

M. Atiqullah et al.

of Et(Ind)2 ZrCl2 and Me2 Si(Ind)2 ZrCl2 will decrease with the increasing concentration of AA. This was supported by the cooligomerization trials.

3. Copolymerization of Olefins with Oxygen-Containing Functional Groups 3.1 Graft Copolymerization of Olefins with MMA Methylmethacrylate (MMA) is very important from the application viewpoint; therefore, in this section, it will be discussed first. The copolymerization of an olefin with MMA using transition metal complexes can proceed via 1,2 insertion or 2,1 insertion. During 1,2 insertion, the potential copolymerization could be inhibited by back chelation of the penultimate carbonyl group with the transition metal atom. Consequently, the olefin and the comonomer may be prevented from access to the vacant coordination sites of the transition metal. However, in case of 2,1 insertion, a metal-oxygen enolate bond is formed. Consequently, olefin will not insert due to endothermicity of the insertion step. The metal-oxygen bond (BDEM-O ) dissociation energy is much higher than that of the metal-carbon bond (BDEM-C ) dissociation energy. See Scheme 3.1 and Fig. 3.1.6

Scheme 3.1. Copolymerization of an olefin with MMA.6

An exception to the above occurs when the metal-oxygen enolate species can rearrange to form another carbon-bonded intermediate. Such a system has been discovered in late transition and rare earth metal complexes.6 The research group of Chung5,45–57 have extensively worked on the copolymerization of α-olefins with MMA using the reactive polyolefin intermediate (precursor) approach. The reactive polyolefin was prepared by copolymerizing an α-olefin with a selected “reactive”

Figure 3.1. Bond energy diagram.6

Synthesis of Functional Polyolefins using Metallocenes

187

comonomer such as 5-hexenyl-9-BBN and p-MeSt. Here, the key factor is to select a functional comonomer which should meet the following criteria: i. It should not complex with the metallocenes and be soluble in hydrocarbon polymerization media; ii. It should have good copolymerization reactivity with α-olefins; and iii. It should be facile and subsequently effective to form a stable in-situ intiator to polymerize MMA. By far the most effective reactive comonomers used for preparing PE-g-MMA are boranes5,45–57 and p-MeSt.18–23 The graft copolymerization of α-olefins with MMA proceeds stepwise. The first step synthesizes PE-containing borane and p-MeSt side groups whereas the next step converts the resulting product to PE-g-PMMA. 3.1.1 Synthesis of Polyolefins with Borane Side Groups. The copolymerization of α-olefins and 5-hexenyl-9-BBN (a higher α-olefin-containing borane moiety) is a convenient way to prepare borane-containing polyolefin.5,45–57 The borane moiety does not pre-complex with metallocene and dissolves well in organic solvents used in polymerization.58 The reactivity ratio of the two monomers governs the structure of the resulting copolymer. The bigger the size of the monomer, the lower is usually the activity. Ethylene is generally five times more reactive than propylene. Therefore, the copolymerization of the borane monomer with ethylene59 will be more difficult than with propylene,51–53 or with an α-olefin (1-butene or 1-octene).60 However, the metallocene catalyst offers better reactivity for copolymerization of borane monomer with ethylene. Chung et al.61 copolymerized ethylene with 5-hexenyl-9-BBN using Et(Ind)2 ZrCl2 and Cp2 ZrCl2 . The borane was incorporated as the pendant side group (Scheme 3.2). The resulting product served as a reactive polyolefin intermediate. See Section 3.1.2. Other findings can be summarized as follows: i. Et(Ind)2 ZrCl2 showed overall satisfactory copolymerization performance at the ambient temperature. ii. The concentration of the borane in the resulting copolymer turned out to be proportional to that in the feed. About 50 to 60 wt% of borane monomer was incorporated into the copolymer in about half an hour. iii. The catalytic activity increased with the increase of borane in the initial feed concentration. iv. Cp2 ZrCl2 incorporated less borane than Et(Ind)2 ZrCl2 . v. The presence of the borane moiety made the copolymer air-sensitive. vi. The molecular weight was high with PDI < 3.

Scheme 3.2. Reaction of ethylene with 5-hexen-9-BBN.20

188

M. Atiqullah et al.

Scheme 3.3. Synthesis of polyethylene-g-MMA copolymer from borane side groups.63,64

Chung and Xu62 copolymerized the olefin such as ethylene with a borane dimer (9-BBN) using Cp∗ 2 ZrMe2 in the presence of methyltri(pentafluorophenyl)borate [MeB(C6 F5 )3 ] cocatalyst. The borane dimer got incorporated into the growing polyethylene chain as a terminal reactive functional group. This finding differs from what was observed with Et(Ind)2 ZrCl2 and Cp2 ZrCl2 . However, the product backbone is comparable with what Pellechia et al.16,17 noted in the copolymerization of ethylene with styrene using Cp∗ TiMe3 -B(C6 F5 )3 . The above was the first step. In the next step, the resulting borane-terminated copolymer PE-t-B was oxidized by NaOH/H2 O2 to the corresponding hydroxyl-terminated PE-t-OH copolymer. Increasing the concentration of the dimer decreased the catalytic activity and the molecular weight Mw of the PE-t-OH copolymer. Therefore, the dimer acted as an in-situ chain-transfer agent. 3.1.2 Synthesis of Polyolefin-g-MMA Copolymer from Borane Side Groups.. The alkyl-9BBN side groups in polyethylene can be spontaneously oxidized even at very low temperature (–65◦ C) to a peroxide species as shown in Scheme 3.3. The insertion of oxygen increases the unfavorable ring strain into the C–B bonds of the bicyclic ring of 9-BBN. This destroys the stable double chair-form structure. The oxidation reaction selectively takes place at the C–B bond in the linear alkyl group.63,64 This produces the peroxyborane (C–O–O–B) (I) species. The peroxy borane (I) behaves differently from regular benzoyl peroxides. It decomposes by itself even at ambient temperature. The decomposition reaction follows the homolytical cleavage of peroxide to generate an alkoxy radical (C–O∗ ) and a borinate radical (B–O∗ ). The alkoxy radical (C–O∗ ) generated in this way is highly reactive; it initiates radical polymerization with methylmethacrylate. The borinate radical (B–O∗ ) is stabilized by the empty p-orbital of boron by back donating electron density. The growing chain (II) then reacts with MMA to extend the polymer chain to form a graft copolymer. The graft length of (PMMA side chain) is basically controlled by the MMA concentration and reaction time. The PE-g-MMA copolymerization results are listed below:57 i. Oxygen affected the graft efficency. Even though the final stoicheometry of oxygen to boron should be 1:1, the best result was obtained when oxygen was introduced slowly so that O B at any time. An excess of oxygen not only poisoned the free radical polymerization but also overoxidized to boronates. Borates are poor free-radical initiators at room temperature.

Synthesis of Functional Polyolefins using Metallocenes

189

Scheme 3.4. Synthesis of PE-g-MMA graft copolymer from p-MeSt side groups.61

ii. The polarity of the solvent played an important role in the graft reaction. THF was found to be a better solvent than the nonpolar benzene which slowed the reaction probably due to low solubility of oxygen. iii. The FTIR spectrum of PE-g-PMMA showed a strong adsorption band at 1,730 cm−1 corresponding to the ester groups. A high concentration (>65 mol%) of PMMA can be incorporated into PE by a small quantity (0.5 mol%) of borane groups. 3.1.3 Synthesis of Polyolefin-Graft-MMA Copolymer from p-MeSt Side Groups. First we shall review the synthesis of PE-g-MMA, then that of PP-g-MMA. The p-MeSt side groups in polyolefins [Section 2.1.1] can be converted into a polyolefin graft copolymer via anionic polymerization. See Scheme 3.4. Generally, a low concentration (< 1 mol%) of p-MeSt in the copolymer is preferred to prepare the graft copolymer because the resulting graft copolymer will have low graft density and long graft length. The p-MeSt side groups in PE can be effectively metallated at ambient temperature.65 The lithiated PE-p-MeSt copolymer contains several polymeric anions that are homogenously distributed in the polymer chain. The lithiated PE-p-MeSt initiates graft copolymerization of methylmethacrylate which is separated from the homopolymer by solvent fractionation. In most cases, less than 10 wt% homopolymer was obtained. A polar solvent like THF gave poor yield at 0◦ C and 25◦ C, while a nonpolar solvent such as cyclohexane gave better yield. The polar solvent may increase the nucleophilicity of the carbanion, which results in more side reactions. It is interesting to note that the anionic polymerization of polar monomers using butyl lithium as an initiator cannot achieve high yields at ambient temperature. Usually very low reaction temperature (< –20◦ C) is required. In lithiated PE-p-MeSt case, the formed polymeric benzylic lithium is much more stable; therefore, it minimized the side reactions. The graft polymerization of MMA in nonpolar cyclohexane is fairly effective and a sufficiently long graft length can be achieved even at ambient temperature. Lu and Chung66 carried out polypropylene-g-MMA copolymerization using poly(propylene-co-p-MeSt) as starting material. The polypropylene copolymer containing p-MeSt was synthesized using a supported Ziegler-Natta catalyst system: (MgCl2 /TiCl4 /ED [external donor]/AlEt3 . The PP-co-p-MeSt was lithiated similar to PE-co-p-MeSt. However, the reaction was conducted at 70◦ C. The anionic graft polymerization of MMA took place at room temperature without any side reaction. The MMA incorporation was obtained in the range of 23 to 49 mol% having Tg 71.8 to 96.2◦ C and Tm 155.8 to 156.8◦ C.

190

M. Atiqullah et al.

3.2 Block Copolymerization of Olefins with MMA The block copolymerization of ethylene with MMA is an important subject. These copolymers have excellent properties as diverse as adhesion, dyeability, printability, printability, gas permeability, and compatibility with other functional polymers. An established technique for improving the interfacial interaction between polymers is to use block and graft copolymer as compatibilizer. The diblock copolymer structure is known to be the most effective of compatibilizers. Usually the compatibility of polymer blends can be improved by adding a small quantity (<1%) of a suitable diblock copolymer, which alters the phase morphology and interfacial adhesion of the blends. It is especially desirable to prepare polyolefin diblock copolymers containing functional polymers such as PMMA, PVA, etc., which can dramatically increase the interaction of polyolefin with a broad range of polymers containing functional groups and other substrates. In general, most block copolymers have been produced by sequential living polymerization processes namely anionic,67–70 group transfer, metathesis,71,72 or controlled radical polymerizations such as RAFT, ATRP, and nitroxyl-mediated.73–80 However, Chung and his group have successfully used boranes,5,45–50,52–57 and p-MeSt18–23 to synthesize graft and block copolymers of α-olefins and polar groups. Recently, Yasuda and his group81,82 have successfully used rare earth metal compounds to produce block copolymers of olefin and polar groups. These two approaches of Chung and his group are discussed below.5,50,59,61,62,83–85 In the first approach, polyolefins having a reactive end group were synthesized, which were further converted to block copolymers. Preparing a polymer having a terminal functional group, which serves as a building block for constructing multi segmented polymers, is a scientific challenge. 3.2.1 Synthesis of Polyolefins with Borane End Groups. The synthesis of polyolefin with borane side groups has already been discussed in Section 3.1.1. What follows discusses how to introduce the borane end groups. The reactive borane groups during polymerization of α-olefins encounter the following:83,84 i. B–H chain transfer reactions, namely hydroboration reaction of the B–H group to α-olefin monomers; and ii. Ligand exchange reaction between borane and aluminum alkyl coinitiator. A borane compound containing B–H groups usually forms a stable dimer in hexane and toluene that are used as polymerization solvents.83,84 No reaction between borane and α-olefin is expected during polymerization. To avoid exchange reactions of boranes with aluminum alkyls, perfluoroborate co-initiatoirs, instead of aluminum alkyl compounds, can be used. Therefore, the B–H moiety dialkyl borane can engage a chain transfer reaction by ligand exchange between B–H and M–C (M: transition metal) as illustrated in Scheme 3.5. If the chain transfer reaction is the dominant termination reaction, each PE chain will have a terminal borane group (PE–t-B), and the molecular weight of the borane-terminated polymer will be inversely proportional to the concentration of the borane chain transfer agent. Xu and Chung83,84 copolymerized ethylene with 9-borabicyclononane (9BBN) (a borane chain transfer agent) using [(Me5 C5 )2 ZrMe]+[MeB(C6 F5 )3 ]− and

Synthesis of Functional Polyolefins using Metallocenes

191

Scheme 3.5. Formation of PE with borane end groups.83,84

[(Me5 C5 )2 ZrMe]+[B(C6 F5 )4 ]− catalyst systems. The polymerization was done at room temperature for 3–5 minutes in toluene at 1.0 atm pressure. The resulting borane-terminated polyethylene was further copolymerized with a dimer of 9-BBN which produced polyethylene with terminal borane groups. This product was oxidized using: i. NaOH and H2 O2 to produce polyethylene with –OH terminal groups; and ii. O2 to produce free radicals as an initiator source for copolymerization with MMA. To maintain a constant [borane]:[ethylene] feed ratio, the reactions were carried out using rapid mixing and a short reaction time (about three to five minutes). The findings are summarized below:83,84 i. The higher concentration of a 9-BBN chain transfer agent lowered the molecular weight of the resulting PE. ii. The polymer weight distribution was narrow which is consistent with a single-site polymerization process. iii. The catalyst activity decreased in the presence of 9-BBN, reflecting the competitive coordination at metallocene active sites between the monomer and the chain transfer agent. iv. The Mn of the resulting polymer ranged from 3,700 to 85,000 with PDI varying from 2.0 to 2.7. Chung et al.85 evaluated the performance of the following three dialkyl borane chain transfer agents: 9-borabicyclononane (9-BBN), dimesitylborane (HB(Mes)2 ), and bis(2,4,6triisopropylphenyl)borane (HB(Trip)2 ), during ethylene polymerization. See Fig. 3.2. The above borane compounds were characterized by 11B NMR in d-toluene. It was observed that 9-BBN and HB(Mes)2 exist in a dimeric form whereas HB(Trip)2 in a monomeric form. The copolymerization of ethylene using 9-BBN, HB(Mes)2 , and HB(Trip)2 in presence of [Cp2 ∗ ZrMe]+[MeB(C6 F5 )3 ]− can be compared as follows. For BH(Mes)2 , the catalyst productivity was comparable with that of the control whereas it dropped for 9-BBN.

192

M. Atiqullah et al.

Figure 3.2. Selected dialkyl borane chain transfer agents.85

BH(Trip)2 formed no polymer. The overall catalyst activity decreased with the increase in borane concentration, which has already been explained earlier.85 The performance of [Cp2 ∗ ZrMe]+[MeB(C6 F5 )3 ]−, [(Ind)2 ZrMe]+[MeB(C6 F5 )3 ]−, and [C5 Me4 SiMe2 (Cp)(NtBu)TiMe]+[MeB(C6 F5 )3 ]− having different spatial openings was compared for ethylene polymerization using 9-BBN and BH(Mes)2 .85 The first two metallocenes with a small spatial opening showed relatively low chain transfer activities, while the third one with a wider opening showed significantly higher chain transfer reactions. 3.2.2 Synthesis of PE-b-MMA Block Copolymer from Borane End Groups.. The boraneterminated polyolefin can be selectively transformed into a polymeric peroxide-containing peroxylborane (C–O–O–BR2 ) moiety. This has been already described in Section 3.2.1. The peroxylborane is a reactive radical initiator even at ambient temperature.85 It undergoes O2 cleavage and generates alkoxy radical (C–O∗ ) and stable borinate radical (∗ O–BR2 ). See Scheme 3.6. The alkoxy radical then initiates MMA polymerization. The performance of PE-t-9BBN and PE-t-B(Mes)2 was compared. For PE-t-B(Mes)2 , almost no PMMA homopolymer was formed. However, for PE-t-9-BBN, a small amount (∼10 wt%) of PMMA homopolymer was formed. The results of PE-b-PMMA copolymer synthesized using PE-t-B(Mes)2 are reviewed as follows. The PE-t-B(Mes)2 intermediate precursor continuously increased its molecular weight during post-modification. However, its PDI remained constant (2.0 to 2.4). PE or PMMA homopolymers did not form. These findings established (a) the existence of a

Scheme 3.6. Formation of PE-b-PMMA copolymer.85

Synthesis of Functional Polyolefins using Metallocenes

193

borane group at each PE chain end, and (b) a living radical polymerization of MMA in the chain propagation process. PP showed similar results.65 The molecular weight increased over two times (Mn of PP-t-B = 13,000, Mn of PP-b-PMMA = 29,000). However, the PDI remained unchanged (∼1.6). Shiono et al.86 converted vinylidene-terminated isotactic PP, to an intermediate MgBrterminated PP. This MgBr-terminated PP was then used to prepare PP-b-MMA copolymer. The vinylidene-terminated isotactic PP was synthesized using Et(H4 Ind)2 ZrCl2 at 30◦ C.87 The vinylidene terminal C=C bond was hydroborated by a borane-dimethylsulfoxide complex. The vinylidene-terminated isotactic PP (1 g) was placed in a Schlenk tube with 10 ml toluene. Then borane-dimethylsulfoxide complex (0.086 mmol) was added to the tube under inert atmosphere. The mixture was stirred continuously at 70◦ C for 2 h. After this, pentane-1,5-diyldi(magnesium bromide) in THF solution (0.173 mmol) was added and it was further stirred continuously for 2 h. The reaction is as follows. PP-BR2 + 2BrMg(CH2 )5 MgBr → PP-MgBr + 2RMgBr + [B{(CH2 )5 }2 ]+MgBr− Borane was quantitatively converted to the Grignard reagent by reaction with pentane1,5-diyldi-(magnesium bromide) in toluene whereas PP was converted to MgBr-terminated PP. This MgBr-terminated PP was copolymerized with MMA as an ionic initiator. 3.2.3 Polyolefin-b-MMA Block Copolymer from p-methylstyrene End Groups. The terminal p-MeSt group is a source of living anionic polymerization. The first step in the synthesis of PP-b-PMMA is lithiation of PP-t-p-MeSt. In this reaction, powder PP-t-p-MeSt was mixed with sBuLi and N,N,N’,N’-tetramethylethylenediamine (TMEDA) in cyclohexane, and was stirred at 60◦ C for 4 h. Then it was filtered and washed using cyclohexane to remove the unreactive sBuLi or TMEDA. This lithiated species (II) initiated anionic polymerization of MMA.26–29 3.2.4 Synthesis of Block Copolymers of Olefin with Alkyl Methacrylates Using Rare-Earth Metal Complexes. Transition metal complexes are apt to poisoning during direct block copolymerization of α-olefin with a polar comonomer. However, organo-rare earth metal complexes can initiate the block copolymerization of α-olefin with polar comonomer like MMA, alkyl-acrylates, etc.88,89 The interesting feature of these complexes is that they act as a catalyst as well as an initiator/activator. Therefore, they do not need any cocatalyst. The organo lanthanide complexes LnR(C5 Me5 )2 [where Ln = Sm, Yb, Lu; R = H, Me] have a unique dual catalytic activity toward polar and nonpolar groups. Alkyl- or hydrido-lanthanide complexes exhibit a versatile catalytic activity for initializing the living polymerization of MMA.88,89 See Scheme 3.7. Very high molecular weight polymer (Mn > 1×106) with narrow polydispersity (PDI < 1.05) was obtained. (C5 Me5 )2 LnR showed excellent ethylene polymerization activity.90–92 Ethylene was copolymerized with MMA in two steps.88 It was first homopolymerized using SmMe(C5 Me5 )2 -THF or [SmH(C5 Me5 )2 ]2 at 20◦ C in toluene under atmospheric pressure. Next, MMA was added. The reaction is shown in Scheme 3.8. Ethylene initially polymerized very rapidly and the reaction completed in 2 min. The polymer properties were as follows: Mn = 10,000, and PDI ∼ = 1.43. However, the subsequent copolymerization with MMA proceeded rather slowly. The reaction was carried out for 2 h at 20◦ C. The resulting polymer was soluble in 1,2,4-trichlorobenzene at 100◦ C but was insoluble in THF or CHCl3 , indicating conversion to the desired block copolymer.

194

M. Atiqullah et al.

Scheme 3.7. Reaction showing the formation of PE-b-PMMA.82,89

Scheme 3.8. Copolymerization of ethylene with MMA using SmMe(C5 Me5 )2 catalyst.82

Repeated fractionations of the block copolymer in hot THF did not change the molar ratio of polyethylene to PMMA though PMMA-PE blend is easily extracted with THF. The molar ratio of PE to PMMA blocks could be controlled by feeding MMA in the range of 100:1 to 100:103, when Mn of the initial PE was ∼10,000. However, this ratio of PE:PMMA decreased with an increase of Mn of the initial PE, especially when Mn exceeded 12,000. These findings can be attributed to the encapsulation of the active sites by polyethylene, thus inhibiting the diffusion of MMA to the active sites. As a result further copolymerization was suppressed. Desurmont et al.93–95 synthesized a hydrogenated complex of samarocene [Me2 Si(C5 H3– 3-Me3 Si)2 SmH(THF)2 ] having a binuclear µ-H structure (See Fig. 3.3). The above complex showed high ethylene polymerization activity (2.7 × 104 g PE/mol Sm h). Polyethylenes with Mn = 3 × 104 to 5 × 104 and PDI ∼ = 1.65 were obtained. This complex also exhibited block copolymerization of ethylene with MMA [PMMA:PE 19:81 mol:mol, Mn = 6 × 104 to 7 × 104, PDI ∼ = 1.68] when ethylene was first polymerized and next MMA was added. However, the reverse order of addition induced no block copolymerization. It produced only PMMA homopolymer. Desurmont et al.93–95 also carried out block copolymerization of 1-pentene or 1-hexene with MMA using the following bridged complexes of yttrium and samarium (See Fig. 3.4).

Figure 3.3. Structure of [Me2 Si(C5 H3– 3-Me3 Si)2 SmH(THF)2 ] catalyst.95

Synthesis of Functional Polyolefins using Metallocenes

195

Figure 3.4. Structure of yttrium and samarium complexes.95

The block copolymerization was carried out in two steps. The first step homopolymerized 1-pentene or 1-hexene at 20◦ C in toluene under atmospheric pressure. In the second step, MMA was added to accomplish block copolymerization. Unlike ethylene, 1-pentene or 1-hexene polymerized rather slowly and the polymerization reaction was completed in 24 to 32 h to give poly(1-pentene) (Mn = 2.8 × 104 and PDI = 1.51) and poly(1-hexene) (Mn = 5.3 × 104 and PDI = 2.46, 90% yield) using the yttrium complex. The activity of the samarium complex was much lower than that of the yttrium one. The as-synthesized polymer showed bimodal MWD. However, when this was washed with hot hexane, a unimodal diblock copolymer was obtained. Desurmont et al.93–95 also prepared tri-block copolymers (ABA type) using ethylene and MMA. Here, two different samarium complexes were used. See Fig. 3.5. Compared to Catalyst 4, Catalyst 3 showed both lower catalytic activity and MMA incorporation, and produced polyethylene of higher molecular weight. However, Catalyst 4 synthesized MMA-co-Ethylene-co-MMA triblock polymer with molecular weight much greater than that obtained by using Catalyst 3.

3.3 Copolymerization of Olefins with Alcohols, Acids, and Ethers Olefins have been copolymerized with alcohols, acids, ethers, and oxazolines using metallocenes, according to the following three routes:

Figure 3.5. Structure of two different samarium complexes.95

196

M. Atiqullah et al. i. Directly copolymerize olefins with alcohols; ii. Use the protection/deprotection approach; and iii. Copolymerize olefins with a reactive group such as borane; then convert it to alcohols.

The polar groups can be protected by either silane or alkylaluminum compounds before polymerization and later on can be converted back to alcohols by washing with a suitable acid. 3.3.1 Direct Copolymerization of Olefins with Alcohols. Aaltonen and Loefgren96 copolymerized ethylene with the methylene spacer-containing 10-undecen-1-ol using (nBuCp)2 ZrCl2 . They found the following: i. Spacer effect: The longer methylene spacer group in the comonomer protected the catalyst from poisoning. ii. Temperature effect: At 60◦ C, an alcohol concentration of 9.9 wt% was incorporated in the product, whereas at 80◦ C, this was 1.2 wt%. Either temperature showed bimodal molar MWD as the alcohol feed concentration was increased. This indicates the formation of two or more active catalyst species. The polymer yield at both temperatures decreased with the increase in the alcohol feed concentration. iii. Reaction time effect: The consumption of ethylene increased with time because of its continued copolymerization with 10-undecen-1-ol. iv. 10-undecen-1-ol effect: The catalyst is deactivated; the Mw decreased; and the MWD broadened with the increase in concentration of 10-undecen-1-ol. When the concentration of 10-undecen-1-ol was increased, its –OH group reacted more with the Me of MAO and its associated/free AlMe3 . As a result, the alkylation of (nBuCp)2 ZrCl2 suffered which affected the generation of the active metallocenium cation; hence, the catalyst deactivated. The latter two phenomena may be attributed to the chain transfer effect of the alcohol as its concentration increased. The bimodal MWD indicates that at least two active species were present. The bimodality was also observed during DSC analysis. Aalotonen et al.97 copolymerized ethylene and propylene with 10-undecen-1-ol using Et(Ind)2 ZrCl2 , Et(Ind)2 ZrCl2 , Me2 Si(Ind)2 ZrCl2 , Me2 Si(2-MeInd)2 ZrCl2 and Me2 Si(2Me-4,5-BenzoInd)2 ZrCl2 . They studied the influence of (a) pretreating the metallocenes with MAO and (b) their structural variation on the above copolymerization reaction. The copolymerization of ethylene with 10-undecen-1-ol showed the following: i. The pretreatment of 10-undecen-1-ol with MAO before initiating polymerization increased the ethylene polymerization rate. ii. The copolymerization activity decreased as Me2 Si[Ind]2 > Me2 Si[2-MeInd]2 > Et[Ind]2 > Me2 Si[2-Me-4,5-BenzoInd]2 ∼ = Ind2 . Me2 Si(2-MeInd)2 ZrCl2 showed almost half the activity of the non-substituted catalyst probably due to sterically more crowded coordination sphere. However, the bulky Me2 Si(2-Me-4,5BenzoInd)2 ZrCl2 showed much higher ethylene homopolymerization activity. This may be attributed to the presence of the electron-donating substituent (Me and Benzo groups). iii. The alcohol incorporation varied as Me2 Si[2-Me-4,5-BenzoInd]2 > Me2 Si[2MeInd]2 > Me2 Si[Ind]2 > Ind2 . The overall incorporation of the alcohol ranged from 0.8 to 13.4 wt%.

Synthesis of Functional Polyolefins using Metallocenes

197

iv. In terms of activity, conversion, and alcohol incorporation, the performance of Me2 Si[2-MeInd]2 and Me2 Si[Ind]2 was better than the remaining ones. v. The Mw can be rated as Ind2 > Me2 Si[2-MeInd]2 > Me2 Si[2-Me-4,5-BenzoInd]2 > Et[Ind]2 . Each metallocene produced high molecular weight polyethylene homopolymer (Mw > 100,000) which significantly decreased upon addition of 10undecen-1-ol. The non-bridged metallocene offered a little broader molecular weight distribution (PDI = 5.0) than the bridged analogues (PDI = 2.0–2.5). vi. The PDI for all the zirconocenes except Ind2 ZrCl2 showed to be comparable and characteristic of single-site catalysts. On the other hand, for copolymerization of propylene with 10-undecen-1-ol, the following findings can be reported: i. The copolymerization activity decreased as Me2 Si[2-MeInd]2 > Me2 Si[2-Me-4,5BenzoInd]2 > Me2 Si[Ind]2 > Et[Ind]2 . ii. The alcohol incorporation varied as Me2 Si[Ind]2 ∼ = Et[Ind]2 > Me2 Si[2-MeInd]2 ∼ = Me2 Si[2-Me-4,5-BenzoInd]2 . iii. In terms of activity, conversion, and alcohol incorporation, the performance of Me2 Si[2-MeInd]2 and Me2 Si[2-Me-4,5-BenzoInd]2 was better than the remaining ones. The overall incorporation of the alcohol ranged from 2.7 to 3.8 wt%. iv. The Mw can be rated as Me2 Si[2-Me-4,5-BenzoInd]2 > Me2 Si[2-MeInd]2 > Me2 Si[Ind]2 > Et[Ind]2 . The addition of a Me substituent in the α–position of each C5 ring dramatically increased Mw . This was attributed to the steric influence of the substituents making chain termination more difficult. v. The PDI for all the zirconocenes showed to be comparable and characteristic of single-site catalysts. Hakala et al.98 and Sepp¨al¨a et al.99 copolymerized several oxygen-containing olefins—alcohols, ketones, ester, and carboxylic functionalities—with propylene using Et(Ind)2 ZrCl2 . See Fig. 3.6. They noted the following: i. All the comonomers significantly decreased the propylene polymerization activity. ii. The structure of the comonomer markedly affected its polymerizability and the catalyst deactivation. The steric protection of the functional group diminished the deactivating influence of the comonomer. The comonomers containing keto or weakly shielded ester groups poisoned the catalyst to the largest extent. The tertiary alcohols were found less detrimental to the catalyst than the primary and secondary ones. Also, the tert-butyl derivative of 10-undecenoic acid was tolerated better than the methyl ester due to the better protective effect of the bulkier tert-butyl group. iii. The structure of the comonomer chain (straight or branched) affected the activity of the metallocene. The deactivation of the catalyst is due to the interaction between the Lewis acid catalyst components and the free electron pairs of oxygen atoms in the comonomer structures. The interaction leading to catalyst deactivation will weaken if the oxygen atom is shielded by substituents added to the adjacent carbon atoms. Therefore, the activity of the metallocene, demonstrated in Fig. 3.7, was found to be more detrimentally affected by straight-chain alcohols than by the branched ones, reflecting the effect of steric hindrance of oxygen-containing group. Also, of the two esters, the tert-butyl derivative was tolerated better than the methyl ester, evidently again due to the better protective effect of the bulkier tert-butyl group (Fig. 3.6).

198

M. Atiqullah et al.

Figure 3.6. Structures of oxygen-containing functional comonomers copolymerized with propylene.98, 99

iv. The longer was the spacer group the better was the incorporation of the comonomer (Fig. 3.7). The highest comonomer content (2.7 wt%) was achieved with 10undecen-1-ol and the highest conversion (12.1 wt%) with 10-undecenoic acid. v. The incorporation of the functional comonomers decreased the melting temperature of the polymer. vi. The molecular weight of the copolymers was slightly lower than that of the propylene homopolymer and the polydispersity index of all the copolymers was typically narrow, as usually evidenced by metallocene catalysts. The aforesaid findings are explained as follows. The interaction between the Lewis acid catalyst components and the free electron pairs of oxygen atoms in the comonomer structures is responsible for deactivating the catalyst. Thus, shielding the oxygen atom by bulky substituents weakens this interaction, which reduces the deactivation. In addition, a longer spacer between the polymerizable double bond and the oxygen atom favored the copolymerizability of the functional comonomer.

Figure 3.7. The effect of shielding and spacer in alcohols on (a) catalyst deactivation and (b) comonomer incorporation.99

Synthesis of Functional Polyolefins using Metallocenes

199

Scheme 3.9. Copolymerization of propylene with 6-t-butyl-2-(1,1-dimethylhept-6-enyl)-4methylphenol using Me2 Si(H4 Ind)ZrCl2 .101

The catalyst deactivation may not be totally avoided when oxygen-containing polar comonomers are added. However, functional groups (such as alcohols and to some extent carboxylic acids) capable of forming stable, protected aluminates are less deactivating than the less acidic and/or less polar groups such as esters.98,99 Also, a selected study on ester groups revealed that the steric protection is important as well. For example, tert-butyl ester deactivated the catalyst to a lower extent than the methyl ester. Kaya et al.100 copolymerized propylene with 10-undecen-1-ol using Et[Ind]2 ZrCl2 . At 30◦ C, they obtained a copolymer that incorporated 0.18 mol% 10-undecen-1-ol, and showed Mw = 16,800 and Tm = 138.2◦ C. Using Me2 Si(H4 Ind)ZrCl2 , Wilen and Nasman101 copolymerized propylene with 6-tbutyl-2-(1,1-dimethylhept-6-enyl)-4-methylphenol, which is a thermooxidative stabilizer. The reaction is shown in Scheme 3.9. The findings can be summarized as follows: i. The synthesized copolymers incorporated 1.3 to 5.5 wt% phenolic units and showed high thermooxidative stability even after prolonged extraction with a mixture of refluxing 2-propanol:cyclohexane (50:50). ii. The addition of the phenolic stabilizer significantly increased the polymerization rate. Thus, the sterically hindered phenolic monomer was found as a polar activator and an effective comonomer to copolymerize with propylene. The initial polymerization rate, compared to that of the homopolymerization of propylene, increased almost 6 times when the phenolic comonomer was added. An increase in activity was also observed when 2,6-di-tert-butylphenol was added during propylene polymerization. The activity enhancement can be attributed to the ability of the phenolic stabilizer to scavenge the free AlMe3 present in MAO.102 iii. The products showed narrow molecular weight distribution, PDI ∼ = 2.0. iv. The incorporation of the comonomer decreased the crystallinity and melting point. Wilen et al.103,104 also copolymerized ethylene with 6-tert-butyl-2-(1,1-dimethylhept6-enyl)-4-methylphenol (a sterically hindered phenolic stabilizer) using Cp2 ZrCl2 , Me2 Si(H4 Ind)2 ZrCl2 , Et(H4 Ind)2 ZrCl2 in the presence of MAO. See Scheme 3.10. The copolymerization was conducted at 20◦ C and 1.6 bar in toluene. This approach of the copolymerization of a stabilizer with an olefin using metallocene proved to be unique for tethering stabilizers on the polyolefin backbone. The following results were obtained: i. The hindered phenol increased up to 3 times the initial rate of copolymerization over that of ethylene homopolymerization for Me2 Si(H4 Ind)2 ZrCl2 and

200

M. Atiqullah et al.

Scheme 3.10. Copolymerization of ethylene with 6-tert-butyl-2-(1,1-dimethylhept-6-enyl)-4methylphenol.103

ii.

iii.

iv. v. vi. vii. viii. ix.

Et(H4 Ind)2 ZrCl2 . On the contrary, the non-bridged Cp2 ZrCl2 showed no such increase in activity. The enhancement in catalyst activity was more for propylene (5-fold increase) than ethylene (2- to 3-fold increase). This finding can be attributed to the prochiral nature of propylene and the removal of AlMe3 via reaction with the phenolic comonomer. The level of comonomer insertion depended on the chirality of the metallocenes used. The comonomer insertion level (0.6–6.7 wt%) differed for Cp2 ZrCl2 . It was 2 to 3 times lower than that obtained with the bridged metallocenes. The phenol incorporation by Cp2 ZrCl2 is lower presumably because it is a poor catalyst for prochiral monomers. This is also reported by Kaminsky et al.105 The molecular weight distribution was well below 3 indicating single site catalysis. The Mn of the products synthesized by the above metallocenes ranged from 20,000 to 28,000 with PDI less than 2.6. The melting point and crystallinity of the resulting copolymers decreased with the increase in phenolic content as a result of increased branching. The thermo-oxidative stability of the copolymers produced was very high. The synthesized product was a random copolymer, which contains phenolic longchain branches. Polymerizations conducted at 20◦ C and 80◦ C produced copolymers with very high molecular weight and lower molecular weights, respectively. All the homo- and copolymer pairs showed fairly similar molecular weights and molecular weight distributions. The overall molecular weight was not influenced by the presence of the phenolic comonomer.

Mullins et al.106 copolymerized ethylene and propylene with an hindered allyl phenol (4-allyl-2,6-di-t-butyl phenol) using Me2 SiCp∗ (NtBu)TiMe2 and B(C6 F5 )3 cocatalyst. Hydrogen was used as a molecular weight regulator. Terpolymerization was also carried out using 1-octene. The molecular weight dropped more with ethylene than with propylene. The resulting polymers were reported to show improved coatability and blending characteristics.

Synthesis of Functional Polyolefins using Metallocenes

201

3.3.2 Copolymerization of Olefins with Acids, Andyrides, and their Derivatives. Maleic anhydride (MA)-modified polyolefins are one of the most important classes of functionalized polyolefins in commercial applications. The reason is that they combine low cost, high activity, and good processibility. They are generally chosen to improve compatibility, adhesion, and paintability of polyolefins. Lu and Chung66,107 synthesized maleic anhydride-modified polypropylene having a well-defined molecular structure using borane-terminated PP as an intermediate. The borane terminated PP was prepared by hydroboration reaction of chain end unsaturated PP (u-PP). The u-PP was suspended in THF and a slight excess of 9-BBN was added to initiate the reaction by stirring the whole mixture at 55◦ C for 5 h. The resulting 9-BBN terminated PP was oxidized in presence of maleic anhydride at room temperature by slowly adding stoichiometric quantity of oxygen (vs borane) for ∼4 h. Hydroboration converted this functionalized chain end-unsaturated polypropylene to the borane-terminated polypropylene. The borane group at the end of the polymer chain was selectively oxidized using oxygen to form a stable polymeric radical. This reacted in-situ with maleic anhydride (MA) to produce maleic anhydride-terminated PP (PP-t-MA) having a single MA unit. However, the polymeric radical also copolymerized with styrene (St) and maleic anhydride to produce the PP-b-StMA di-block copolymer. Because of the low tendency of homopolymerization by MA, the PP structure incorporated a low concentration of MA to form the final PP-b-MA copolymer. However, the addition of a small amount of styrene in the PP-B-MA mixture significantly increased the incorporation of MA into the copolymer backbone. The addition of styrene extended the PP chain end, forming an alternating styrene and MA (St-MA) copolymer. In other words, a diblock copolymer of PP-b-(St-MA) was obtained, containing both PP and St-MA segments. The reaction is shown in Scheme 3.11. Kaya et al.100 copolymerized propylene with the methylene spacer-containing 10undecenoyl chloride and 10-undecenoic acid using Et[Ind]2 ZrCl2 . The incorporation of the functional copolymers was 0.14–0.20 mol% while the melting point ranged from 136.2 to 138.7◦ C. The copolymerization with the undecenoyl chloride R-COCl is of interest for the synthesis of polymers with different functionalities through post-modification. The addition of an appropriate polymerization termination agent can incorporate different functional groups in the copolymer. For instance, the addition of water, amines, or alcohols to the polymer solution after the completion of polymerization results in copolymers with carboxylic acid, amide, or ester functionalities, respectively. In the above study, the 1H NMR spectra analysis of the copolymer obtained with R-COCl revealed that free carboxyl acidic groups formed in the polymer backbone during the termination of the polymerization by adding water to the system.

Scheme 3.11. Reaction scheme showing the synthesis of PP-b-(St-MA) copolymer.107

202

M. Atiqullah et al.

The synthesis of highly soluble reactive intermediate precursor and functional polyolefins without degradation or crosslinking is a challenge. Zheng et al.38,39 found that ethylene-allylbenzene copolymer originally prepared by Byun et al.34,40 can overcome this situation. They incorporated glutaric anhydride (GA), succinic anhydride (SA), phthalic anhydride (PA), and chlorosulfonic acid at the para-postion of the pendant phenyl ring of the ethylene-allylbenzene copolymer. Friedel-Crafts (FeC) acylation reaction, in the presence of anhydrous aluminum chloride in carbon disulfide introduced the aforesaid anhydrides. The melting temperatures (Tm s) of the acylated copolymers with GA first decreased slowly; then increased significantly with the increase of the amount of carboxyl acid groups.38 The reaction of the ethylene-allylbenzene copolymer with chlorosulfonic acid in 1,1,2,2-tetrachloroethane followed by hydrolysis introduced the acid functionality. The melting temperature increased with the degree of sulfonation. The sulfonated copolymers increased the degradation temperature from 444 to 460◦ C and the surface hydrophilicity compared to the base copolymer.38 3.3.3 Copolymerization of Olefins with Ethers. The copolymerization of olefins with ethylene oxide synthesizes amphiphilic copolymers. Amphiphilic copolymers contain hydrophobic and hydrophilic blocks and show interesting surface properties. They are ideal candidates for many applications such as emulsifiers, dispersants, stabilizers, antifoaming agents in aqueous solutions, and compatibilizers in polymer blends and composites. Lu et al.108 carried out diblock copolymerization of ethylene oxide with polyethylene, syndiotactic polystyrene, poly(ethylene-co-1-octene) and poly(ethylene-co-styrene). The copolymerization was done in two steps; in the first step, borane-terminated polyolefin was synthesized using metallocenes and borane chain transfer agents. In the second step, the borane terminal group was converted to an anionic (—O−K+) terminal group for the subsequent ring-opening polymerization of ethylene oxide. The reactions are detailed in Scheme 3.12. The yield and incorporation of ethylene oxide mildly increased with the initial feed concentration. However, the molecular weight increased far greater. 3.3.4 Copolymerization of Olefins with Oxazolines. Kaya et al.100,109 used Et[Ind]2 ZrCl2 and Me2 Si[2-Me-4,5-BenzoInd]2 ZrCl2 to copolymerize propylene with the following functional comonomers: 2-(9-decene-1-yl)-1,3-oxazoline, 2-(9-decen-1-yl)-4,4-dimethyl-1,3oxazoline, and 2-(4-(10-undecene-1-oxo)phenyl)-1,3-oxazoline. The structures of these comonomers are given in Fig. 3.8. Up to 0.52 mol% oxazoline was incorporated into the polypropylene backbone. The Mn values (8.0 × 103 to 12.0 × 103) evidence the formation

Scheme 3.12. Diblock copolymerization of ethylene oxide with polyethylene, syndiotactic polystyrene, poly(ethylene-co-1-octene) and poly(ethylene-co-styrene).108

Synthesis of Functional Polyolefins using Metallocenes

203

Figure 3.8. Structures of the comonomers: 2-(9-decen-1-yl)-1,3-oxazoline (R- Ox1), 2-(9-decen1-yl)-4,4-dimethyl-1,3-oxazoline (R-Ox2), 2-[4-(10- undecene-1-oxy)phenyl]-1,3-oxazoline (ROx3).100

of approximate cooligomeric products. The acid-functionalized copolymer was obtained by hydrolyzing the 10-undecenoyl chloride copolymer solution. The steric hindrance in the oxazoline ring of the comonomer R-Ox2 potentially reduced catalyst poisoning compared to other oxazolines. An increase in the polymerization temperature decreased the molecular weight, crystallinity, and the melting point of the resulting copolymers.

4. Nitrogen-Containing Polar Functional Groups Nitrogen-containing polar functional groups cover amines, acryl amides, imides, and so on. Polymers containing amino groups have been widely studied and are useful functional polyolefins. This may be attributed to the chemical and structural versatility of the amino functional group. The reactions of amines include those of bases and nucleophiles. 4.1 Copolymerization of Olefins with Amines The structural and functional versatility of amines leads to diverse applications that range from polymer precursors for other functional polymers to ion exchange resins, analytical and preparative chromatography, solid phase synthesis and catalysis, drug delivery, metal ion extraction (hydrometallurgy), light harvesting, and chemical sensors. Amines can enhance adhesion and miscibility in polymer blends and composites, thereby improving the physicalmechanical properties of the polymer. Amines are introduced to the polyolefin backbone either by copolymerization of aminecontaining monomers,110,111 or post-functionalization of polyolefins.112 Stehling et al.110 copolymerized α-olefins such as propylene or 4-methylpentene with alkene-substituted alkoxy amines using [Et(H4Ind)2 ZrMe2 ]/[HNPhMe2 ]+[BPh4 ]−. The copolymerization of propylene with the alkoxyamine offered a product with Mn = 28,000, and PDI = 1.8. The incorporation of the alkoxy amine into the copolymer was well correlated to the feed ratio. The reaction is detailed in Scheme 4.1. The alkoxy amine pendant group in polyolefin backbone was used to initiate free radical polymerization. The above copolymer was also heated at 123◦ C with styrene in presence of acetic anhydride to form graft copolymer. The resulting graft copolymer showed Mn of 210,000 and PDI of 2.0.

204

M. Atiqullah et al.

Scheme 4.1. Copolymerization of α-olefins such as propylene or 4-methylpentene with alkenesubstituted alkoxy amines using [Et(H4 Ind)2 ZrMe2 ]/[HNPhMe2 ]+[BPh4 ]−.112

Stehling et al.111 also polymerized functionalized α-olefins such as 5-amino-1-pentenes and 4-amino-1-butenes using dimethyl zirconocenes [(Me5 C5 )2 ZrMe2 , Et(Ind)2 ZrMe2 , (Ind)2 ZrMe2 , Et(H4 Ind)2 ZrMe2 , iPr(tBuCp)(Flu)ZrMe2 ] activated with anilinium borate [HNMe2 Ph]+[B(C6 F5 )4 ]−. The polymerization was done in toluene at room temperature for 1 h. (Me5 C5 )2 ZrMe2 /borate showed the highest polymerization activity for 5-(N,Ndiisopropylamino)1-pentene. (Me5 C5 )2 ZrMe2 /borate was 4 times more active than the (Me5 C5 )2 ZrMe2 /MAO system. The polymer product was characterized by 1H and 13 C NMR spectroscopy. 5-(N,N-diisopropylamino)-1-pentene polymerized to isotactic polyaminopentene by [Et(Ind)2 ZrMe2 ], syndiotactic polymer by [iPr(tBuCp)(Flu)ZrMe2 ], and atactic polymer by (Ind)2 ZrMe2 . The Mn varied from 800 to 8,000. Shiono et al.113,114 synthesized the atactic and isotactic polypropylenes having terminal vinylidine groups using Cp2 ZrCl2 and Et(H4 Ind)2 ZrCl2 . Et(H4 Ind)2 ZrCl2 synthesized isotactic PP in toluene at 20◦ C whereas Cp2 ZrCl2 prepared atactic PP at 0◦ C. The products having vinylidene end groups were first treated with excess borane in benzene, then with excess 1-hexene. The interaction of borane with vinylidine end groups alkylated the borane. The resulting trialkylborane was disproportionated with boron trichloride in xylene at 110◦ C to obtain alkyldichloro borane. This was reacted with 1-butylazide to synthesize polypropylenes having a 1-butylamino group at the end of the polymer chain. The yield was over 80%. The polymer was characterized using FTIR and 1H and 13C NMR spectroscopy. The Tm varied from 115.6 to 116.4◦ C. Shiono et al.,115 in another study, first produced atactic polypropylene macromer (PPM) which was later copolymerized with propylene using Me2 Si[2-Me-4,5-BenzoInd]2 ZrCl2 . The objective was to synthesize isotactic PP with an atactic side chain by copolymerization of propylene with the atactic PPM. The atactic PPM was synthesized using Cp∗ 2 ZrCl2 /MAO and liquid propylene. Here, Cp∗ 2 ZrCl2 /MAO produced vinyl-terminated PP via β-Me transfer, which in contrast to vinyledene-terminated PP, acted as the desired PPM for copolymerization with propylene. It showed a number average molecular weight of 630 and polydispersity index of 2.4. The PPM content in the final copolymer was up to 1.3 mol%. The Mn of the copolymer varied from 45,000 to 48,000. Dong et al.29 synthesized isotactic polypropylene with terminal amine groups using Me2 Si[2-Me-4-Ph(Ind)]2 ZrCl2 . Propylene was copolymerized with an amino-substituted

Synthesis of Functional Polyolefins using Metallocenes

205

styrene [4-{2-[N,N-bis(trimethylsilyl)amino] ethyl}styrene] which also acted as a chain transfer agent. The amino substituent was protected by trimethyl silane. Therefore, the first product was PP-t-St-NSi2 which was finally converted into amine styrene-terminatedpolypropylene, PP-t-St-NH2 , by reacting with hydrochloric acid. The polymer produced showed Mn in the range of 24,000 to 59,000 with PDI 2.1 to 2.5. The Tm varied from 156 to 159◦ C. Hackman et al.116 copolymerized propylene with isocitronellene using Et[Ind]2 ZrCl2 . The lower temperature (30◦ C) favored isocitronellene incorporation more than the higher temperature (60◦ C) however converse was the effect on the activity. The higher temperature lowered the molecular weight. The isocitronellene content in the resulting copolymer varied from 7.8 to 15.6 mol%. 4.2 Copolymerization of Olefins with Acrylonitrile Chung et al.61 copolymerized polyethylene with acrylonitrile using PE-co-p-MeSt as an intermediate. The p-MeSt in PE was effectively lithiated at ambient temperature. The details are given in Section 2.1.1. The resulting lithiated PE-p-MeSt copolymer contained several polymeric anions that were homogenously distributed in the polymer chain. See Scheme 4.2. The lithiated PE-p-MeSt initiated graft polymerization of acrylonitrile. The resulting polymer products were solvent fractionated to separate the homopolymer. In most cases, less than 10 wt% of the homopolymer was obtained. Hexane mildly increased the yield; the incorporation of acrylonitrile was more with hexane than THF. Overall the graft copolymerization of acrylonitrile is less efficient than that of styrene. Polar THF gave poor yield while the nonpolar cyclohexane improved the results. The polar solvent may increase the nucleophilicity of the carbanion, which causes more side reactions. The anionic polymerization of acrylonitrile using butyl lithium as an initiator could not offer high polymer yield at ambient temperature. Usually very low reaction temperature (< –20◦ C) is required. In lithiated PE-p-MeSt, the formed polymeric benzylic lithium is much more stable and therefore, it minimizes the side reactions. The acrylonitrile content varied from 15.0 to 51.2 mol%. Lu and Chung66 graft copolymerized polypropylene with acrylonitrile using poly(propylene-co-p-MeSt) as the starting material. The synthesis of PP-co-p-MeSt and its lithiation is already detailed in Section 2.2.2. The anionic graft copolymerization of acrylonitrile took place at room temperature without any side reaction. The acrylonitrile incorporation reached up to 50 mol%.

Scheme 4.2. Copolymerization of polyethylene with acrylonitrile using PE-co-p-MeSt as an intermediate.61

206

M. Atiqullah et al.

4.3 Copolymerization of Olefins with Carbazole Mustonen et al.117 copolymerized ethylene with 9(bicycle[2,2,1]hept-5-en-2-ylmethyl)9H-carbazole (BHMCZ) using Ph2 C(Flu)CpZrCl2 and Ph2 C(Ind)(Cp)ZrCl2 . The polymerization conditions and the π -carboxylic ligand type influenced the formation of the cooligomers. The Flu and Ind ligands produced Mw of 6,250 and 70,200, respectively.

5. Copolymerization of Olefins with Halogen-Containing Polar Functional Groups Shiono and Soga87,118 synthesized terminally halogenated isotactic polypropylene. First, propylene was homo-polymerized using Et(H4 Ind)2 ZrCl2 . The product having terminal C C groups was hydroaluminated by di-isobutylaluminum hydride (isoBu2 AlH) to produce polypropylene having aluminum at one end. This aluminum-terminated polypropylene was next halogenated to produce halogen-terminated polypropylene. See Scheme 5.1. Bruzaud et al.119 carried out homo-, co- and terpolymerization of 11-chloro undec-1ene with ethylene, propylene, and 1-hexene using Et(Ind)2 ZrCl2 . The homo-polymerization of 11-chloroundec-1-ene was done in various solvents. Polymerization did not occur in CH2 Cl2 solvent; however, in toluene or heptane it proceeded to complete conversion. The copolymerization of 1-hexene was also done with 11-chloroundec-1-ene in heptane at molar ratio of 50:60. The copolymer obtained showed Mn of 9,000 with PDI ≈ 2.0. The homo and copolymerization of 5-chloropent-1-ene was also tried. No polymerization occurred, indicating that the catalyst was poisoned. The terpolymerization of 11-chloroundec-1-ene with ethylene and propylene in heptane was conducted at 20◦ C by keeping the ethylene pressure at 1.5 bar while that of propylene at 3.5 bar. The concentration of 11-chloroundec-1-ene was increased up to 0.15 mol/l. The incorporation of the above-chlorinated monomer was maximum 2.0 mol% with Mn of 66,000. The polymers produced in this way contained chlorine pendant groups. These chlorocontaining polymers were first treated with potassium benzoate under phase transfer condition to produce esterified polymer which was further hydrolyzed with potassium hydroxide to produce polymers having hydroxyl pendant groups.

6. Copolymerization of Olefins with Silane-Containing Functional Groups Fu and Marks120 polymerized ethylene with PhSiH3 using (Me5 C5 )2 LnH and Me2 Si(Me4 C5 )2 LnH lanthanocenes. The Mn of the resulting polymers varied from 400 to 98,000 with PDI ranging from 1.8 to 4.9. This shows the promise of lanthanocenes to synthesize oligomeric to polymeric products.

Scheme 5.1. Synthesis of halogen-terminated polypropylene.87,118

Synthesis of Functional Polyolefins using Metallocenes

207

Scheme 6.1. Synthesis of silyl-caped polyolefins either (Me5 C5 )2 LnH or Me2 Si(Me4 C5 )2 LnH.120,121

Fu and Marks120 and Koo et al.121 have reported efficient and selective silanolytic (PhSiH3 , nBuSiH3 , and PhCH2 SiH3 ) chain transfer in organo-lanthanide-catalyzed homogeneous ethylene polymerization and ethylene copolymerization with several α-olefins. A series of silyl-capped polyolefins were produced. In case of homo- and copolymerization of ethylene with 1-hexene using either (Me5 C5 )2 LnH or Me2 Si(Me4 C5 )2 LnH, the primary aryl silane (PhSiH3 ) and the alkyl-silanes (nBuSiH3 , PhCH2 SiH3 ) functioned as efficient chain transfer agents. The Mn ranged from 1,100 to 3,200. The reaction mechanism is shown in Scheme 6.1. Marks and Koo122 copolymerized propylene, 1-hexene, and styrene with PhSiH3 using Me2 SiCp∗ (NtBu)TiMe2 and borate cocatalyst [Ph3 C]+[B(C6 F5 )]−. The objective was twofold: a. Synthesize silyl-terminated polymer; and b. Evaluate the chain transfer capability of the experimental silane. The chain transfer capability of PhSiH3 was affected by the monomer type as follows: propylene (Mw = 43,000), 1-hexene (Mw = 2,500), and styrene (Mw = 72,000). It appears that 1-hexene also acted an in-situ chain-transfer agent. Makio et al.123 copolymerized ethylene with n-hexyl-SiH3 using the supported catalyst system SiO2 /MAO/Cp2 ZrCl2 . The activity and molecular weight decreased with the increase of the silane amount in the fresh feed. The product molecular weight varied from 6,000 to 7,000. Arriola et al.124 cooligomerized propylene with allyldimethylsilane in the presence of hydrogen. They used Me2 Si(Me4 C5 )(tBuN)Ti(1,3-pentadiene) and (PhF5 )3 B cocatalyst. The molecular weight of the resulting product varied from 78,000 to 232,000.

7. Copolymerization of Olefins with Dienes and Cyclic Olefins The copolymerization of ethylene or α-olefins with cyclic olefins produces cycloolefin copolymers (COC), a new amorphous thermoplastic material. COC polymers are characterized by excellent transparency and very high, long-life service temperatures. They are solvent resistant and can be melt-processed.

208

M. Atiqullah et al.

Marathe and Sivaram125 copolymerized ethylene with 5-vinyl-2-norbornene (VNB) using Cp2 ZrCl2 . The copolymer produced showed pendant vinyl groups, which were oxidized to epoxy functional groups. See Scheme 7.1. The VNB content in the synthesized copolymer ranged from 6 to 14 mol%. The EVNB copolymer was then oxidized using m-chloroperbenzoic acid to produce the epoxy functional groups. The pendant vinyl groups of the resulting copolymer were also oxidized to hydroxyl functional groups as follows. The copolymer was first hydroborated with 9-BBN, then oxidized using NaOH and H2 O2 . See Scheme 7.2. Radhakrishnan and Sivaram126 copolymerized ethylene with a symmetrical diene namely 2,5-norbornadiene (NBD) using Cp2 ZrCl2 , (nBuCp)2 ZrCl2 , Et(Ind)2 ZrCl2 and Me2 Si(Cp)2 ZrCl2 . See Scheme 7.3. Ethylene readily copolymerized with NBD by Cp2 ZrCl2 through one of the double bonds. Copolymers with as high as 19 mol% NBD were synthesized without any cross linking which was evidenced by the solubility of the resulting copolymers in toluene at room temperature. (nBuCp)2 ZrCl2 increased the catalyst activity as well as the copolymer molecular weight, maintaining the same level of NBD incorporation. In case of Et(Ind)2 ZrCl2 , the catalyst activity as well as the incorporation of NBD with increase in molecular weight was observed. Me2 SiCp2 ZrCl2 produced an insoluble copolymer of ethylene and NBD.

Scheme 7.1. Copolymerization of ethylene with 5-vinyl-2-norbornene which was subsequently oxidized to epoxy functional groups.125

Scheme 7.2. Copolymerization of ethylene with 5-vinyl-2-norbornene which was subsequently oxidized to hydroxyl functional groups.125

Scheme 7.3. Copolymerization of ethylene with a symmetrical diene namely 2,5-norbornadiene.135

Synthesis of Functional Polyolefins using Metallocenes

209

The insoluble copolymer showed no unsaturation in FT-IR spectrum, indicating the participation of both endocyclic double bonds in copolymerization that cross-linked the various polymer chains. Monkkonene and Pakkanen127 also copolymerized ethylene with 2,5-norbornadiene using Cp2 ZrCl2 and [Ph2 C(Ind)(Cp)ZrCl2 ]. Cp2 ZrCl2 offered higher activity and molecular weights with less incorporation of NBD than [Ph2 C(Ind)(Cp)ZrCl2 ]. Ethylene copolymerized with NBD in the presence of Cp2 ZrCl2 through one of the double bonds. This produced only one type of copolymer structure with unsaturation on the incorporated NBD. However, [Ph2 C(Ind)(Cp)ZrCl2 ]-catalyzed copolymerization proceeded through both the double bonds. As a result a mixture of copolymers having saturated as well as unsaturated NBD was obtained. The activity of both the catalysts dropped with the increase in the NBD:ethylene feed ratio. The polymers with high NBD content were amorphous and transparent. Uozumi et al.128,129 copolymerized ethylene with 1,9-decadiene using Me2 Si(Flu)2 ZrMe2 . They noted that the combined increase of the diene and the polymerization time increased the copolymer yield and the weight average molecular weight which ranged from 328,000 to 450,000. Schiffino and Crowther130 synthesized an elastomeric terpolymer that consisted of ethylene, propylene, and a diene (ethylidene-norbornene, ENB, using Me2 Si(2, 4Me2 C5 )(Flu)ZrCl2 . The Mn of the resulting products ranged from 15,853 to 19,314; however, the PDI varied from 10.27 to 12.17. The EBN and PP content varied from 2.71−3.15 wt%, and 10.52−11.64 wt%, respectively.

8. Application of Protection and Deprotection Concepts to Synthesize Functional Polyolefins The copolymerization of olefins with polar comonomers encounters several difficulties. The polar groups partly deactivate the catalyst. The transition metals in the metallocene catalysts are often killed by the protic functionality and/or poisoned by heteroatoms such as N, O, etc. High incorporation level of polar monomers usually decreases the molar mass of the polymer.99 However, efforts to prevent deactivation of metallocenes during copolymerization with polar monomers are in progress.131–134 In general, the synthesis of functional polyolefins has involved modifying the catalyst or protecting the monomer itself. The polar comonomers successfully copolymerize with olefins when the Lewis base feature of the comonomer is reduced by masking them with aluminum alkyls, or silanes before copolymerization.134 This step is called protection. After copolymerization, the copolymer is treated with a suitable acid or other chemicals which regenerate these oxygen functional groups. The latter step is called deprotection. The functional groups of polar comonomers have been protected using pre-treatment of the comonomers with alkylaluminum and alkylsilyl compounds. What follows summarizes these protection methods.

8.1 Protection of Functional Groups through Pretreatment with Alkylaluminum Compounds Here, the polar monomers are pretreated with excess trialkylaluminum compounds such as (TIBA) or methylaluminoxane (MAO). However, this approach lowers effective comonomer incorporation into the polymer chain.

210

M. Atiqullah et al.

Scheme 8.1. Copolymerization of ethylene with TMA-protected 5-Norbornene-2-methanol using Cp2 ZrCl2 , Et(Ind)2 ZrCl2 and Me2 SiCp2 ZrCl2 .135

Radhakrishnan and Sivaram135 copolymerized ethylene with TMA-protected bicyclo [2.2.1]hept-5-enen-2-methanol (5-Norbornene-2-methanol) using Cp2 ZrCl2 , Et(Ind)2 ZrCl2 and Me2 SiCp2 ZrCl2 . The reaction is shown in Scheme 8.1. The addition of acidic methanol to terminate the polymerization also acted as a deprotection reagent and converted the attached aluminum to alcohol. Me2 Si(Cp)2 ZrCl2 incorporated the maximum comonomer (6.2 mol%) which decreased the catalytic activity. However, the higher Al:Zr ratio increased the activity as well as comonomer incorporation. This observation is quite opposite to the ethylene/bicyclic olefin copolymerization, where the comonomer incorporation decreased with the increase in Al:Zr ratio. The increase in copolymerization temperature also increased the activity. The DSC curve showed two separate melting peaks, indicating a mixture of copolymer with the homopolymer of ethylene. Marques et al.136 synthesized co- and terpolymers of ethylene, propylene, and TMA-protected polar vinyl comonomers having OH and COOH functional groups using Et(Ind)2 ZrCl2 (1), Me2 Si(Me4 Cp)(NtBu)TiCl2 (2), and (2-MeBenzoInd)2 ZrCl2 (3), all activated by MAO. The vinyl comonomers comprised 5-hexen-1-ol and 10-undecen-1-ol and 10-undecenoic acid. TMA and MAO were used to protect the respective functional groups. The above catalysts produced functionalized co- and terpolymers by direct polymerization of ethylene/propylene/hydroxyl-α-olefins. However, Et(Ind)2 ZrCl2 showed appreciable activities for direct polymerization of ethylene, propylene, and carboxy-α-olefins. (2-MeBenzoInd)2 ZrCl2 exhibited better tolerance toward hydroxyolefins. MAO was not as effective as TMA to protect the alkoxy groups by preventing them from binding to the active site. The overall findings can be summarized as follows: i. The presence of the hydroxyl olefins did not significantly decrease the molecular weight of the resulting co- and terpolymers. ii. The molecular weight of the polymers obtained with Me2 Si(Me4 Cp)(NtBu)TiCl2 were slightly lower (Mn = 17.5 × 103 to 54 × 103) than those obtained with Et[Ind]2 ZrCl2 (Mn = 24.7 × 103 to 197 × 103). However, the polydispersities remained in the same range (from 2.0 to 3.9). iii. The molecular weight of the polymers obtained with (2-MeBenzoInd)2 ZrCl2 were, in general, higher than those obtained with the other two catalysts (Mn = 92.8 × 103 to 309.8 × 103). The increasing trend of Mn Me2 Si(Me4 Cp)(NtBu)TiCl2 ) < Et[Ind]2 ZrCl2 < MeBenzoInd)2 ZrCl2 may be attributed to the decreasing space available around the metal center, which is imposed by the ligand architecture. The less space that is available, the less susceptible is the metal center to β-hydrogen elimination; hence, a higher Mn is achieved. Goretzki and Fink132 copolymerized ethylene with 10-undecen-1-ol and 5-norbornene2-methanol protected by trimethyl and triisopropyl silyl groups using homogenous as well as supported Me2 Si(Ind)2 ZrCl2 and iPr(Cp)(Ind)ZrCl2 .

Synthesis of Functional Polyolefins using Metallocenes

Figure 8.1. A terpolymer backbone of an terpolymer—significance of the carbon atoms.138

211

ethylene/norbornene/norbornene

derivative

Protection by trimethyl silane decreased the catalyst activity whereas triisopropyl silane showed good catalyst activity and comonomer insertion. The incorporation of 10-undecen1-oxytriisopropyl silane in the copolymer by the supported catalysts was significantly lower than that by the homogeneous analogues. Goretzki and Fink133 copolymerized ethylene with 10-undecen-1-ol, 5-(N,Ndiisopropylamino)-1-pentene [a sterically hindered amine], and 5-norbornene-2-methanol using homogenous as well as supported Me2 Si(Ind)2 ZrCl2 , iPr(Cp)(Ind)ZrCl2 and iPr(3Me-Cp)(Flu)ZrCl2 ). Here, the functional groups were protected by triisobutyl aluminum (TIBA). This produced alkoxy aluminum alkyls which maintained the polymerization activity and introduced good comonomer insertion. At the end of polymerization, the free hydroxyl groups were regenerated using acidic methanol. The following was observed: i. Pretreating the comonomers with triisobutylaluminum (TIBA) prevented rapid catalyst deactivation. ii. 5-(N,N-diisopropylamino)-1-pentene could be copolymerized without previous complexing with TIBA, because the basic nitrogen of the amino group is mainly protected by the isopropyl groups. iii. The molecular weight of the resulting copolymers varied as a function of the comonomer and catalyst type. See below. Comonomer

Catalyst

10-undecen-1-ol Me2 Si(Ind)2 ZrCl2 /MAO/SiO2 i Pr(3-Me-CpFlu)ZrCl2 5-norbornene-2-methanol 5-(N,N-diisopropylamino)-1-pentene Me2 Si(Ind)2 ZrCl2 Me2 Si(Ind)2 ZrCl2 /MAO/SiO2

Mw 54,200 78,200 to 111,400 21,000 to 146,000 269,000 to 290,000

Alexander and Fink137 copolymerized and terpolymerized ethylene with norbornene having polar groups of alcohol and acid using iPr(Cp)(Ind)ZrCl2 . The polar groups of norbornene were passivated by reacting with triisobutyl aluminum (TIBA). The polar groups were incorporated in the range of 5–12 mol%. In addition to TIBA, trialkyl silanes were also used as protecting agents. The silanes incorporated 5–6 mol% polar groups. Wendt and Gerhard138 synthesized functionalized copolymers and terpolymers of ethylene with TIBA-protected functional comonomers using iPr(Cp)(Ind)ZrCl2 (Fig. 8.1). Protection using TIBA prevented the catalyst from deactivation. The functional comonomers included the norbornene derivatives, 5-norbornene-2-methanol, and 5-norbornene-2carboxylic acid.

212

M. Atiqullah et al.

Scheme 8.2. Copolymerization of ethylene with 5-hexene-1-ol pretreated with trimethyl aluminum (TMA) using Et(Ind)(Flu)ZrCl2 .139

The overall findings are summarized below: i. Pre-treatment (protection) of 5-norbornene-2-carboxylic acid with TIBA increased the copolymerization activity with ethylene. However, this increase depended on the feed pressure of ethylene. Molecular weight in the cooligomeric product range was obtained. The average copolymer composition was about 4.5 mol%. ii. Pre-treatment (protection) of 5-norbornene-2-carboxylic acid with TIBA also increased the activity of the above system during terpolymerization with norbornene. Here, norbornene was incorporated more than the acid analogue. It also increased Mw . iii. In ethylene/norbornene/5-norbornene-2-methanol terpolymerization, the activity of the catalyst system was not markedly reduced with the increasing content of 5norbornene-2-methanol in the feedstock. Norbornene was incorporated more than the methanol analogue. iv. Ethylene/norbornene/5-norbornene-2-methanol terpolymerization offered higher activity than the corresponding terpolymerization with the norbornene acid analogue even without protection by TIBA. Mw increased by about two-fold depending on the feed composition. What follows summarizes the copolymerization of 5-norbornene-2-methyleneoxytriethylsilane (TES) and 5-norbornene-2-methyleneoxy-tert-butyldimethylsilane (TBDMS) with ethylene. The protection group was removed by treating the resulting copolymer with a dilute solution of hydrochloric acid in methanol. TBDMS offered higher Mw than TES. The terpolymerization of ethylene/norbornene/5-norbornene-2-methyleneoxyTBDMS using iPr(Cp)(Ind)ZrCl2 can be listed as follows:138 i. TBDMS offered higher activity and molecular weight than TES. ii. In either case, norbornene was incorporated more than norbornene methanol. However, when incorporating both of these two monomers the trialkyl silanes showed a comparable performance independent of the structure. Hagihara et al.139 copolymerized ethylene with 5-hexen-1-ol that was pretreated with TMA which effectively masked the hydroxyl group. Et(Ind)(Flu)ZrCl2 was used as the catalyst. The reaction is shown in Scheme 8.2. The copolymerization activity highly depended on the ratio of [5-hexene-1-oxyl group]:[TMA]. The resulting copolymer incorporated 5-hexen-1-ol up to 50 mol% with almost an alternating sequence. This catalyst system has also been reported to copolymerize ethylene with α-olefins in an alternating manner. The copolymer was insoluble in hot

Synthesis of Functional Polyolefins using Metallocenes

213

toluene; however, it dissolved in polar THF, indicating the presence of a polar group in the copolymer backbone. The PDI was in the range of 1.4 to 2.0. The increasing concentration of TMA lowered the molecular weight of the resulting copolymers due to chain transfer to the Al atom.

8.2 Protection of Functional Groups through Pretreatment with Alkylsilyl Compounds Kesti et al.132 polymerized silyl-protected alcohols, amines, and nonconjugated dienes using selected cationic Group IV zirconocene catalysts. They found that silyl ethers more easily poisoned [(H4 Ind)ZrMe]+X− than [Cp2 ∗ ZrMe]+X−. The chiral [(H4Ind)ZrMe]+X− actively polymerized 4-(tert-butyldimethylsilyloxy)-1,6-heptadiene and 5-(diisopropylamino)-1-pentene but not 4-(trimethylsilyloxy)-1,6-heptadiene or 4(tert-butyldimethylsilyloxy)-1-pentene. Steric effect may protect the polar group. However, the high steric interactions between the protected monomer and the catalyst ligand may decrease the insertion of the polar comonomer into the polymer chain.132,134 Where special trialkylsilyl compounds were utilized as alternative protection groups for norbornenemethanol with lower steric demand to evaluate this effect and to investigate alternative protection groups allowing for higher incorporation rates. The importance of sterical contribution of the protecting group to limit the accessibility of oxygen atom (catalyst poisoning moiety) to the metallocene metal was investigated by copolymerizing ethylene with the polar norbornene derivatives.140,141 The experimental metallocenes were iPr(Cp)(Ind)ZrCl2 , iPr(3-iPrCp)(Ind)ZrCl2 , and iPr(3t BuCp)(Ind)ZrCl2 . Using various trialkylsilyethers with high sterical demand, Fink et al.140 reported high activities and comonomer incorporation (up to 12 mol%) in co- and terpolymerization of these norbornene derivatives with ethylene and norbornene. Figure 8.2 shows the various trialkylsilyl-protected norbornene derivatives that were used in this study. The use of metallocenes with differently substituted Cp ligands affected the polymerization activity due to the various steric interactions between the ligand and the polar norbornene derivatives. The hindering of the formation of 3 (Scheme 8.3) is envisioned to be the key to prevent the catalyst from deactivation. The increasing sterical demand by the protection groups as well as the catalysts formed the inhibited Zr-O-complex and made the norbornene-olefin insertion more favorable. A relationship between the catalyst activity and the steric demand of the protecting group was observed. The kinetic investigations point to a reversible deactivation reaction, during which a bond between the oxygen atom of the polar norbornene derivative and the center of the active catalyst is formed that competes with the olefin coordination and the subsequent insertion. What follows summarizes the results of copolymerization of ethylene with norbornenemethanol protected by iso-propyldimethyl (IPDMS), trialkylsilyl (TES), tertbutyldimethylsilyl (TBDMS), thixyldimethylsilyl (TDMS), and triisopropylsilyl (TIPS), catalyzed by iPr(Cp)(Ind)ZrCl2 . The protecting silyls decreased the activity as follows:141 IPDMS (1a) < TES (1b) < TBDMS (1c) < TDMS (1d) < TIPS(1e) The above order shows that the degree of catalyst deactivation depended on the structure of the protection group. The activity also decreased with the increasing content of norbornene derivative feed composition. The Mw ranged from 1,600 to 12,000.

214

M. Atiqullah et al.

Figure 8.2. The structure of alkyl silanes used to protect norbornenemethanol.140, 141

Hakala et al.134 copolymerized ethylene with oxygen-containing comonomers such as 10-undecen-1-ol (I), 10-undecenyl methyl ether (II), 10-undecenyl triethyl silyl ether (III), and 1-undecene (IV) using Et(Ind)2 ZrCl2 . Figure 8.3 shows the structures of these comonomers. The objective was to investigate how the molecular structure around the oxygen atom influences the comonomer incorporation. The comonomer with hydroxyl (I) or ether (II) functionality copolymerized with ethylene under mild conditions offering moderate catalyst activity. However, this activity was much lower than the ethylene

Figure 8.3. Structures of the long-chain oxygen-containing comonomers: (I) 10- undecen-1-ol, (II) 10-undecenyl methyl ether, (III) 10-undecenyl trimethyl silyl ether, and (IV) 1-undecene.134

Synthesis of Functional Polyolefins using Metallocenes

215

Scheme 8.3. Insertion mechanism including deactivation reaction between polar norbornene derivative (comonomer 1) and the metallocene catalyst system 2.140,141

homopolymerization activity. The conversions of the functional comonomers (I and II) were up to 40% whereas that of IV was up to 75%. The catalytic activity for all the comonomers was almost comparable indicating that the trimethyl silyl groups did not act as effective protective groups. The NMR characterization of the final polymer revealed that final functional group in the copolymer of silyl ether was hydroxyl; however, in methyl ether it remained unchanged. The substitution of the hydroxyl group with a trimethylsilyl or methyl ether group did not affect the copolymerization behavior of the comonomers. Similar observations were reported by Goretzki and Fink.132 The copolymerization of functional vinyl comonomers by metallocene catalysts is possible when their Lewis basicity is reduced by masking the functional moiety with alkylaluminums or by adding silyl or other protecting groups to the heteroatom.112,131,132,136 In the copolymerization with 10-undecenyl methyl ether

216

M. Atiqullah et al.

(II), the methyl group remained unreacted. This demonstrates that the protection of the oxygen atom via the formation of aluminum alkoxide is not a prerequisite for comonomer incorporation. The oxygen-functionalized comonomers significantly decreased the polymerization activity. Substituting the hydrogen atom of the hydroxyl group of comonomer I with larger methyl or trimethylsilyl groups did not affect the magnitude of catalyst deactivation. MAO did not show any noteworthy differences in the ability of the alcohol I and ethers II and III to copolymerize. As the comonomer copolymerized and even the methyl ether group remained unreacted in the polymerizations reveal that the protection of the oxygen atoms via the formation of aluminum alkoxides, as in the reaction between the hydroxyl group and MAO, is not necessary for comonomer incorporation. The addition of nonfunctionalized IV to the reaction medium was found to increase the polymerization activity. The molecular weights of all the functionalized copolymers were markedly lower than that of the ethylene homopolymer. This effect was more pronounced with the functional comonomers (I, II, and III) than with the nonfunctional analogue (IV). The melting point of the copolymers also decreased. This indicates lower crystallinity due to the incorporation of the side chains. Some of the copolymers with higher amounts of hydroxyl functionality were not completely soluble in 1,1,2,2-tetrachloroethane or 1,2,4-trichlorobenzene. In contrast, all the copolymers with ether functionality were completely soluble. Goretzki and Fink132 used (a) Me2 Si(Ind)2 ZrCl2 and Me2 Si(Ind)2 ZrCl2 /MAO/SiO2 to copolymerize ethylene with 10-undecene-1-oxytrimethylsilane and 10-undecene1-oxytriisopropylsilane, and (b) iPr(Cp)(Ind)ZrCl2 and iPr(Cp)(Ind)ZrCl2 /MAO/SiO2 to copolymerize ethylene with 5-norbornene-2-methyleneoxytrimethylsilane and 5norbornene-2 methyleneoxytriisopropylsilane. They found the following. The trimethylsilyl (TMS) protecting group could not prevent the catalyst from deactivation caused by the addition of the above polar comonomers. The steric effect of TMS was considered not large enough to protect the catalytic active species against the coordination of these polar comonomers. However, protection with the triisopropyl (TIPS) group, which has a higher steric effect, retained good catalyst activity and comonomer content. Therefore, the steric effect of the protecting group strongly influences the polymerization activity and effective comonomer incorporation. The copolymers of ethylene and 5-norbornene-2-methyleneoxytriisopropylsilane, symthesized by iPr(Cp)(Ind)ZrCl2 and the supported iPr(Cp)(Ind)ZrCl2 /MAO/SiO2 , showed two melting points. This can be attributed to the bimodal molecular weight distribution of the resulting copolymers. Novak and Tanaka142 copolymerized methacrylates with ethylene using metallocenes by masking the functional group of the acrylate through preparing nonenolizable forms of this comonomer. This means that the comonomer was prevented in this way from forming enol or enolate intermediates which are of lower energy and are incapable of inserting olefins. This was achieved by converting the acrylate to the corresponding carboxylate salt, which was complexed with Ti(III) as shown in Figure 8.4. Ethylene was copolymerized at room temperature with acrylic acid-titanocene complex and with methacrylic acid-titanocene complex (0.5–37%) in the presence of [Cp2 TiMe]+[MeB(C6 F5 )3 ]− to give titanocene-complexed products (55–98% yield relative to ethylene homopolymerization rated as 100%). The acrylic monomers did not decrease the catalyst activity. The titanocene protective groups were removed by washing with aqueous acid to synthesize copolymers with carboxyl groups.142 See Scheme 8.4. Kesti et al.131 surveyed the copolymerization of olefins with nitrogen-containing polar groups using zirconocenes. Tert-amine-functionalized olefins can be copolymerized in the

Synthesis of Functional Polyolefins using Metallocenes

217

Figure 8.4. Structures of the metallocene–acrylate complexes.142

presence of steric hindrance around the nitrogen atom. As in the case of alcohols,134 the amine can be protected by trimethylsilyl groups.142 Schneider et al.143 copolymerized ethylene with N,N-bis(trimethylsilyl)-1-amino-10undecene using Me2 Si(Benz[e]Ind)2 ZrCl2 which, upon hydrolysis of the silylated amines, synthesized short-chain branched linear low-density polyethylene (LLDPE) having pendent aminoalkyl groups. The incorporation of 1-amino-10-undecene varied between 6 and 19 wt%, which significantly influenced the properties of the resulting LLDPEs. Bis-silylation overcame poisoning of the catalyst because bis-silylated amines are much weaker Lewis bases. The effect of substituents on amines on the copolymerization activity showed that the comonomer where the nitrogen atom was surrounded by the most bulky groups such as tBu and Bz less deactivated the catalysts. However, the comonomer incorporation dropped.99 Dong et al.29 copolymerized propylene with styrene derivatives carrying Cl, OH and NH2 functional groups using Me2 Si(2-Me-4-Ph-Ind)2 ZrCl2 in the presence of hydrogen as a chain transfer agent. The polar OH and NH2 groups on styrene were passivated by trialkyl silane groups before copolymerization. The silane group effectively protected –OH and –NH2 functional groups and was easily deprotected by washing with an acid.

Scheme 8.4. Ethylene/acrylate copolymerization using metallocene catalyst complex.142

218

M. Atiqullah et al.

9. Conclusions and Recommendations From the application viewpoint, the catalytic activity and the product properties (average molecular weight, polydispersity index (PDI), and level of comonomer incorporation) are important determinants. In this conclusion, we summarize the advantages and limitations of metallocenes to synthesize functional polyolefins. The advantages are as follows. Metallocenes, depending on their types and structures, as well as those of the comonomer, can synthesize products having varying molecular weights, with 2.0 < PDI < 3.0. The molecular weights can range from macromeric/oligomeric to high values. The macromeric/oligomeric products are potentially suitable for the development of additives whereas the high molecular weight products may add to the application profile of commodity polyolefins, including the synthesis of compatibilizers. On the other hand, the limitations concern significant decrease in the catalyst activity (much below that of homopolymerization); low comonomer incorporation (up to 15 mol%); requirement of at least an equimolar quantity of an Al alkyl to pretreat an OH-containing comonomer, which make the process less economical and pose the problem of Al removal and other residues from the polymer; widespread implementation in existing processes that require heterogeneous (immobilized) catalysts, etc. To overcome these limitations, research continues in parallel using the less oxophilic late-transition metal catalysts. However, comprehensive work, particularly with respect to a wide range of comonomers, remains to be done in this area as well. Finally, we observe the lack of adequate research in the following areas—synthesis of easily soluble functional cooligomer/copolymer; products with a uniform distribution of the comonomer; establishment of relation among catalyst structure, and the various steps of copolymerization (initiation, propagation, and chain termination); degradation and stabilization study of the functional copolymer and correlation of the same with the metallocene structure; and application of supported metallocenes to synthesize the resulting polymers. Therefore, we recommend that the future research on this subject be directed toward these areas, as well as to the minimization of multi-step synthesis, and the development of MAO cocatalyst formulation to improve the activity and comonomer incorporation.

Acknowledgement The authors thankfully appreciate the support provided by the Research Institute and the Center of Research Excellence in Petroleum Refining & Petrochemicals (CoRE-PRP, established by the Saudi Ministry of Higher Education), King Fahd Univeristy of Petroleum & Minerals, Dhahran, Saudi Arabia. The financial support provided by Ciba Plastic Additive Segment at Basel, Switzerland, under Project Number CRP 2219, is gratefully acknowledged. The technical assistance of Mr. Anwar Hossaen is appreciated.

Glossary of Symbols AA Al:Zr ATRP 9-BBN B(C6 F5 )3 B(C6 F5 )4

Allyanisole Aluminum to zirconium ratio Atom transfer radical polymerization 9-borabicyclononane Tri(pentafluorophenyl)borane Tetra(pentafluorophenyl)borane

Synthesis of Functional Polyolefins using Metallocenes BDEM−C BDEM−O BHMCZ BR2− O∗ (nBuCp)2 ZrCl2 s

Bu-Li Bu-SiH3 −CH=CHCH2 -Ph (SiMe2 )(C5 Me4 )(NtBu) (SiMe2 )(C5 Me4 )(NtBu)TiCl2

C6 H5 CH2 SiH3 −CH2 -CH3 CH2 Cl2 [(SiMe2 )(C5 Me4 )(NtBu)TiMe]+[MeB(C6 F5 )3 ]−

13

C NMR C-O∗ COC C-O-O-BR2 C-O-O-B Cp∗ 2 ZrCl2 Cp∗ 2 ZrMe2 [Cp2 ∗ ZrMe]+[B(C6 F5 )4 ]−

[Cp2 ∗ ZrMe]+[MeB(C6 F5 )3 ]−

[Cp2 ∗ ZrMe]+X− Cp∗ TiMe3 Cp2 ZrCl2

219

Bond dissociation energy of metalcarbon bond Bond dissociation energy of metaloxygen bond 9-(bicycle[2,2,1]hept-5-en-2ylmethyl)- 9H-carbazole Borinate radical Bis(n-butylcyclopentadienyl) zirconium dichloride Secondary butyl lithium Butylsilane Phenyl propylene Dimethylsilyl tetra methyl cyclopentadienyl-tertiary butyl ammonium Dimethylsilyl tetra methyl cyclopentadienyl-tertiary butyl ammonium titanium dichloride Benzyl silane Ethyl end group Dichloromethane Ion pair of dimethylsilyl tetra methyl cyclopentadienyl (tertiary butyl ammonium) methyl titanium cation and methyl tri(pentafluorophenyl) borate anion Carbon-13 nuclear magnetic resonance Alkoxy radical Cycloolefin copolymers Peroxyl borane group Peroxy borane Bis(pentamethyl cyclopentdienyl) zirconium dichloride Bis(pentamethyl cyclopentdienyl) dimethyl zirconium Ion pair of bis(pentamethylcyclopentadienyl) methyl zirconium cation and tetra (pentafluorophenyl) borate anion Ion pair of bis(pentamethylcyclopentadienyl) methyl zirconium cation and methyl tri (pentafluorophenyl) borate anion Ion pair of bis(pentamethylcyclopentadienyl) methyl zirconium cation and an anion Pentamethyl cyclopentdienyl trimethyl titanium Bis(cyclopentdienyl) zirconium dichloride

220 DSC 1,4-DVB EP-DVB EO-DVB Et(Ind)2 ZrCl2 Et(Ind)2 ZrMe2 Et(Ind)(Flu)ZrCl2 Et(H4 Ind)2 ZrCl2 Et(H4 Ind)2 ZrMe2 Et2 AlCl EVNB FeC FT-IR GC-MS H2 O2 HB(Mes)2 HB(Trip)2 11 B NMR 1 H NMR [HNPhMe2 ]+[B(C6 F5 )4 ]− [HNPhMe2 ]+[BPh4 ]− HOMO Ind2 ZrCl2 [(H4 Ind)2 ZrMe]+X− [(Ind)2 ZrMe]+[MeB(C6 F5 )3 ]−

IPDMS Pr(3-Me-Cp)(Flu)ZrCl2

i

i

Pr(Cp)(Ind)ZrCl2

M. Atiqullah et al. Differential scanning calorimeter 1,4-Divinyl benzene Ethylene propylene and 1,4-divinylbenzene Ethylene 1-octene and 1,4-divinylbenzene Ethylene bis(indenyl) zirconium dichloride Ethylene bis(indenyl) dimethyl zirconium Ethylene (indenyl) (fluorenyl) zirconium dichloride Ethylene bis(tetrahydroindenyl) zirconium dichloride Ethylene bis(tetrahydroindenyl) dimethyl zirconium Diethyl aluminum chloride Ethyl-5-vinyl-2-norbornene copolymer Friedel Crafts Fourier transform infrared Gas chromatograph with mass spectrometer Hydrogen peroxide Dimesityl borane Bis(2,4,6-triisopropylphenyl) borane Boron 11 nuclear magnetic resonance Proton nuclear magnetic resonance Ion pair of dimethyl anilinium cation and tetra (pentafluorophenyl) borate anion Ion pair of dimethyl anilinium cation and tetra phenyl borate anion Highest occupied molecular orbital Bis(indenyl) zirconium dichloride Ion pair of bis(tetrahydroindenyl) methyl zirconium cation and any anion Ion pair of bis(indenyl) methyl zirconium cation and methyl tri (pentafluorophenyl) borate anion Isopropyldimethyl silane Isopropylidine (3-methylcyclopentadienyl) (fluorenyl) zirconium dichloride Isopropylidine (cyclopentadienyl) (indenyl) zirconium dichloride

Synthesis of Functional Polyolefins using Metallocenes i

Pr(tBuCp)(Flu)ZrMe2

isoBu2 AlH kp ktr LLDPE LnR(C5 Me5 )2 LUMO MA MAO Me [MeB(C6 F5 )3 ]− (2-MeInd)2 ZrCl2 (2-MeBenzoInd)2 ZrCl2 (Me5 C5 )2 ZrMe2 Me2 Si(Cp)(NtBu)TiCl2

Me2 Si(Cp)2 ZrCl2 Me2 Si(2-Me-4,5-benzoInd)2 ZrCl2

Me2 Si(2-Me-4-Ph-Ind)2 ZrCl2 Me2 Si(Ind)2 ZrCl2 Me2 Si(2-MeInd)2 ZrCl2 Me2 Si(C5 H3– 3-Me3 Si)2 SmH(THF)2

Me2 Si(C9 H6 )2 ZrCl2 Me2 Si(Cp∗ )(NtBu)TiMe2

Me2 Si(Flu)2 ZrMe2 Me2 Si(Ind)2 ZrCl2

221

Isopropylidine (tertiary butyl cyclopentadienyl) (fluorenyl) dimethyl zirconium Di-isobutylaluminum hydride Rate of propagation constant Rate of termination constant Linear low-density polyethylene Bis(pentamethylcyclopentadienyl) alkyl lanthanum Lowest unoccupied molecular orbital Maleic anhydride Methyl aluminoxane Methyl Methyl tri(pentafluorophenyl)borate anion Bis(2 methylindenyl)zirconium dichloride Bis(2-methylbenzoindenyl) zirconium dichloride Bis(pentamethylcyclopentadienyl) dimethyl zirconium Dimethylsilylene (cyclopentadienyl) amido tertiary butyl titanium dichloride Dimethyl silylene bis(cyclopentadienyl) zirconium dichloride Dimethyl silylene bis(2-methyl-4,5benzoannelated indenyl) zirconium dichloride Dimethyl silylene bis(2-methyl-4phenyl indenyl) zirconium dichloride Dimethyl silylene bis(indenyl) zirconium dichloride Dimethyl silylene bis(2methylindenyl) zirconium dichloride Dimethylsilylene bis(3-trimethylsilyl cyclopentadienyl) samarium hydride adduct with tetrahydrofuran Dimethylsilylene bis(indenyl) zirconium dichloride Dimethylsilylene (pentamethyl cyclopentadienyl) (tertiary butyl ammonium) dimethyl titanium Dimethylsilylene bis(fluorenyl) zirconium dichloride Dimethylsilylene bis(indenyl) zirconium dichloride

222

M. Atiqullah et al.

Me2 Si(H4 Ind)2 ZrCl2 Me2 Si(Me4 C5 )(NtBu)Ti(1,3-pentadiene) Me2 Si(Me4 Cp)(NtBu)TiCl2

(Me5 C5 )2 LnH Me2 Si(Me4 C5 )2 LnH MgCl2 MMA Mn Mw MWD NaOH NBD n BuSiH3 NH2 NMR −OCH3 PDI PE-b-PMMA PE-g-MMA PE-p-MeSt PE-t-B PE-t-OH PhCH2 SiH3 Ph2 C(Flu)(Cp)ZrCl2 Ph2 C(Ind)(Cp)ZrCl2 Ph3 C (PhF5 )3 B PhSiH3 PMMA p-MeSt PP u-PP PP-b-MA PP-B-MA PP-b-MMA

Dimethylsilylene bis(tetrahydroindenyl) zirconium dichloride Dimethylsilylene (tetramethyl cyclopentadienyl) (tertiary butyl ammonium) 1,3-pentadiene titanium Dimethylsilylene (tetramethyl cyclopentadienyl) (tertiary butyl ammonium) titanium dichloride Bis (pentamethyl cyclopentadienyl) lanthanium dihydride Dimethylsilylene bis(tetramethyl cyclopentadienyl) lanthanum dihydride Magnesium dichloride Methylmethacrylate Number average molecular weight Weight average molecular weight Molecular weight distribution Sodium hydroxide 2,5-Norbornadiene n-butylsilane Amine group Nuclear magnetic resonance Anisole methoxy Poly dispersity index Polyethylene block polymethylmethacrylate Polyethylene graft methylmethacrylate Polyethylene para methylstyrene copolymer Borane terminated polyethylene Hydroxyl terminated polyethylene Benzylsilane Diphenyl methyl (fluorenyl) (cyclopentadienyl) zirconium dichloride Diphenyl methyl (indenyl) (cyclopentadienyl) zirconium dichloride Triphenyl carbon Tri(pentafluorophenyl)borate Phenyl silane Polymethylmethacrylate Para methylstyrene Polypropylene Chain end unsaturated polypropylene Polypropylene block maleicanhydride Borane terminated propylenemaleicanhydride copolymer Polypropylene block methylmethacrylate

Synthesis of Functional Polyolefins using Metallocenes PP-b-StMA PP-g-MMA PPM PP-t-MA PP-t-p-MeSt PP-t-St-NH2 i-Pr(Cp)(Ind)ZrCl2 Pst-b-PMMA PSt-t-B PSt-t-OH PVA RAFT R-COCl R-COOH ROH R-Ox SiO2 SMA [SmH(C5 Me5 )2 ]2 SmMe(C5 Me5 )2 -THF

TBDMS TDMS TES Tg THF TIBA TiCl3 TiCl4 TIPS Tm TMA TMEDA UV VNB Zr-H Zr-C –ϕ-CH3

223

Polypropylene block styrene maleicanhydride Polypropylene graft methylmethacrylate Polypropylene macromer Maleicanhydride terminated polypropylene Paramethylstyrene terminated polypropylene Amine styrene terminated polypropylene Isopropylene cyclopentadienyl indenyl zirconium dichloride Polstyrene block polymethylmethacryle Borane-terminated polystyrene Hydroxyl-terminated polystyrene Polyvinyl alcohol Reversible addition fragmentation chain transfer Alkyl acid chloride Carboxylic acid Alcohol Alkyl-1,3-oxazoline Silica Styrene maleicanhydride Bis(bis(pentamethylcyclopentadienyl) samarium hydride) Bis(pentamethylcyclopentadienyl) methyl samarium adduct with tetrahydrofuran Tertiarybutyldimethylsilane Thixyldimethylsilane Triethylsilane Glass transition temperature Tetrahydrofuran Triisobutyl aluminum Titanium trichloride Titanium tetrachloride Triisopropylsilane Melt temperature Trimethyl aluminum N,N,N ,N -tetramethyl ethylene diamine Ultraviolet 5-vinyl-2-norbornene Zirconium hydrogen bond Zirconium− carbon bond Para methyl phenyl end group

224

M. Atiqullah et al.

References 1. Patil, A. O. “Functional polyolefins,” Chemical Innovation, 2000, 30,19–24. 2. Aggarwal, S. L. Block Copolymers; Plenum Press: New York, 1970. 3. Riess G.; Periard, J.; Bandereet, A. Colloidal and Morphological Behaviour of Block and Graft Copolymers; Plenum Press: New York, 1971. 4. Lohse D.; Datta, S.; Kresge, E. “Graft copolymer compatibilizers for blends of polypropylene and ethylene-propylene copolymers,” Macromolecules, 1991, 24, 561–566. 5. Chung, T. C.; Lu, H. L.; Janvikul, W. “Metallocene catalysts and borane reagents in the block/graft reactions of polyolefins”, Polym. Mat. Sci. Eng., 1995, 73,463–464. 6. Boffa, L. S.; Novak, B. M. “Copolymerization of polar monomers with olefins using transitionmetal complexes”, Chem. Rev., 2000, 100, 1479–1493. 7. Atiqullah, M. “Heterogeneous olefin polymerization metallocene catalysts: The application and role of Group 14 Lewis acid silica support modifiers”, In Progress in Heterogeneous Catalysis; Marmaduke, D. L., Ed.; Nova Science: New York, 2008. 8. Janiak, C. “Metallocene catalysts”, In Kirk-Othmer Encyclopedia of Chemical Technology; Seidel, A., Ed.; 2006, 16, 79–125. 9. Severn, J. R.; Chadwick, J. C.; Duchateau, R.; Friederichs, N. “Bound but not gagged”Immobilizing single-site α-olefin polymerization catalysts,” Chem. Rev., 2005, 105, 4073–4147. 10. Chen, E. Y. X.; Marks, T. J. “Cocatalysts for metal-catalyzed olefin polymerization: activators, activation processes, and structure-activity relationships,” Chem. Rev., 2000, 100, 1391– 1434. 11. Olabisi, O.; Atiqullah, M.; Kaminsky, W. “Group 4 metallocenes: supported and unsupported,” J. Macromol. Sci.-Rev Macromol. Chem. Phys., 1997, C37(3), 519–554. 12. Ribeiro, M. R.; Deffieux, A.; Portela, M. N. “Supported metallocene complexes for ethylene and propylene polymerizations: preparation and activity,” Ind. Eng. Chem. Res., 1997, 36, 1224–1237. 13. Sinn, H. J. “Proposals for structure and effect of methylaluminoxane based on mass balances and phase separation experiments,” Macromol. Symp. 1995, 97, 27–52. 14. Soares, J. B. P.; Hamielec, A. E. “Metallocene/aluminoxane catalysts for olefin polymerization. A review,” Polym. React. Eng., 1995, 3, 131–200. 15. Gupta, V. K.; Satish, S.; Bhardwaj, I. S. “Metallocene complexes of Group 4 elements in the polymerization of monoolefins,” J. Macromol. Sci.-Rev. Macromol. Chem. Phy., 1994, C34, 439–514. 16. Pellechia, C.; Mazzeo, M.; Grutter, G. J. “Synthesis of 4-phenylbutyl-branched polyethylene from an ethylene-styrene feed by a half-titanocene catalyst,” Makromol. Rapid Commun., 1999, 20, 337–340. 17. Pellechia, C.; Pappalardo, D.; Oliva, L.; Mazzeo, M.; Grutter, G. J. “Selective co-oligomerization of ethylene and styrene by half titanocene catalysts and synthesis of polyethylenes with 4-aryl1-butyl branches,” Macromolecules, 2000, 33, 2807–2814. 18. Chung, T. C.; “Lu, H. L. α-olefin-p-alkylstyrene copolymers, process and catalysts for their manufacture and functionalized copolymers thereof,” 1996, US Patent No. 5,543,484. 19. Chung, T. C.; Lu, H. L. “Synthesis of poly(ethylene-co-methylstyrene) copolymers by metallocene catalysts with constrained ligand geometry,” J. Polym. Sci., Part A: Polym. Chem., 1997, 35, 575–579. 20. Chung, T. C.; Lu, H. L. “Kinetic and microstructure studies of poly (ethylene-co-pmethylstyrene) copolymers prepared by metallocene catalysts with constrained ligand geometry,” J. Polym. Sci., Part A: Polym. Chem., 1998, 36, 1017–1029. 21. Chung, T. C.; Lu, H. L. “Functionalized α-olefin/para-alkylstyrene copolymers,” 1999, US Patent No. 5,866,659. 22. Chung, T. C. “Metallocene based reactive polyolefin copolymers containing p-methylstyrene,” In Metallocene based polyolefins, Scheirs, J.; Kaminsky, W, Eds.; 1999, Wiley: New York, 293–318.

Synthesis of Functional Polyolefins using Metallocenes

225

23. Chung, T. C.; Lu, H. L. “Functionalized α-olefin/para-alkyl styrene terpolymers,” 2000, US Patent No. 6,015,862. 24. Chung, T. C.; Dong J. Y. “A novel consecutive chain transfer reaction to p-methylstyrene and hydrogen during metallocene-mediated olefin polymerization,” J. Am. Chem. Soc., 2001, 123, 4871–4876. 25. Dong J. Y.; Chung, T. C. “Synthesis of polyethylene containing a terminal p-methylstyrene group: metallocene-mediated ethylene polymerization with a consecutive chain transfer reaction to p-methylstyrene and hydrogen,” Macromolecules, 2002, 35, 1622–1631. 26. Lu, B.; Chung, T. C. “Synthesis of PP graft copolymers via anionic living graft-from reactions of polypropylene containing reactive p-methylstyrene units,” J. Pol. Sci.: Part-A Polym. Chem., 1999, 37, 4176–4183. 27. Chung, T. C., Dong, J. Y. (The Penn State Research Foundation). “Polyolefin containing a terminal phenyl or substituted phenyl group and process for preparing the same,” Nov. 2002, US Patent No: 6,479,600. 28. Chung, T.-C.; Dong J. Y. (The Penn State Research Foundation). “Polyolefin containing a terminal phenyl or substituted phenyl group and process for preparing the same,” Sept. 2002, US Patent No: 20,020,132,943. 29. Dong, J. Y.; Wang, Z. M.; Hong, H.; Chung, T. C. “Synthesis of isotactic polypropylene containing a terminal Cl, OH or NH2 group via metallocene mediated polymerization/chain transfer reaction,” Macromolecules, 2002, 35, 9352–9359. 30. Chung, T.-C.; Dong, J. Y. (The Penn State Research Foundation). “Polyolefin graft copolymers derived from linear copolymers of alpha-olefins and divinylbenzene having narrow molecular weight and composition distributions and process for preparing the same,” July 2001, US Patent No: 6,265,493. 31. Chung, T. C.; Dong, J. Y. (The Penn State Research Foundation). “Linear copolymers of alphaolefins and divinylbenzene having narrow molecular weight and composition distributions, graft copolymers derived therefrom, and process for preparing the same,” July 2002, US Patent No: 6,414,102. 32. Chung T. C.; Dong, J. Y. “Synthesis of linear ethylene/divenylbenzene copolymers by metallocene catalysis,” Macromolecules, 2002, 35, 2868–2870. 33. Dong, J. Y.; Chung, T. C.; Wang, H. C.; Datta, S. “Synthesis of linear polyolefin elastomers containing divinylbenzene units and applications in crosslinking, functionalization and graft reactions,” Macromolecules, 2003, 36, 6000–6009. 34. Byun, Doo-Jin; Shin, Dong-Keun; Kim, S. Y. “Copolymerization of ethylene with allylbenzene using rac-ethylenebis(indenyl)zirconium dichloride/methylaluminoxane as a catalyst,” Macromol. Rapid Commun., 1999, 20, 419–422. 35. Byun, Doo-Jin; Kim, S. Y. “Selective chain transfer reactions in metallocene catalyzed copolymerization of ethylene with allylbenzene,” Macromolecules, 2000, 33, 1921–1923. 36. Resconi, L.; Piemontesi, F.; Franciscono, G.; Abis, L.; Florani, T. “Olefin polymerization at bis(pentamethylcyclopentadienyl)zirconium and –hafnium centers: Chain transfer mechanism.” J. Am. Chem. Soc. 1992, 114, 1025–1032. 37. Weng, W.; Markel, E. J.; Dekmezian, A. H. “Synthesis of vinyl-terminated isotactic poly(propylene),” Macromol. Rapid Commun. 2000, 21, 1103–1107. 38. Zheng, Yi; Li, Pan; Yan-Guo Li; Yue-Sheng, Li “Synthesis and characterisation of novel functional polyolefin containing sulfonic acid groups,” Eur. Polym. J., 2008, 44, 475–482. 39. Zheng, Yi; Yanguo, Li; Li, Pan; Yuesheng, Li “Synthesis of novel functional polyolefin containing carboxylic acid via Friedel-Crafts acylation reaction,” Polymer, 2007, 48, 2496–2502. 40. Byun, D. J.; Choi, K. Y.; Kim, S. Y. “Synthesis of phenol group-containing polyethylenes via metallocene-catalyzed ethylene-allylanisole copolymerization,” Macromol. Chem. Phy., 2001, 202, 992–997. 41. Atiqullah, M.; Akhtar, M. N.; Tinkl, M.; Ahmed, N. “The role of allylanisole in metallocenecatalyzed propylene polymerization and synthesis of end-capped oligomers,” Macromol. React. Eng., 2008, 2, 334–338.

226

M. Atiqullah et al.

42. Spaleck, W.; Kuber, F.; Winter, A.; Rohrmann, J.; Bachmann, B.; Antberg, M.; Dolle, V.; Paulus, E. F. “The influence of aromatic substituents on the polymerization behavior of bridged zirconocene catalysts,” Organometallics, 1994, 13, 954–963. 43. Kawaoka, A. M.; Marks T. J. “Organolanthanide-catalyzed synthesis of phosphine-terminated polyethylenes,” J. Am. Chem. Soc., 2004, 126, 12764–12765. 44. Kawaoka, A. M.; Marks, T. J. “Organolanthanide-catalyzed synthesis of phosphine-terminated polyethylenes. Scope and mechanism“, J. Am. Chem. Soc., 2005, 127, 6311–6324. 45. Chung, T. C. “Functional polyolefins via borane monomers,” Chemtech, 1991, 27, 496–499. 46. Chung, T. C. “Synthesis of functional polymers via borane monomers and metathesis catalysts,” J. Mol. Catal., 1992, 76, 15–31. 47. Chung, T. C. “Functionalized polyolefin prepared by the combination of borane monomers and transition metal catalysts,” Macromol. Symp., 1995, 89,151–162. 48. Chung, T. C. “Recent developments in the functionalization of polyolefin utilizing boranecontaining copolymers,” Trends in Polym. Sci., 1995, 3, 191–198. 49. Chung, T. C. “Recent Developments in the functionalization of polyolefins utilizing boranecontaining copolymers,” Trends in Polym. Sci., 1995, 3, 191–198. 50. Chung, T. C. “Synthesis of functional polyolefin copolymers with graft and block structures,” Prog. Polym. Sci., 2002, 27, 39–85. 51. Chung T. C.; Rhubright, D. “Synthesis of functionalized polypropylene,” Macromolecules, 1991, 24, 970–972. 52. Chung T. C.; Rhubright, D. “Kinetic aspects of the copolymerization between α-olefins and borane monomers in Ziegler-Natta catalyst,” Macromolecules, 1993, 26, 3019–3025. 53. Chung T. C.; Rhubright, D. “Polypropylene-graft-polycaprolactone: Synthesis,” Macromolecules, 1994, 27, 1313–1319. 54. Chung T. C., Jiang, G. J. “Synthesis of poly(1-octene-g-methyle methacrylate) copolymers. Synthesis of poly(1-octene-g-methyl methacrylate)copolymers,” Macromolecules, 1992, 25, 4816–4818. 55. Chung T. C.; Jiang, G. J.; Rhubright, D. “Olefin graft copolymers prepared using boranecontaining olefin backbone polymer,” 1994, US Patent No. 5,286,800. 56. Chung T. C.; Jiang, G. J.; Rhubright, D. “Olefin graft copolymers prepared using boranecontaining olefin backbone polymer,” 1995, US Patent No. 5,401,805. 57. Chung T. C.; Rhubright, D.; Jiang, G. J. “Synthesis of polypropylene-graft-poly(methyle methacrylate) copolymers by the borane approach,” Macromolecules, 1993, 26, 3467– 3471. 58. Chung, T. C. “Synthesis of polyalcohols via Ziegler-Natta polymerization,” Macromolecules, 1988, 21, 865–869. 59. Chung, T. C.; Lu, H. L. “Functionalization and block reactions of polyolefins using metallocene catalysts and borane reagents,” J. Mol. Catal. A: Chem., 1997 115, 115–127. 60. Ramakrishnan S.; Berluche, E.; Chung, T. C. “Functional group-containing copolymers prepared by Ziegler-Natta process,” Macromolecules, 1990, 23, 378–382. 61. Chung, T. C.; Lu, H. L.; Ding, R. D. “Synthesis of polyethylene-g-polystyrene and polyethyleneg-poly(p-methylstyrene) graft copolymers,” Macromolecules, 1997, 30, 1272–1278. 62. Chung, T-C.; Xu G. (The Penn State Research Foundation). “Process for preparing polyolefin diblock copolymers involving borane chain transfer reaction in transition metal-mediated olefin polymerization,” June 2001, US Patent No: 6,248,837. 63. Chung, T. C.; Janvikul, W.; Lu, H. L. “A novel stable radical initiator based on the oxidation adducts of alkyl-9-BBN, J. Am. Chem. Soc., 1996, 118, 705–706. 64. Chung, T. C.; Janvikul, W. “Borane-containing polyolefins: synthesis and applications,” J. Organomet. Chem., 1999, 581, 176–187. 65. Chung, T. C.; Lu, H. L.; Janvikul, W. “A novel synthesis of PP-b-PMMA copolymers via metallocene catalysis and borane chemistry,” Polymer, 1997, 38, 1495–1502. 66. Lu, B.; Chung, T. C. “New maleic anhydride PP copolymer with block structure: Synthesis and application in PP/polyamide reactive blends,” Macromolecules, 1999, 32, 2525–2533.

Synthesis of Functional Polyolefins using Metallocenes

227

67. Szwarc, M. “Living polymers. Their discovery, characterization, and properties,” J. Pol. Sci., Part A: Polym. Chem., 1998, 36, ix–xv. 68. Young, R. N.; Quirk, R. P.; Fetters, L. J. “Anionic polymerizations of nonpolar monomers involving lithium,” Advances in Polym. Sci., 1984, 56, 1–90. 69. Miyamoto M.; Sawamoto, M.; Higashimura, T. “Synthesis of telechelic living poly(vinyl ethers),” Macromolecules, 1985, 18, 123–127. 70. Kenedy J. P. “Thermoplastic elastomers of isobutylene and process of preparation,” 1990, US Patent No. 4,946,899. 71. Risse W.; Grubbs, R. H. “A novel route to block copolymers by changing the mechanism from living ring-opening metathesis polymerization of cyclic olefins to aldol condensation polymerization of silyl vinyl ethers,” Macromolecules, 1989, 22, 1558–1562. 72. Schrock R. R.; Yap K. B.; Yang D. C.; Sitzmann H.; Sita L.R.; Bazan, G. C. “Evaluation of cyclopenten-based chain-transfer agents for living ring opening metathesis polymerization,” Macromolecules, 1989, 22, 3191–3200. 73. Matyjaszewski, K.; Xia, J. “Atom transfer radical polymerization,” Chem. Rev., 2001, 101, 2921–2990. 74. Hawker, C.J.; Bosman, A.W.; Hart, E. “New polymer synthesis by nitroxide mediated living radical polymerizations,” Chem. Rev., 2001, 101, 3661–3688. 75. Hadjichristidis, N.; Iatrou H.; Pitsikalis, M.; Mays, J. “Macromolecular architectures by living and controlled/living polymerizations,” Prog. Polym. Sci., 2006, 31, 1068–1132. 76. Tsarevsky, N.V.; Matyiaszewski, K. “Green” atom transfer radical polymerization: from process design to preparation of well-defined environmentally friendly polymeric materials”, Chem. Rev., 2007, 107, 2270–2299. 77. Moad, G.; Rizzardo, E.; Thang, S.H. “Radical addition fragmentation chemistry in polymer synthesis”, Polymer, 2008, 49, 1079–1131. 78. Zetterlund P.B.; Kagawa Y.; Okubo M. “Controlled/Living polymerization in dispersed systems,” Chem Rev., 2008, 108, 3747–3794. 79. Cunningham M. F. “Controlled/living radical polymerization in aqueous dispersed systems,” Prog. Polym. Sci., 2008, 33, 365–398. 80. Debuigne, A.; Poli, R.; Jerome, C.; Jerome, R.; Detrembleur, C. Overview of cobalt-mediated radical polymerization: Roots, state of the art and future prospects. Prog. Polym. Sci., 2009, 34, 211–239. 81. Yasuda, H. “Rare earth metal intiated polymerization of (meth) acrylates and block copolymerization of olefins with polar monomers,” J. Polym. Sci., Part A: Polym. Chem., 2001, 39, 1955– 1959. 82. Yasuda, H.; Yamamoto, H.; Yokota, K.; Nakmura, S. “Synthesis of monodispersed high molecular weight polymers and isolation of an organolanthanide(III) intermediate coordinated by a penultimate poly(MMA) unit,” J. Am. Chem. Soc., 1992, 114, 4908–4910. 83. Xu, G.; Chung, T. C. “Borane chain transfer agent in metallocene-mediated olefin polymerization. Synthesis of borane terminated polyethylene and diblock copolymers containing polyethylene and polar monomer,” J. Am. Chem. Soc., 1999, 121, 6763–6764. 84. Xu, G.; Chung, T. C. “Synthesis of syndiotactic pPolystyrene (s-PS) containing a terminal polar group and diblock copolymers containing s-PS and polar polymers,” Macromolecules, 1999, 32, 8689–8692. 85. Chung, T. C.; Yingying, G.; Xu. L.; Youliang, H. “Metallocene mediated olefin polymerization with B-H chain transfer agents: Synthesis of chain end functionalized polyolefins and diblock copolymers,” Macromolecules, 2001, 34, 8040–8050. 86. Shiono, T.; Yoshihide, A.; Soga, K. “Synthesis of isotactic polypropylene-block-poly(methyl methacrylate) using magnesium bromide-termainted isotactic polypropylene,” Macromolecules, 1994, 27, 6229–6231. 87. Shiono T.; Soga, K. “Synthesis of terminally aluminum-functionalized polypropylene,” Macromolecules, 1992, 25, 3356–3361.

228

M. Atiqullah et al.

88. Yasuda, H.; Furo, M.; Yamamoto, H. “New Aapproach to block copolymerization of ethylene with alkyl methacrylates and lactones by unique catalysis with organolanthanide complexes,” Macromolecules, 1992, 25, 5115–5116. 89. Yasuda H.; Yamamoto, H.; Yamashita, M.; Yokota, K.; Nakmura, S. “Synthesis of high molecular weight poly(methyl methacrylate) with extremely low polydispersity by the unique function of organolanthanide(III) complexes,” Macromolecules, 1993, 26, 7134–7143. 90. Jeske, G.; Lauke, H.; Mauermann, P. N.; Sepston, H.; Schumann, T.; Marks, J. “Highly reactive organolanthanides. Systematic routes to and olefin chemistry of early and late bis(pentamethylcyclopentadienyl) 4f hydrocarbyl and hydride complexes,” J. Am. Chem. Soc., 1985, 107, 8091–8103. 91. Jeske, G.; Lauke, H.; Mauermann, H.; Schumann, H.; Marks T. J. “Highly reactive organolanthanides. A mechanistic study of catalytic olefin hydrogenation by bis(pentamethylcyclopentadienyl) and related 4f complexes,” J. Am. Chem. Soc., 1985, 107, 8111–8118. 92. Ballard, D. G. H.; Courtis, A.; Holton, J.; McMeeking, J.; Pearce, R. “Lanthanide catalysts for the polymerization of olefins,” Conf. Eur. Plast. Caoutch., [C. R.], 1978, 5(1), A4/1–A4/7. 93. Desurmont, G. et al. “New approach to block copolymerization of ethylene with polar monomers by unique catalytitic function of organolanthanide complexes,” J. Polym. Sci., Part A: Polym. Chem., 2000, 38, 4095–4109. 94. Desurmont G.; Tokimitsu, T.; Yasuda, H. “First controlled block copolymerization of higher 1-olefins with polar monomers using metallocene type single component lanthanide initiators,” Macromolecules, 2000, 33, 7679–7681. 95. Desurmont, G.; Li, Y.; Yasuda, H. “Reaction pathways for the formation of binuclear samarocene hydride from monomeric alkyl samarocene derivative and the effective catalysis of samarocene hydride for the block copolymerization of ethylene with polar monomers,” Organometallics, 2000, 19, 1811–1813. 96. Aaltonen, P.; L¨ofgren, B. “Synthesis of functional polyethylenes with soluble metallocene/methylaluminoxane catalyst,” Macromolecules, 1995, 28, 5353–5357. 97. Aaltonen, P.; Fink, G.; L¨ofgren, B.; Sepp¨al¨a, J. “Synthesis of hydroxyl group containing polyolefins with metallocene/methylaluminoxane catalysts,” Macromolecules, 1996, 29, 5255–5260. 98. Hakala, K.; L¨ofgren, B.; Helaja, T. “Copolymerizations of oxygen-functionalized olefins with propylene using metallocene/methylaluminoxane catalyst,” Eur. Polym. J., 1998, 34, 1093– 1097. 99. Sepp¨al¨a, J.; Loefgren, B.; Hakala, K.; Lipponen, S.; Helaja, T.; Anttila, U. “Functional copolymers by using metallocene catalysis enabling new applications,” Polym. Mat. Sci. and Eng., 2001, 84, 257–258. 100. Kaya, A.; Jakisch, L.; Komber, H.; Voigt, D.; Pionteck, J.; Voit, B.; Schulze, U. “Synthesis of various functional propylene copolymers using rac-Et[1-Ind]2 ZrCl2 /MAO as the catalyst system,” Macromol. Rapid Commun., 2001, 22, 972–977. 101. Wilen, C. E.; Nasman, J. H. “Polar activation in copolymerization of propylene and 6-t-butyl-2(1,1-dimethylhept-6-enyl)-4-methylphenol using Me2 Si(H4Ind)ZrCl2 /MAO catalyst system,” Macromolecules, 1994, 27, 4051–4057. 102. Busico, V.; Cipullo, R.; Cutillo, F.; Friedrichs N.; Rona, S.; Ronca, S.; Wong, B. “Improving the performance of methylalumoxane: A facile and efficient method to trap free trimethylaluminum,” J. Am. Chem. Soc. 2003, 125, 12402–12403. 103. Wilen, C. E.; Luttikhedde, H.; Hjertberg, T.; Nasman J. H. “Copolymerization of ethylene and 6-tert-butyl-2-(1,1-dimethylhept-6-enyl)-4-methylphenol over three different metallocenealuminoxane catalyst system,” Macromolecules, 1996, 29, 8569–8575. 104. Kaminsky, W.; Engehausen, R.; Zoumis, K.; Spaleck, W.; Rohrmann, J. “Standardized polymerizations of ethylene and propene with bridged and unbridged metallocene derivatives: A comparison,” Macromol. Chem., 1992, 193, 1643–1651.

Synthesis of Functional Polyolefins using Metallocenes

229

105. Wilen, C. E.; Auer, M.; Stranden, J.; N¨asmann, J. H.; Rotzinger, B.; Steinmann, A.; King, R.E.; Zweifel H.; Drewes, R. “Synthesis of novel hindered amine light stabilizers (HALS) and their copolymerization with ethylene or propylene over both soluble and supported metallocene catalyst systems,” Macromolecules, 2000, 33, 5011–5026. 106. Mullins, M. J.; Soto, J.; Nickias, P. N. (The Dow Chemical Company), “Incorporation of functionalized comonomers in polyolefins”, Dec., 2000, US Patent No: 6,166,161. 107. Lu, B.; Chung, T. C. “Maleic anhydride modified polypropylene with controllable molecular structure: New synthetic route via boron-terminated polypropylene,” Macromolecules, 1998, 31, 5943–5946. 108. Lu, H. L.; Hong, S.; Chung, T. C. “Synthesis of polypropylene-co-methylstyrene copolymers by metallocene and Ziegler-Natta catalysts“, J. Polym. Sci. Part-A: Polym. Chem., 1999, 37(15), 2795–2802. 109. Kaya, A.; Jakisch, L.; Komber, H.; Pompe, G.; Pionteck, J.; Voit, B.; Schulze, U. “Synthesis of oxazoline functionalized polypropene using metallocene catalysts,” Macromol. Rapid Commun., 2000, 21, 1267–1271. 110. Stehling, U. M.; Stein, K. M.; Kesti, M. R.; Waymouth, R. M. “Metallocene/borate catalyzed polymerization of amino-functionalized α-olefins,” Macromolecules, 1998, 31, 2019–2027. 111. Stehling, U. M.; Stein, K. M.; Fischer, D.; Waymouth, R. M. “Metallocene/Borate-Catalyzed Copolymerization of 5-N,N-Diisopropylamino-1-pentene with 1-Hexene or 4-Methyl-1pentene,” Macromolecules, 1999, 32, 14–20. 112. Stehling, U. M.; Malmstrom, E. E.; Waymouth, R. M.; Hawker, C. J. “Synthesis of poly(olefin) graft copolymers by a combination of metallocene and living free radical polymerization techniques,” Macromolecules, 1998, 31, 4396–4398. 113. Shiono, T.; Kurosawa, H.; Ishida, O.; Soga, K., “Synthesis of polypropylenes functionalized with secondary amino groups at the chain ends,” Macromolecules, 1993, 26, 2085–2089. 114. Shiono, T.; Kurosawa, H.; Soga, K. “Synthesis of isotactic polypropylene functionalized with a primary amino group at the initiation chain end,” Macromolecules, 1994, 27, 2635–2637. 115. Shiono, T.; Azad, S. M.; Ikeda, T. “Copolymerization of atactic polypropylene macromonomer with propene by an isospecific metallocene catalyst,” Macromolecules, 1999, 32, 5723–5727. 116. Hackman, M.; Repo, T.; Jany, G.; Rieger, B. “Zirconocene-MAO catalyzed homo- and copolymerizations of linear asymmetrically substituted dienes with propene: a novel strategy to functional (co)poly(α-olefin)s,” Macromol. Chem. Phy., 1998, 199, 1511–1517. 117. Mustonen, I.; Hukka, T.; Pakkanen, T. “Synthesis, characterization and polymerization of the novel carbozole-based monomer 9-(bicycle[2.2.1]hept-5-en-2-ylmethyl)-9H-carbazole,” Macromol. Rapid Commun., 2000, 21, 1286–1290. 118. Shiono, T.; Soga, K. “Syntheiss of terminally halogenated isotactic polypropylene)s using hydroalumination,” Makromol. Rapid Commun., 1992, 13, 371–376. 119. Bruzaud, S.; Cramail, H., Duvignac, L.; Deffieux, A. “ω-Chloro-α-olefins as co- and termonomers for the synthesis of functional polyolefins”, Macromol. Chem. Phy., 1997, 198, 291–303. 120. Fu, P. F.; Marks, T. J. “Silanes as transfer agents in metallocene-mediated olefin polymerization. Facile in-situ catalytic synthesis of silyl-terminated polyolefins,” J. Am. Chem. Soc., 1995, 117, 10747–10748. 121. Koo, K.; Fu, P. F.; Marks, T. J. “Organolanthanide mediated silanolytic chain transfer processes. Scope and mechanism of single reactor catalytic routes to silnaopolyolefins,” Macromolecules, 1999, 32, 981–988. 122. Marks, T. J.; Koo, K. (Northwestern University). “Silyl-terminated polymer and method for preparing silyl-terminated polyolefins,” June, 2000, US Patent No: 6,077,919. 123. Makio H.; Koo, K.; Marks, T. J. “Silanolytic chain transfer in olefin polymerization with supported single site ziegler-natta catalysts,” Macromolecules, 2001, 34, 4676–4679. 124. Arriola, D. J.; Bishop, M. T.; Campbell, Jr. R. E.; Devore, D. D.; Hahn, S. E.; Ho, T. H. T.; McKeand, J.; Timmers F. J. (The Dow Chemical Company), “Silane functionalized olefin interpolymer derivatives”, US Patent No: 6,624,254, Sept. 2003.

230

M. Atiqullah et al.

125. Marathe, S.; Sivaram, S. “Regioselective copolymerization of 5-vinyl-2-norbornene with ethylene using zirconocenes-methylaluminoxane catalysts: A facile route to functional polyolefins,” Macromolecules, 1994, 27, 1083–1086. 126. Radhakrishnan, K.; Sivaram, S. “Copolymerization of ethylene with 2,5-norbornadiene using a homogeneous metallocene/MAO catalyst system,” Macromol. Chem. Phys., 1999, 200, 858–862. 127. Monkkonene, K.; Pakkanen, T. T. “Synthesis and characterization of poly(ethylene-conorbornadiene),” Macromol. Chem. Phy., 1999, 200, 2623–2628. 128. Uozumi, T.; Miyazawa, K.; Sano, T.; Soga, K. “Alternating copolymerization of ethylene and 1-octene with meso-[dimethylsilylbis(2-methyl-1-indenyl)]zirconium dichloridemethylaluminoxane as catalyst system,” Macromol. Rapid Commun., 1997, 18, 883–889. 129. Uozumi, T.; Tian, G.; Ahn, C.-H.; Jin, J.; Tsubaki, S.; Sano, T.; Soga, K. “Synthesis of functionalized alternating olefin copolymer and modification to graft copolymer by hydrosilylation,” J. Polym. Sci. Part A: Polym. Chem., 2000, 38, 1844–1847. 130. Schiffino, R. S.; Crowther, D. J. (Exxon Mobil Chemicals Patents Incorporated). “Ethylene copolymerization process”, March, 2004, US Patent No: 6,713,574. 131. Kesti, M. R.; Coates, G. W.; Waymouth, R. M. “Homogeneous Ziegler-Natta polymerization of functionalization monomers catalyzed by cationic Group IV metallocenes,” J. Am. Chem. Soc., 1992, 114, 9679–9680. 132. Goretzki, R.; Fink, G. “Homogeneous and heterogeneous metallocene/MAO catalyzed polymerization of trailkylsilyl-protected alcohols,” Macromol. Rapid Commun., 1998, 19, 511–515. 133. Goretzki, R.; Fink. G. “Homogeneous and heterogeneous metallocene/MAO-catalyzed polymerization of functionalized olefins,” Macromol. Chem. Phy., 1999, 200, 881–886. 134. Hakala, K.; Helaja, T.; Lofgren, B.. (2000). “Metallocene/methyl aluminoxane-catalyzed copolymerizations of oxygen-functionalized long-chain olefins with ethylene,” J. Pol. Sci., Part A: Polym. Chem., 2000, 38, 1966–1971. 135. Radhakrishnan, K.; Sivaram, S. “Copolymerization of ethylene with bicyclo[2.2.1]hept-5-ene2-methanol: a facile route to hydroxyl functional polethylenes,” Macromol. Rapid Commun., 1998, 19, 581–584. 136. Marques, M. M.; Correia, S. G.; Ascenso, J. R.; Ribeiro, A. F. G.; Gomes, P. T.; Dias, A. R.; Foster, P.; Rausch, M. D.; Chien, J. C. W. “Polymerization with TMA-protected polar vinyl comonomers. I. Catalyzed by Group 4 metal complexes with d-diimine-type ligands,” J. Pol. Sci., Part A: Polym. Chem. 1999, 37(14), 2457–2469. 137. Alexander, R.; Fink, G. “Homogeneous metallocene/MAO catalyzed polymerizations of polar norbornene derivatives: copolymerization using ethene and terpolymerizations using ethene and norbornene,” Macromol. Chem. Phy., 2000, 201, 1365–1373. 138. Wendt, R. A.; Gerhard, F. “Homogeneous metallocene/MAO-catalyzed polymerizations of polar norbornene derivatives: copolymerizations using ethene, and terpolymerizations using ethene and norbornene,” Macromol. Chem. Phy., 2000, 201, 1365–1373. 139. Hagihara, H.; Murata, M.; Uozumi, T. “Alternating copolymerization of ethylene and 5-Hexen1-ol with [Ethylene(1-indenyl)(9-fluorenyl)]-zirconium dichloride/methylaluminoxane as the catalyst,” Macromol. Rapid Commun., 2001, 22, 353–357. 140. Fink, G.; Wendt, R. A.; Angermund, K.; Jensen, V. R.; Thiel, W. “2: copolymerizations using ethene, and terpolymerizations using ethene and norbornene,” Polym. Mat. Sci. Eng., 2001, 84, 255–256. 141. Wendt, R.; Angermund, A., K.; Jensen, V.; Thiel, W., Fink, G. “Ethene copolymerization with trialkylsilyl protected polar norbornene derivates,” Macromol. Chem. Phy., 2004, 205(3), 308–318. 142. Novak, B. M.; Tanaka, H. “The “if you can’t beat them, join them” approach to olefin-polar comonomer polymerization using metallocene catalysts”, Polym. Mat. Sci. Eng., 1999, 80, 45–55. 143. Schneider, M. J.; Schafer, R.; Mulhaupt, R. “Aminofunctional linear low density polyethylene via metallocene-catalysed ethene copolymerization with N,N-bis(trimethylsilyl)-1-amino-10undecene,” Polymer, 1997, 38, 2455–2459.

Polymer Reviews, 50:231–234, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.493625

Perspective Polymer Microscopy: Current Challenges 2 ¨ LAWRENCE F. DRUMMY1 AND CHRISTIAN KUBEL 1

Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright Patterson AFB, Ohio 2 Karlsruhe Institute of Technology, Institute of Nanotechnology, Eggenstein-Leopoldshafen, Germany

Keywords polymer, microscopy, atomic force microscopy, transmission electron microscopy

The growing demand for lightweight and multifunctional materials systems is pervasive across a multitude of technologies and applications. These systems invariably contain polymeric or organic components, which, through intricate synthetic control and processing techniques, are tailored for optimized properties by tuning the structure from the molecular level to the nano- and bulk scale. The need for an improved morphological understanding across length scales will grow as materials systems become increasingly complex and with that, the need for a sustained investment in education, research, and technology for morphology tools such as microscopy in the field of polymer science and engineering will grow.1 Microscopy has played a leading role in the development of several polymeric material classes such as semicrystalline polymers,2 block copolymers,3 polymer nanocomposites,4 and rigid rod polymers.5 Currently, polymers and macromolecules are used for an incredibly diverse set of materials applications ranging from structural to electrical, both in synthetic and natural materials systems. Microscopy is critical for understanding the structure of these materials, and although significant steps forward have been made in polymer microscopy, there are several major challenges that limit future prospects. This introduction article briefly outlines current challenging areas in polymer microscopy, and the articles in this special issue provide significant steps forward in addressing these challenges.

1. Challenge 1: Preserving Sample Structure The nature of all microscopic techniques is such that it is necessary to perturb the sample in order to characterize it. In scanning force microscopy, for example, the tip can damage the Received May 7, 2010; accepted May 11, 2010. Address correspondence to Lawrence Drummy, Materials and Manufacturing Directorate, Air Force Research Laboratory, 2941 Hobson Way, Wright Patterson AFB, OH 45433. E-mail: [email protected]

231

232

L. F. Drummy and C. K¨ubel

sample, effectively changing the material that is being imaged or characterized. The article by McConney et al.6 in this issue examines this effect in the context of characterizing the elastic modulus of polymeric materials using surface force spectroscopy (including force–distance curves, elastic modulus measurements, and adhesion measurements). Preparation of representative materials surfaces and cross sections or generation of ultrathin films from complex structures such as devices and bulk multicomponent polymeric materials is a major challenge. Preserving the original structure throughout the sample preparation chain can be as much of a limitation as resolution, sample stability, contrast, and the optics of the microscope itself. Focused ion beam (FIB) would allow us to study selected interfaces directly, examining the effects of constraints on polymers and structural changes at surfaces and interfaces (organic–inorganic), which would be a significant advance, although the effect that ion beam damage in the FIB has on the polymer structure is not known in detail.7 Cryomicroscopy is an invaluable tool for nanostructured liquids, hydrated materials, and other macromolecular-based materials not stable in vacuum. A wide range of soft materials systems have been investigated using this technique, with a wealth of morphological information extracted, and this will continue to expand as more laboratories get access to the equipment and hands-on experience with cryopreservation techniques. Current challenges in cryomicroscopy include distortion or collapse of some structures due to confinement in a thin, electron transparent liquid layer, generation of contrast from typically low-Z materials, as well as beam damage. The article by Zhong and Pochan8 in this issue discusses these challenges, as well as techniques such as cryo-TEM tomography and the multitude of polymeric architectures that have been imaged using cryo-TEM in recent years.

2. Challenge 2: Generating Interpretable Contrast In scanning probe microscopy,6 numerous signals can be extracted from a single material/experiment to obtain information about the mechanical, electrical, chemical, and magnetic properties. In electron microscopy, contrast generation from organics and light elements (without the use of staining) has undergone incremental improvements based on control of the electron energy9 and thus the electron–sample interaction and the image formation method: Defocus,10 Zernike phase-plate,11 energy filtered transmission electron microscopy (TEM),12 and Z-contrast high angle annular dark field transmission electron microscopy (HAADF-STEM).13 By increasing the scattering contrast between two phases, the necessary illumination intensity can be reduced while maintaining the same detectability. Thus, the total number of electrons that pass through the sample can be reduced, preserving the structure or allowing for higher electron optical magnification to be used.14 The most significant leaps forward in polymer electron microscopy may be a combination of a phase-plate with Cs -corrected optics. Other approaches to improve contrast while managing sample damage include improvements in detector efficiency, where a lower electron dose could be used to achieve a certain signal level in the image. Also, the actual critical dose of the polymer or the electron dose a certain material can withstand without significant detectable damage can actually be increased by certain methods. These include a reduction in temperature, changing the dose rate, varying the incident electron energy, averaging over many structures (single-particle analysis), and possibly using a pulsed electron beam15 in which the pulse time is shorter than the associated time scale for material damage.16 As is discussed in the articles by Libera and Egerton,17 Martin et al.,18 and Kolb et al.,19 generating interpretable contrast in organic materials is intimately coupled with beam damage challenges, and improvements

Polymer Microscopy: Current Challenges

233

in one or both can significantly increase the amount of directly interpretable microstructural information that can be extracted.

3. Challenge 3: Image Segmentation and Quantitative Microscopy Once detectable contrast between features is generated, for quantitative analysis in most cases those features must be segmented. Image segmentation is the key challenge currently limiting the accessibility of quantitative information from microscopy, and although materials scientists are beginning to use advanced techniques, most of the mathematics and algorithm development has come from the field of signal processing. Modern algorithms minimize the need for user supervision and can extract features from images without the loss of resolution or introduction of artifacts typically associated with traditional image processing. For example, anisotropic or stabilized inverse diffusion equations have been successfully used to detect boundaries in images while maintaining sharpness and definition there.20,21 The set of equations acts as an unstable inverse intensity diffusion near edges and as a stable linear diffusion in regions without edges. This has the effect of both noise removal and edge enhancement in the image. In addition, alternative noise reduction and segmentation techniques based on the watershed algorithm and curvature/energy methods are currently being actively explored. Nevertheless, for complex materials with varying composition and irregular features, automated image segmentation in two dimensions and three dimensions is still the main limiting factor for a full quantitative analysis. In summary, though microscopy has directly contributed to many of the most significant advances in polymer science and engineering, significant barriers remain for the development of the field. As research in general becomes more interdisciplinary, and polymeric materials become increasingly complex, development of microscopy tools will be critical for developing the next generation of polymeric materials.

References 1. Ober, C. K.; Cheng, S. Z. D.; Hammond, P. T.; Muthukumar, M.; Reichmanis, E.; Wooley, K. L.; Lodge, T. P. “Research in macromolecular science: Challenges and opportunities for the next decade,” Macromolecules, 2009, 42, 465–471. 2. Tsukruk, V. V.; Reneker, D. H. “Surface morphology of syndiotactic polypropylene singlecrystals observed by atomic force microscopy,” Macromolecules, 1995, 28, 1370–1376. 3. Spontak, R.; Fung, J. C.; Braunfeld, M. B.; Sedat, J. W.; Agard, D. A.; Ashraf, A.; Smith, S. D. “Architechture-induced phase immiscibility in a diblock/multiblock copolymer blend,” Macromolecules, 1996, 29, 2850–2856. 4. Drummy, L. F.; Wang, Y. C.; Schoenmakers, R.; May, K.; Jackson, M.; Koerner, H.; Farmer, B. L.; Maruyama, B.; Vaia, R. “Morphology of layered silicate- (NanoClay-) polymer nanocomposites by electron microscopy and small-angle X-ray scattering,” Macromolecules, 2008, 41, 2135–2143. 5. Martin, D. C.; Thomas, E. L. “Ultrastructure of poly(para-phenylenebenzobisoxazole) fibers,” Macromolecules, 1991, 24, 2450–2460. 6. McConney, M. E.; Singamaneni, S.; Tsukruk, V. V. “Probing soft matter with the atomic force microscope: force spectroscopy and beyond,” Polymer Reviews, 2010, 50(3), 235–286. 7. Brostow, W.; Gorman, B. P.; Olea-Mejia, O. “Focused ion beam milling and scanning electron microscopy characterization of polymer plus metal hybrids,” Materials Letters, 2007, 61, 1333–1336. 8. Zhong, S.; Pochan, D. “Cryogenic transmission electron microscopy for direct observation of polymer and small molecule materials and structures in solution,” Polymer Reviews, 2010, 50(3), 287–320.

234

L. F. Drummy and C. K¨ubel

9. Drummy, L. F.; Yang, J.; Martin, D. C. “Low-voltage electron microscopy of polymer and organic molecular thin films,” Ultramicroscopy, 2004, 99, 247–256. 10. Handlin, D. L.; Thomas, E. L. “Phase contrast imaging of styrene-isoprene and styrene-butadiene block copolymers,” Macromolecules, 1983, 16, 1514–1525. 11. Tosaka, M.; Danev, R.; Nagayama, K. “Application of phase contrast transmission microscopic methods to polymer materials,” Macromolecules, 2005, 38, 7884–7886. 12. Du Chesne, A. “Energy filtering transmission electron microscopy of polymers—benefit and limitations of the method,” Macromolecular Chemistry and Physics, 1999, 200, 1813–1830. 13. Loos, J.; Sourty, E.; Lu, K.; de With, G.; Bavel, S. V. “Imaging polymer systems with highangle annular dark field scanning transmission electron microscopy,” Macromolecules, 2009, 42, 2581–2586. 14. Yakovlev, S.; Libera, M. “Dose-limited spectroscopic imaging of soft materials by low-loss EELS in the scanning transmission electron microscope,” Micron, 2008, 39, 734–740. 15. Shorokhov, D.; Zewail, A. H. “4D Electron imaging: Principles and perspectives” Physical Chemistry Chemical Physics, 2008, 10, 2879–2893. 16. Grubb, D. T. “Radiation damage and electron microscopy of organic polymers,” Journal of Materials Science, 1974, 9, 1715–1736. 17. Libera, M.; Egerton, R. “Advances in the transmission electron microscopy of polymers,” Polymer Reviews, 2010, 50(3), 321–339. 18. Martin, D. C.; Wu, J.; Shaw, C. M.; King, Z.; Spanninga, S. A.; Richardson-Burns, S.; Hendricks, J.; Yang, J. “The morphology of poly(3,4-ethylenedioxythiophene),” Polymer Reviews, 2010, 50(3), 340–384. 19. Kolb, U.; Gorelik, T. E.; Mugnaioli, E.; Stewart, A. “Structural characterization of organics using manual and automated electron diffraction,” Polymer Reviews, 2010, 50(3), 385–409. 20. Dong, X.; Pollak, I. “Multiscale segmentation with vector-values nonlinear diffusions on arbitrary graphs,” IEEE Transactions on Image Processing, 2006, 15, 1993–2005. 21. Frangakis, A. Noise Reduction and Segmentation Techniques Developed for Multidimensional Electron Microscopy of Biological Specimens; Ph.D. thesis, Technical University Munich, 2001.

Polymer Reviews, 50:235–286, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.493255

Reviews Probing Soft Matter with the Atomic Force Microscopies: Imaging and Force Spectroscopy MICHAEL E. McCONNEY, SRIKANTH SINGAMANENI, AND VLADIMIR V. TSUKRUK School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia The development of atomic force microscopy has evolved into a wide variety of microscopy and characterization techniques well beyond conventional imaging. The focus of this review is on characterization methods based on the scanning probe and their application in characterizing physical properties of soft materials. This consideration is broken into three major categories focusing on mechanical, thermal, and electrical/magnetic properties in addition to a brief review of high-resolution imaging. Surface spectroscopy is discussed to great extent and consideration includes procedural information, common pitfalls, capabilities, and their practical application in characterizing soft matter. Key examples of the method are presented to communicate the capabilities and impact that probe-based characterization techniques have had on the mechanical, thermal, and electrical characterization of soft materials. Keywords atomic force microscopy, force spectroscopy, scanning thermal microscopy, kelvin probe force microscopy, scanning probe microscopy, polymers

1. Introduction to Atomic Force Microscopy Imaging The invention of scanning tunneling microscopy (STM) in early 1980 by Rohrer and Binnig at the IBM Zurich Laboratories led to a fast establishment of a new class of microscopy known as scanning probe microscopy (SPM) over the past three decades.1–6 Overcoming the limitations of STM in their application to nonconductive materials, atomic force microscopy (AFM) was introduced as a logical next step in SPM techniques, thereby greatly expanding the imaging and probing capabilities.7 The ongoing development of SPM and nanotechnology remain deeply intertwined and mutually augmented. SPM techniques have several common components, including an ultrasharp probe, sensing elements (Figure 1A), a piezo scanner tube, and a computer-controlled feedback

Received January 20, 2010; accepted April 21, 2010. Address correspondence to Vladimir V. Tsukruk, School of Materials Science and Engineering, Georgia Institute of Technology, 771 Ferst Dr., Atlanta, GA 30332-0245. E-mail: [email protected]

235

236

M. E. McConney et al.

Figure 1. (a) Schematic depicting the AFM tip interacting with the sample surface and the most common optical technique employed to detect the deflection of the microcantilever. (b) Interaction force–separation distance plot showing the long range attractive regime (noncontact) and short-range repulsive regime (contact). (c) Schematic showing the microcantilever interaction with the sample in the three basic imaging modes of operation of AFM.

loop. One key feature that set SPM-based techniques apart from other microscopy techniques is the use of ultrasharp probes. Apart from imaging the properties with nanoscale resolution, one of the important developments is the manipulation of matter on the surface using a scanning probe. Furthermore, as natural succession to their application as force transducers in AFM, microcantilevers are being extensively investigated as a new platform for transduction in sensing technology in chemical, biological, and thermal sensing.8–13 The unprecedented lateral and vertical resolution offered by SPM techniques enables the visualization of micro-, nano-, and molecular-scale structure of polymer surfaces and interfaces. Under special conditions, atomic resolution is even attainable with SPM.14 Other outstanding advantages of SPM include true three-dimensional (3D) topology, minimal sample preparation, and imaging under a wide variety of environments, including ambient conditions, fluidic conditions, gases, and under different temperatures. Various SPM techniques enable simultaneous probing of the different properties, such as structural, mechanical, electrical, thermal, or magnetic properties with nanoscale resolution. These microscopy methods continue to provide invaluable insight into the understanding structure–property relationship of these materials at nanoscale. SPM can be used to manipulate and pattern soft matter by applying normal and shearing forces and modifying surface topography by repeated scanning.15,16

Imaging and Force Spectroscopy

237

Based on cantilever dynamics, AFM operation can be generally divided into static and dynamic modes as shown in Figure 1C. Dynamic modes involve oscillating the cantilever, usually near its resonance frequency. Under dynamic modes, the resonant frequency, amplitude, and phase of the oscillation change due to the interaction between the tip and the sample. Dynamic modes can be carried out in several variations including amplitude modulation (AM-AFM) and/or frequency modulation (FM-AFM). The most common type of dynamic mode AFM, called tapping mode or intermittent contact, is a simple and robust amplitude-modulated AM-AFM technique. On the other hand, in the static mode, the tip is raster-scanned across the surface and the deflection of the cantilever is maintained constant by the feedback control. The readers are referred to several reviews for detailed information regarding AFM-based imaging and discussion of static and dynamic modes.17,18 Under ambient conditions the magnitude of the tip-to-sample force in the contact mode is typically between 1 and 100 nN. This force for a regular tip (radius of few nanometers) results in a pressure of few GPa, which is on the order of yield stress of glassy polymers, thus often causing plastic deformation. On the other hand, the forces are greatly reduced to 0.1–1 nN by performing the scanning in fluid (water, organic solvents, etc.) because the capillary forces are significantly minimized. Overall, imaging in contact mode involves relatively large shear forces, frequently resulting in the damage and distortion of soft surfaces, making it unfavorable for polymeric and biological samples and it is thus employed only in some special cases (e.g., for friction force microscopy, see below). In order to prevent surface damage caused by contact imaging, noncontact modes were developed.19,20 Generally, noncontact modes operate with the probe scanning about 5–40 nm above the sample surface, perturbed by the attractive van der Waals forces between the tip and the sample; see Figure 1B. In order to overcome the limitation of the relatively weak tip–sample interaction force observed under static noncontact mode, the cantilever is set to oscillate at or slightly off of the resonance frequency of the cantilever. The lateral resolution of the dynamic mode is typically limited to 0.5 nm for topography and around 10 nm for other properties. Dynamic modes can reduce the typical operational forces by at least one order of magnitude compared to the contact mode (usually well below 1 nN). It virtually eliminates the shear force associated with the lateral raster scanning and reduces the tip sample contact duration by two orders of magnitude. Noncontact modes have been applied for studying a wide variety of materials such as metals, semiconductors, polymers, and biological materials. These modes offer unique advantages for probing the soft polymeric and biological samples compared to contact AFM. Although in a practical version of noncontact mode, so-called tapping mode, forces are considered minimal, they are nonetheless substantial and might result in surface modification and damage, especially in hard tapping.21 In general, the phase shift at modest tapping forces is proportional to the stiffness of a material, but stiffness is dependent on the contact radius, which is generally larger for softer materials under the same forces. On the other hand, these tip–surface contact area issues are less important under medium-tapping forces, compared to hard tapping forces. Furthermore, because the tip–surface contact area can significantly affect the phase shift angle, it is important to consider the effects of topography when interpreting phase images.22 To date, tapping mode has been extensively employed for imaging a wide variety of polymer surfaces such as hard, glassy polymers; crystalline polymers; rubbers; gels; polymer fibers; polymer blends; block copolymers; and polymer composites. Apart from tracking the surface topography using the weak van der Waals forces, noncontact dynamic mode is employed for probing other weak forces such as electrostatic and magnetic, as discussed later.

238

M. E. McConney et al.

Figure 2. AFM images revealing the conformations of adsorbed P2VP chains: (a) pH 3.89, extended coils; (b) pH 4.04, intermediate state; (c) pH 4.24, compact coils. Plots depicting the (d) RMS end-to-end distance and (e) RMS radius of gyration of P2VP single molecules adsorbed on mica surface versus pH. (Obtained from Roiter and Minko29 with permission from the American Chemical Society.)

One common approach to using lift mode involves a special raster scan where each line is scanned twice before the next line is scanned. In the first line scan, the topography is scanned in a conventional manner, such as tapping mode, and then the probe is lifted by a set amount (several nanometers) and the probe retraces the previous topographic line scan, which thereby effectively eliminates the topographical contributions to these other signals. For comprehensive review of the basic AFM modes of imaging and their application to the various classes of polymers the readers are referred to corresponding reviews and books on the subject.6,17,18,23–28 One very recent notable study was the use of AFM for revealing the conformation of a single polymer chain directly in fluid.29 Using light tapping mode (98% free amplitude) imaging under controlled pH, Minko et al. observed the conformation change in poly(2vinylpyridine) (P2VP) chains adsorbed on atomically flat mica substrates (Figures 2A–C). The P2VP chains exhibited a sharp globule to coil transition with a change in the pH from 4.0 to 3.8. Analysis of the AFM images clearly revealed that the protonation of the P2VP chains (with change in pH) dramatically altered the RMS end-to-end distance and the radius of gyration (Figures 2D and E).29 Tsukruk and coworkers have performed ambient and in-fluid tapping mode imaging of the surface morphology of the mixed covalently grafted brush layer about 5 nm thick composed of Y-shaped binary molecules polystyrene (PS) and poly-(acrylic acid) (PAA; Figure 3).30,31 The surface topography images revealed the nanoscale network-like surface morphology formed by coexisting stretched soluble PAA arms and collapsed insoluble PS chains in water. Exposure to different fluids (selective solvents for individual or either

Imaging and Force Spectroscopy

239

Figure 3. Top: Schematics of molecular transformations and AFM images of Y-shaped amphiphilic brushes combining two dissimilar hydrophobic and hydrophilic polymer chains (polystyrene [PS] and poly-(acrylic acid) [PAA]). Bottom: AFM images collected in light tapping mode in different solvents. The images clearly reveal the switching surface morphology depending on the quality of the solvent for individual components of the mixed brushes (adapted from Lin et al.32 Copyright American Chemical Society).

blocks) resulted in dramatic reorganization of the Y-shaped brushes. The structural organization of the brushes ranged from a soft repellent layer covered by swollen PS arms in toluene to an adhesive, mixed layer composed of coexisting swollen PAA and collapsed PS arms in water (Figure 3).32 The motion of macromolecules, polymers, and biomolecules can be observed in real time with AFM.33–37 This technique has been particularly useful in observing the molecular motion mechanisms of proteins. This technique has also been used to observe the reputation of polymers. The reptation of isolated isotactic poly(methyl methacrylate) (it-PMMA) chains deposited on a mica substrate was imaged in the tapping mode (Figure 4). The thin water layer (0.1 nm) adsorbed on the substrate accelerated the reptation movements. The reptation movements were also observed in the noncontact mode (frequency modulation

240

M. E. McConney et al.

Figure 4. 2D (top row) and 3D (bottom row) time-lapse AFM images showing the movements of an isotactic-PMMA chain on mica at lower humidity (34% RH). The arrow indicates movements of a loop along the chain. (Obtained from Kumaki et al.38 with permission from the American Chemical Society.)

mode) in which the tip force acting on the chains is smaller compared to that in the tapping mode.38 Figure 4 shows the detailed conformational changes of it-PMMA chains at 34% RH. The loop indicated by the arrow in the AFM image moved along the chain as shown by the 2D (top) and 3D (bottom) AFM in Figure 4. Though conventional AFM microscopy modes offer unprecedented vertical and lateral resolution, these techniques provide no information about subsurface features, except in the case of features very shallowly buried below the surface. Subsurface features (defects, fillers, and the like) can be nondestructively imaged using methods based on acoustic microscopy in which an acoustic (ultrasonic) wave is transmitted through the sample and

Imaging and Force Spectroscopy

241

the amplitude and phase of the acoustic wave are monitored to image the subsurface features. Ultrasonic force microscopy (UFM) is a robust technique developed for subsurface imaging and can be considered as a modification of the standard contact mode of AFM where the sample is oscillated at a high frequency (compared to the resonance frequency of the cantilever) by an additional piezo-resonator.39,40 The microcantilever exhibits nearly 102–104 times higher dynamic stiffness at frequencies much higher than the primary resonance frequency. The fundamental principle involves working in the inertial regime (high dynamic stiffness) of the cantilever and sensing the nonlinearity of the tip surface interaction. The sample oscillating at these higher frequencies exerts a constant additional force on the apparently stiff cantilever, elastically indenting itself into the tip. The modulation of ultrasonic waves passing through the sample thickness due to the varying local stiffness and buried features are detected as modulation of the cantilever deflection. Apart from subsurface imaging, UFM has been employed to probe the local mechanical properties of thin polymer films and composites, especially materials with high elastic moduli.41,42 A major issue with UFM for imaging the subsurface features is the nonlinear tip–sample interaction, which is extremely sensitive to the elastic and viscoelastic properties of the surface. Furthermore, the method is not ideal choice for soft polymeric and biological samples due to the relatively large forces of interaction between the tip and the sample. Overcoming these limitations, scanning near-field ultrasound holography (SNFUH) has been developed in which two ultrasonic waves are setup one from underneath the sample (2.1 MHz) and the other from the cantilever (2.3 MHz), forming a standing wave.43 The phase and amplitude of the sample scattered ultrasound wave, manifested as perturbation to the surface acoustic standing wave, are mapped to unveil the subsurface features.

2. Mechanical Characterization of Polymer Surfaces The AFM is capable exerting and detecting forces orders of magnitude lower than that of the chemical bonds.44 The photodetector has sub-Angstrom sensitivity, resulting in the theoretical ability to measure forces down to 0.1 pN, but noise from thermal, electronic, and optical sources limits the force sensitivity in ambient conditions to about 1 pN, with practical limits closer to 5 pN.44 Therefore, it should be should be quite evident that AFM has the potential to address materials and molecules with minimal forces over minimal surface areas. This part has a major section dedicated to force spectroscopy due to the ubiquitous nature of this method and because there are several techniques that use common fundamentals related to force spectroscopy. 2.1. Probes for Characterizing Mechanical Properties Regular AFM probes are fabricated from silicon or silicon nitride with typical radii of 10–20 and 20–30 nm, respectively. Silicon nitride probes are preferred for very stiff surfaces (the elastic modulus higher than 3 GPa). For these probes, at regular forces exerted during probing mechanical properties the diameter of the contact area usually does not exceed 1–3 nm and thus mechanical or adhesive properties can be probed with near-molecular resolution.45 However, the use of these highly hydrophilic tips is generally feasible for relatively stiff materials (usually with the elastic modulus higher than 1 MPa) with nonhydrophilic and low-adhesive surfaces. In the case of hydrophilic, highly compliant materials (e.g., hydrogels) with sticky surfaces, regular tips are prone to contamination and easy piercing. In these cases, colloidal probes and chemically modified tips should be used.

242

M. E. McConney et al.

Colloidal probes are fabricated by carefully gluing microspherical particles onto the end of a tipless cantilever.46,47 The microparticles are available through several commercial sources and colloidal probes themselves are commercially available as well. Microparticles with a diameter of a few micrometers from silica and borosilicate glass are most commonly used and have roughness below 1 nm for a square micrometer area, acceptable for most measurements of soft materials. Force spectroscopy performed with such probes is usually called colloidal force spectroscopy (CFS). The colloidal probes have several advantages over conventional probes for very compliant materials. A major advantage is that the applied forces per a unit area are significantly lower than conventional probes, thus allowing for probing very compliant materials such as hydrogels with the elastic modulus well below 1 MPa and down to a fraction of kPa and even few Pas.48 By applying less force per unit area, the total applied force can be much higher without plastically deforming the surface or damaging the probe, which provides higher resolution in force/area per a force curve by sacrificing lateral spatial resolution. It is very important to note that probing depths are highly dependent on the probe radius, and therefore colloidal probes are inappropriate for characterizing the stiffness or elastic modulus of ultrathin compliant films. Furthermore, the microparticle radius quoted by the manufacturer is generally quite accurate compared to conventional probes and can be easily verified with SEM. The preservation and well-defined tip shape allow for very good analysis with contact mechanics models that assume spherical shape of the probe, such as the Hertzian approximation (see below). However, care should be taken in preparation to ensure good particle–cantilever contact and that the probing particle surface is not covered with glue. It is also possible that the mechanical properties of the glue between the sphere and cantilever can be sampled instead of the sample itself when measuring stiff samples, such as reinforced polymers. Chemical modification of probes is generally used to enhance or reduce tip–sample interactions, which can be useful for a variety of applications including chemical force microscopy and chain-pulling experiments.49 Probes are usually modified with self-assembled monolayers (SAMs) with thiol chemistry on gold precoated tips or silane chemistry on native silicon oxide surface. Thiol-based surface modification involves coating tips with an adhesion layer followed by a gold coating. Silane modification can be done directly on silicon and silicon nitride tips after thorough cleaning.50 Thiol modification involves noncovalent bonding, which leads to a limited lifetime.50 Though thiol SAMs are an important tool for surface scientists, they are poor surface modifiers for applications involving relatively high forces, such as contact mode technique. On the other hand, silane-based modifications involve covalent bonds, which are quite robust and long-lasting.50 Unfortunately, silane modification involves relatively stringent reaction conditions and is somewhat difficult to initially optimize to achieve single monolayer coverage. The reaction is very sensitive to water presence, so the relative humidity has to be limited to a few percent and dry solvents must be used. On the other hand, thiol modification is relatively straightforward and can be conducted under ambient conditions. The ease of thiol tip modification has led to its extensive use even in contact mode and friction modes, causing widespread characteristic artifacts to be generated. 2.2. Force Spectroscopy 2.2.1. Principles of Force Spectroscopy. Surface force spectroscopy (SFS) is a powerful method to probe the nanomechanical and adhesive properties of surfaces, such as

Imaging and Force Spectroscopy

243

quantification of the elastic modulus, adhesion, chemical binding, inter-/intramolecular forces, resilience, elasticity, and more. Modified SFS techniques are also quite useful for electrical and thermal characterization of materials. The so-called pulling-off version of SFS is widely utilized for investigation of protein unfolding, brush stretching, and other tensile-related mechanical properties of individual molecules and requires usually special tip modification with selectively binding groups. Discussion of this approach can be found in some recent papers and reviews and will be not discussed in this review.49,51–57 AFM-indentation based methods that involve plastic deformation can be used to investigate material properties and material failure mechanisms. Indentation methods offer an alternative to elastic SFS measurements, thereby avoiding the difficulty arising from minimizing applied loads. Although AFM indentation-based measurements are an invaluable tool in polymer material analysis, this subject will not be discussed in this review; the reader is referred to relevant papers and reviews.58–61 As a surface-based technique, SFS is well suited to study the effect of free surfaces and confined surfaces on polymeric properties, which can be quite different from bulk properties. Force spectroscopy measurement is a multistep process, which should be done with great care to ensure accurate results and avoid misleading results. Therefore, it is quite important to fully understand the process and the sources of error. Furthermore, like many experimental methods, practice and experience with known samples is invaluable. Every sample behaves somewhat differently and therefore there is usually a learning curve associated with each new sample.62,63 A single force–distance curve is a plot of tip–sample force vs. piezoelement movement (Figure 5A). In Figure 5A is an ideal force–distance curve plotted in the conventional trace–retrace manner. The x-axis can be generally understood as the distance between the tip and the surface. First, the vertical piezoelement is moved in the extension direction, which is depicted in the solid line in Figures 5A and B. In the curve, line 1–2 is called the extension zero-line, which corresponds to the region when the sample is not in contact with the tip but is moving toward the probe. Line 2–3 corresponds to the “jump to contact” region (also known as the snap-to region), when the probe is initially attracted to the sample surface, thereby bending the cantilever downward. The surface is also deformed slightly toward the tip in the snap-to section of the curve. This snap-to section corresponds to an unstable displacement of the tip, where the movement of the free end of the cantilever cannot be directly related to the movement fixed end and therefore the sample penetration is not directly measured. The deflection of the cantilever, when in contact with the sample surface, is indicated by line 3–4. In this region, as the piezoelement moves the sample surface closer toward the cantilever, the cantilever passes from being bent downward through the zero deflection to being bent upward. This region is linear for purely elastic deformation with a slope directly related to surface stiffness. For infinitely stiff substrates that are utilized for sensitivity calibration, the slope is 1, which reflects the fact that the cantilever deflection is exactly equal to the piezoelement displacement. In the case of time-dependent surface deformation (viscoelasticity), nonuniform deformation, or plastic deformation this region becomes highly nonlinear. Point 4 indicates the end of the tip extension sequence and the beginning of the tip retraction sequence. Ideally, lines 3–4 and 4–5 will partially overlap and have the same slope during extension and retraction. Generally, the line 5–6 region represents the force of adhesion, “pulling forces,” or the “snap from contact” event. It is vertical in ideal cases but can display complex shapes in special cases (e.g., “sawtooth” shape for multiple chain unfolding events). Piezoelement hysteresis can be noticeable in this region at high

244

M. E. McConney et al.

Figure 5. (a) An ideal force–distance curve as explained in the text. (b) The deflection data from (a) plotted with respect to time. (c) A schematic explaining the different regions of the force curve. Note that all the numbers in a–c correspond to each other. It is also important to carefully note that in the schematic, the laser spot, cantilever deflection and sample height (piezoelement position) correspond to their positions of the force curve.

frequencies (tens of Hz) appearing as a small initial upward deflection upon the start of retraction. Line 6–7 is again a region where the cantilever is free from the contact with the surface. It is important to note that the applied force is indirectly measured by the AFM by relating the photodiode signal to cantilever deflection and relating cantilever deflection to the applied force. The cantilever spring constant must also be calibrated for each set of force measurements. Force spectroscopy mapping (sometimes referred to as force–volume mode) is a spatial map of force–distance curves collected across a selected surface area. This force–distance curve matrix can be used for sampling statistics, as well as relating surface features to mechanical properties. Calibrating the photodiode sensitivity involves obtaining force curves on a material with a stiffness that is much greater than the cantilever stiffness and therefore can be considered “infinitely hard.” Typically, a freshly cleaned piece of silicon wafer using piranha solution is employed.50 Silicon substrates are immersed for 30 min, followed by several washings under deionized water and then drying under filtered dry nitrogen gas. It is important to be stringent with cleaning of the calibration sample, to ensure that no surface contaminants interfere with the accuracy of the photodiode calibration. Error in the photodiode sensitivity

Imaging and Force Spectroscopy

245

causes a shift in all of the data introducing systematic error which can be very significant and requires postmeasurement verification. The photodiode sensitivity calibration is based on relating the known movement of the vertical piezoelement to the cantilever deflection on stiff substrates (e.g., glass or silicon with the elastic modulus of 170 GPa). Several common pitfalls are related to calibrating the photodiode sensitivity, which are mentioned here. Firstly, it is important to have a freshly cleaned sample to prevent surface contaminants common under conventional lab conditions, which can invalidate the assumption that the penetration is zero. Another common pitfall is related to the thermal drift of the piezoelement. As measurements are performed, the piezoelement warms and the response of the piezoelement will drift. The piezoelement movement is calibrated when the scanner is warm and therefore the photodiode calibration should also be performed when the scanner is warm. Thermal drift is not a problem for z-closed loop scanners because the piezoelement movement is independently measured. For scanners without z-closed loop, the scanner can be warmed by “exercising” the piezoelement. In order to prevent tip damage the scanner can be exercised in free air by false engaging. To warm the head, one can perform force curves with relatively large ramps in air for 20–30 min to warm the head. There are several well-developed methods for measuring cantilever spring constants: the most common methods are the added-mass method,64 geometry-based methods,65–68 the spring-on-spring method,69 and the thermal tuning method.70 Special developments in the form of calibration plots and modified equations have been suggested for more complicated cases such as gold-sputtered silicon nitride cantilevers.67,71 Although the developed equations are quite good at expressing the spring constant based on the cantilever geometry, there is usually a significant difference between theoretical spring constants and experimentally measured spring constants.68 The spring-on-spring method can be used to measure cantilever spring constants by performing force curves on a previously calibrated cantilever. This method has good accuracy and can be estimated as roughly 10% when performed with care; it is also relatively easy to perform and is useful in the case when thermal tuning sweep cannot cover the resonance frequency of cantilevers. The thermal tune is generally easier to perform but is not available on all microscopes and for a whole range of relevant frequencies.72,73 It is well known that the finite tip-end dimensions (usually within 5–30 nm) distort the feature sizes of images within nanoscale features because of shape convolution (sometimes called dilation or convolution). Furthermore, tip shape is often the source of common scanning artifacts, such as doubled features or asymmetric tip. Although this is a common imaging problem, in this section we are concerned with tip dimension measurements regarding the tip–sample contact area during force spectroscopy measurements. The size and shape of the SPM tip must be known to quantify the applied force per area. Several methods have been used to measure the tip size and shape. SEM has been used with relatively good success, although the resolution is practically limited to 2–3 nm. Often the imaging should be performed on conductive tips and the accelerating voltage should be limited to avoid charging. It should also be noted that SEM can often lead to the formation of carbon-based structures on the surface from surface contaminations. Another common method involves calculating tip dimensions from images obtained by scanning samples with known dimensions under tapping mode (Figure 6). Often the nanoparticles are embedded in a poly-lysine coating or attached to amine-terminated SAM, which when scanned appears to help to remove tip contamination and prevent nanoparticle rolling and detachment. Scanning standard gold nanoparticles of diameters from 5 to 30 nm that are tethered to a modified atomically flat mica or silicon surface has proven to be quite

246

M. E. McConney et al.

Figure 6. AFM tip deconvolution and dilated image of gold nanoparticle (bottom); AFM image of gold nanoparticles with diameter 20 nm (top).

accurate at characterizing the very end most portion of the tip. The tip can also be characterized with transmission electron microscopy, by measuring the shadow created by the tip with higher resolution. Another method, so-called direct tip imaging, involves scanning microfabricated calibration samples with sharp features available commercially.74,75

Imaging and Force Spectroscopy

247

2.2.2. Elastic Modulus via Force Spectroscopy. A common misconception is that accurate (usually better than ±50%) quantitative elastic modulus data cannot be obtained from force spectroscopy. This belief comes from the inability to measure the tip surface contact radius in real time and a cumbersome, extremely time-consuming experimental routine that is rarely followed properly without devastating shortcuts. Furthermore, the unstable nature of the snap-to region prevents the exact knowledge of the contact point and creates a certain discrepancy in the initial penetration thus affecting a long chain of calculation. Ease of damaging soft surfaces is a common problem for these materials. Fortunately, for all practical purposes this problem is not as critical or devastating as perceived. Overall, when all steps are performed with care, the resulting data have shown very good agreement with known elastic modulus data for known materials. Overall, quality SFS results should be considered pretty accurate within 20% deviation beyond initial engagement instabilities, as has already been demonstrated for a number of soft materials.24,76,77 There are many tasks for which SFS elastic modulus measurements with nanoscale resolution are the only viable option, but other options should always be considered when high spatial resolution is not required, such as buckling-based metrology (BBM).78 BBM is much less time consuming and has about the same accuracy as SFS. BBM is very appropriate for ultrathin polymeric samples. Furthermore, generally buckling is used for homogenous samples, although in certain cases can be used to measure the modulus of individual components.79 Freely suspended films can also be characterized using the socalled bulging approach.80 However, every technique has its own set of limitations and issues to consider. Nonetheless, if one must use SFS for elastic modulus measurements there are several critical things to do in order to get high-quality quantitative results, including using a cantilever with an appropriate stiffness, stringently avoid tip damage, preventing sample damage, performing calibrations carefully, and analyzing data properly. Although technical steps for these measurements are well known and documented in multiple notes and manuals, here we will list major steps and offer critical evaluations of uncertainties, issues, and important details that are rarely discuss in casual texts. These subjects will be discussed in detail in the following subsections, except for photodetector and sensitivity calibrations, routines that have already been discussed. 2.2.2.1. Choosing Appropriate Cantilever Spring Constants. In order to properly probe the relative stiffness or quantify the elastic modulus of a surface it is imperative to use a probe with an appropriate spring constant–tip radius combination for the sample with particular stiffness. This strict requirement is a product of inherent nature of cantilever-based transduction; specifically, the fact that the applied force (deflection) sensitivity is inversely proportional to the surface deformation (penetration) sensitivity. Relative stiffness and elastic modulus are measures of the penetration versus deflection; therefore, the implication of this seesaw relationship between the sensitivities is that the ideal ratio of deflection to penetration is 1. Furthermore, measurements with deflection-to-penetration ratios of less than 1 or more than 10 results in the stiffness or modulus going to zero or infinity, respectively, because of instrument limitations. Figure 7 is a generalized graph that indicates the appropriate range of spring constants vs. sample elastic moduli if a standard AFM tip is utilized. This graph is based upon aforementioned criteria verified with actual measurements and is a crude guide for initially choosing the appropriate cantilever spring constants when the elastic modulus can be estimated. This graph is inappropriate for probes with large tip radii, colloidal probes. An

248

M. E. McConney et al.

Figure 7. A graph indicating the upper and lower limits of appropriate cantilever spring constants as a function of sample elastic modulus. The top inset shows that the unstable region above the upper limit corresponds to probing where the cantilever deflection is much larger than the sample penetration. The bottom inset shows that the unstable region below the lower limit corresponds to probing where the cantilever deflection is much larger than the sample penetration. (Adapted from Tsukruk et al.90 Copyright Wiley-Blackwell.)

estimation of the samples’ elastic modulus is required to choose an appropriate probe, which adds to the learning curve associated with measuring new samples. 2.2.2.2. Avoiding Tip Damage. Avoiding tip damage is extremely important to obtaining robust quantitative elastic modulus data, because tip–sample contact models typically require hemispherical (or paraboloid) tip shape and the tip radius. Therefore, when small indentation depth and lateral resolution are not critical and specimens are compliant, colloidal probes should be used. Silicon nitride tips are still sharp but much more resilient than silicon tips. To preserve the tip shape, first, great care should be taken when engaging on surfaces, especially stiffer ones. Engaging in contact mode requires properly setting the difference between the deflection offset and deflection setpoint. The deflection offset is the difference between laser light shined on the top half and bottom half of the quadrant photodiode, measured in units of voltage. The deflection setpoint is a user-defined value, which when engaging is used to define the relative amount of cantilever deflection until the system considers itself engaged. In other words, when engaging, the microscope will continue to move the cantilever toward the sample until the deflection offset matches or exceeds the value of the deflection setpoint. It is important to not engage too hard, or one will destroy the tip. Therefore, the safest approach is to set this difference to be fairly small, which will

Imaging and Force Spectroscopy

249

likely falsely engage the tip at first. Then systematically slightly increase the difference, until the probe properly engages on the surface. Usually the deflection setpoint is kept at zero and the deflection offset is altered, which helps to ensure that the laser spot is near the center of the photodetector when the cantilever is engaged. Ensuring that the laser spot is in the center of the photodetector is important. Checking for a false engagement is relatively easy, by changing the deflection setpoint by a small amount (∼0.02 V) in contact scanning mode with the x-y range set to zero and checking for a noticeable change in the z-piezoelement position. It should be noted that this check will not work when the tip has been engaged too hard. Proper engagement on the sample should result in the ability to reach the sample with the piezoelement during extension and pull off the surface of the sample during retraction. It is most common to crack and destroy the tip when calibrating photodetector sensitivity when performing SFS measurements on a hard substrate (elastic modulus above 10 GPa). Unfortunately, it is even more common that the tip shape will not be exactly hemispherical after performing sensitivity, even with utmost care. It is quite difficult to accurately calibrate the photodetector without destroying the tip, especially when using ultrasharp tips with small radius of curvature. A good alternative in avoiding tip damage from sensitivity calibration is offered in a recent method developed involving thermal tuning to estimate sensitivity, without the need to perform force curves.81 This is a little less of an issue for adhesion measurements because typically relatively soft cantilevers can be calibrated with minimal forces, but nonetheless tip shape is just as critical in this case. This problem is much less critical when microscopic colloidal probes are used, where the contact area is much larger. After carefully engaging, the calibration force curves must be obtained extremely carefully. The scanner should be set to take individual curves as opposed to continuously taking curves. The trigger should be extremely small to avoid excessive deformation. That said, the trigger is directly dependent on the photodetector sensitivity. Therefore, initially one should assume a value slightly higher than typical sensitivity to avoid large forces in the first few curves. Typically, triggers should be set to 5–10 nm or less for relatively stiff surfaces and even below 1 nm for very stiff cantilevers. Relatively small ramp sizes should be selected to increase the number of data points in the contact region of the curve and to help prevent tip damage. The number of performed force curves should be minimized; a few (5–10) repeatable measurements at given location is usually sufficient. The engagement and sensitivity measurement should be performed at least several times in different locations to ensure accuracy and eliminate site-specific deviations. It is common to destroy a tip or two in order to estimate the parameters to get force curves in a safe manner. It should be stated that much of these problems can be avoided by performing the sensitivity after the force curves are obtained, but there are some disadvantages to this approach, which are discussed in detail in the section on issues regarding the execution of these measurements. 2.2.2.3. Avoiding Surface Damage. To measure the linear elastic modulus it is imperative to avoid plastically deforming the sample surface. Trigger values should be set by keeping in mind the applied force, which ideally should not exceed a few nN. It is critically important that the sample should be imaged in tapping mode before and after (should be zoomed out prior to scanning) the force curves are obtained. If plastic deformation occurs it will appear in the zoomed-out image as an array of indentation marks. Furthermore, plastic deformation can often be recognized in the force curves as a leveled-off slope at a fairly uniform deflection and from hysteresis between approaching and retracting portions of force–distance curves.

250

M. E. McConney et al.

2.2.2.4. Execution of Measurements. The order of execution that the elastic modulus measurements are performed is critically important. A common procedure execution can be described by the following steps: 1. Clean silicon immediately before measurements as described in sensitivity calibration. 2. Warm scanner head for at least 15 min, depending on the system; always ensure that the head is warm before taking any SFS data. 3. Perform sensitivity calibration, verify linearity, and optimize photodiode signal. 4. Characterize tip shape and radius. Proceed if tip shape is hemispherical. 5. Estimate initial cantilever spring constant (e.g., by using manufacturing data). 6. Image sample surface at several locations and magnifications. 7. Perform force curves and examine deflection–penetration ratio to ensure that cantilever spring constant is appropriate (note that, without measuring sensitivity first, this is just a gross estimation). 8. Scan surface again in zoom-out mode to verify absence of indentation marks. 9. If appropriate, repeat the tip calibration and check the tip shape again. 10. If there is any change in the total photodetector sum, repeat the photodetector calibration. 11. Measure the exact value of cantilever spring constant. 12. Conduct data processing and analysis of the results. An alternative to the execution listed above would be to perform the sensitivity at the end of the measurements, which prevents tip damage to great extent. But without knowing the sensitivity before the SFS measurements, the trigger and the deflection–penetration ratio can only be roughly estimated by doing sensitivity on a tip from the same box before the measurements and being careful to put the laser spot on the same part of the cantilever. As mentioned earlier, an alternative sensitivity calibration method developed by Higgins et al., which involves thermal tuning, can be used to avoid tip damage.81 The trigger, penetration–deflection ratio, and total penetration are very important and so by not knowing sensitivity one is essentially going at it blind, hoping for the best. So, there is a trade-off, because significant time can be wasted when inappropriate experimental conditions are used due to a lack of knowledge of the sensitivity. When the sensitivity is obtained after the measurements, data processing is required to apply the correct sensitivity. If the sensitivity is performed after the measurements to ensure tip preservation, then steps 1–3 should be moved and can replace step 9. It is important to note changes in the laser spot, typically observed as a change in the detector sum, because this is an indication that the sensitivity calibration changed over the course of the experiments and thus the measurements are void. There are several other things to note when performing measurements, including that the scanner warming is not critical for scanners with a z-closed loop. The order of the cantilever spring constant calibration is not too critical, although often knowledge of the sensitivity is required depending on the calibration method, and one may damage the tip performing the tip-on-tip method of calibration. A second tip size calibration may be necessary in the case of a mistake leading to larger forces, if the sample is very stiff or if there is an indication of tip contamination. If the tip size noticeably changes in the course of probing, everything should be redone with a new probe. 2.2.2.5. Tip–Surface Contact Models for Elastic Modulus. Calculating elastic moduli from applied loading force and sample penetration data involves applying a model to account for the tip–surface contact area. Here, the most basic and common models are

Imaging and Force Spectroscopy

251

presented briefly, specifically the Hertzian model, Sneddon model, and Johnson-KendalRoberts. These models are inappropriate for materials with nonlinear elasticity, such as certain gels and biological materials. The reader is referred to several reviews for detailed information regarding nonlinear elastic contact models.82,83 Typically, equations are derived from a quasi-static spring-on-spring model or force balance approach as expressed by zdefl k = P (h)

(1)

where P is the applied force and h is the sample penetration as already defined.84 Assuming a spherical tip, flat surface, and no plastic deformation, one can define an effective spring constant or stiffness for a material as: E ∂P = 2r (2) kM = ∂h 1 − ν2 where r is the tip–surface contact radius, E is the material elastic modulus, and ν is the material Poisson’s ratio.85 Unfortunately, even after determining the tip radius, there is currently no known way to measure the contact radius at nanoscale in real time as the measurements are performed. Instead, contact mechanics models are used for fair estimation, which generally differ in the approaches on considering tip–surface interaction’s contribution to the contact area.62,63 The most popular Hertzian contact mechanics model is applicable for small deformation and it assumes that the adhesion forces are zero and that at zero applied load the contact area is also zero, all of those being far from true in most SFS measurements. However, in the vast majority of practical cases these contributions can be ignored or proper corrections can be made. The force as a function of penetration depth described by the Hertzian model is P =

4 1/2 3/2 R h E 3

where R is the tip radius and E is the composite modulus defined as 1 3 1 − νS 1 − νT = + E 4 ES ET

(3)

(4)

where the subscript S and the subscript T refer to sample- and tip-related variables, respectively. The modulus associated with the probe is generally assumed to be much larger than the elastic modulus of the surface, which is surely true for all polymeric surfaces. Therefore, a simplified equation for the elastic modulus based on the Hertzian approximation can be expressed as: dP 3 1 − ν2 (5) E= 4 R 1/2 d(h3/2 ) When plotting the penetration raised to the 2/3 power versus the deflection (or the applied force) a straight line should result from data taken with a spherical tip and little to no surface–tip interaction. Poisson’s ratio is usually taken as known bulk values typically ranges between 0.3–0.5 (about 0.5 for most of elastic materials/scenarios) are modest considering overall minor contribution in Eq. (5). The Sneddon model is another popular model that can be utilized to describe tips with an elliptic paraboloid shape and for significant deformations. In this approach, the

252

M. E. McConney et al.

paraboloid function Z = bX2 is utilized to describe the contact area as:

a=

(6)

h 2b

(7)

where R in Eq. (5) can be replaced with 1/(2b) to calculate the elastic modulus. The Johnson-Kendal-Roberts (JKR) model includes an adhesive contribution, which can be expressed in terms of a reduced load, PJKR . The elastic modulus from the JKR can be described by the modified Hertzian relationship between the load and the contact area as: dPJKR 3 1 − ν2 (8) E= 4 R 1/2 d(h3/2 ) where the reduced load, PJKR , is defined as: POff 3/2 PJKR = √ P1 3

(9)

where POff is the force associated with the snap-from portion of the force curve, line 5–6 in Figure 5, and P1 is defined as: P1 = (3P2 − 1)

1 9

13 (P2 + 1)

(10)

where P2 is defined as: P2 =

Zdefl +1 Zadh

12 (11)

where Z adh is the cantilever deflection associated with the snap-from, line 5–6 in Figure 5. As mentioned earlier, there is a discrepancy regarding the initial deformation at snap in, or the zero contact point. This point is usually taken as either the snap-to point (the minimum deflection point in the extension curve) or the zero deflection point after the snap-to in the deflection curve, the imaginary intersection point between line 1–2 and line 3–4 in Figure 5. The initial penetration overestimates the modulus and as the penetration depth increases the measured modulus will steadily decrease to the “true value.” If the total deformation well exceeds (two to three times) the initial contact penetration, the true value of the elastic modulus can be obtained anyway. This is a usual case for elastic materials where overall elastic deformation of 10–100 nm utilized for data analysis is much higher the initial deformation of 0.5–3 nm. 2.2.3. Examples of SFS Measurements. There are many different ways to utilize the elastic modulus and surface stiffness measurement capabilities of SFS. Choi et al. demonstrated the variation of the elastic modulus in periodic polymer structures fabricated by multi–laser beam interference lithography.86 The variation in the elastic modulus of the SU8 microstructures was believed to be to due to the periodic variation in the cross-linking density resulting from the light intensity distribution. These measurements avoided effects of geometry by careful control of the probed depth. On the other hand, macroscopic deformation measurements (tensile test and peel test) were performed to reveal the ductile failure

Imaging and Force Spectroscopy

253

Figure 8. Rubber microphase inclusion in glassy polymer matrix for PS-PB blend (AFM topography slice, top, left) along with examples of force mapping (3-D topography, middle; topography, top; modulus, bottom) and surface histograms for adhesion (right, top) and modulus (right, bottom). Temperature variation of the elastic modulus for glassy and rubbery phases are presented as well (left, bottom). (Adapted from Tsukruk et al.91 Copyright Elsevier.)

and necking of the thin nanoscale struts.87,88 SFS measurements can also be very useful to probe phase transitions by performing measurements with varying probing frequencies and/or sample temperatures as demonstrated in several cases.89,90 For instance, we studied the surface distribution of the adhesive forces and elastic moduli for heterogeneous glassy–rubbery polymer films.91 Micromechanical properties of polystyrene–polybutadiene (PS-PB) thin films were probed in the range of temperatures. We demonstrated that for heterogeneous films fabricated from polymer blends, the micromapping of surface properties can be obtained concurrently for glassy and rubber phases as well as across the interface with a lateral resolution better than 100 nm (Figure 8). Histograms of the surface distribution display two very distinctive maxima for both adhesive forces and the elastic moduli, which allows concurrent measurements of micromechanical properties of glassy and rubber phases. Glass transition temperature of glassy matrix and flow temperature of the rubber phase can be also detected by this technique by measuring the surface distribution of elastic modulus in a range of temperatures. Both temperatures (glassy and rubbery phases) derived from these mapping were demonstrated to be close to the known values (Figure 8). The nanomechanical behavior of molecularly thick (8–10 nm) compliant polymeric layers with the nanodomain microstructure from grafted block copolymer, poly[styreneb-(ethylene-co-butylene)-b-styrene] (SEBS or Kraton), was probed with micromechanical surface analysis based on scanning probe microscopy.92 The micromapping with high lateral resolution (below 8 nm per pixel) revealed the bimodal character of the nanomechanical response with different elastic moduli shown by the rubber matrix and the glassy nanodomains (Figure 9). High-resolution probing showed virtually constant elastic response for the compliant layer compressed to 60% of its initial thickness followed by a sharp increase of the resistance when the tip reached within 3 nm from a stiff solid substrate.

254

M. E. McConney et al.

Figure 9. Top: AFM images (topography and phase of SEBS layer). Bottom: Surface distribution of apparent elastic moduli for the polymer layer collected with 64 × 64 resolution, size is 500 × 500 nm, lighter areas correspond to higher moduli along with histogram of the elastic modulus obtained from micromapping. (Adapted from Luzinov et al.92 Copyright Elsevier.)

Application of the double-layer deformational model allowed the estimation of the actual elastic moduli of different nanophases within the grafted polymer monolayer: 7 ± 3 MPa for the rubber phase and 20 ± 7 MPa for the glassy domains (Figure 9). Relatively high elastic modulus of the rubber matrix is caused by a combination of chemical crosslinking/branching and spatial confinement within <2Rg layer. On the other hand, the observed low modulus of the glassy nanodomains can be attributed to both low molar weight of PS blocks and the presence of rubber layers in the probed volume. The approach developed for the microindentation of layered elastic solids was adapted to analyze SPM probing of ultrathin (1–100 nm thick) polymer films on a solid substrate.93,94 The model for analyzing nanoindentation of layered solids was extended to construct twoand tristep graded functions with the transition zones accounting for a variable gradient between layers. This “graded” approach offered a transparent consideration of the gradient of the mechanical properties between layers (Figure 10). By adapting this approach we considered polymer layers with elastic moduli ranging from 0.05 to 3000 MPa with different architecture in a solvated state and in a dry

Imaging and Force Spectroscopy

255

Figure 10. Top: Experimental loading curve (circles), fitting with the trilayered model (solid line, almost completely buried by experimental data points) and Hertzian model (dot line) for polymer brush in good solvent. Bottom: Experimental depth distribution of the elastic modulus for the polymer brush layer (circles) and the best fitting with the trilayered model (solid line) showing slight increase in the elastic modulus near the surface and sharp increase in proximity to a stiff substrate. (Adapted from Kovalev et al.93 Copyright Materials Research Society.)

256

M. E. McConney et al.

state, respectively.93 The most sophisticated case of a trilayered polymer film with overall thickness of 20–50 nm and a combination of hard–soft–hard interlayers was also successfully treated within this approach. In all cases, a complex shape of corresponding loading curves and elastic modulus depth profiles obtained from experimental data were analyzed with the graded functions with nanomechanical parameters (elastic moduli and transition zone widths) close to independently determined microstructural parameters (thickness and composition of layers) of the layered materials. The elastic properties of dendritic (hyperbranched) molecules with diameter below 3 nm have been probed with SFS, which allows for the micromapping of the surface stiffness and adhesion with nanoscale resolution.95,96 To anchor dendritic molecules to hydroxyl terminal groups and reduce tip–molecule interactions, a modification of the silicon surface with an amine-terminated SAM and the AFM tips with methyl-terminated SAMs was used in this study. The nanomechanical response was analyzed in terms of sequential deformation of dendritic molecules and alkyl-silane monolayers (Figure 11). We observed much lower adhesion in the surface areas occupied by dendrimers as well as lower elastic modulus in these areas in comparison with the surrounding surface of SAM. This difference is caused by the reduced contact area between tip and 3-nm-diameter molecules and their high compliance in comparison with alkylsilane SAMs. Higher stiffness was also revealed for molecules within long-chain aggregates compared to individual molecules and small aggregates. Though traditionally SFS is most commonly used to probe glassy and rubbery materials, it is also used to probe materials with elastic moduli in the range of Pa-KPa, including hydrogels and biological materials. SFS has proved to be an invaluable technique in probing biological materials and a number of significant results can be found in literature, but only a few examples will be presented here. Although overall methodology for probing biological and synthetic materials is based upon similar fundamentals, specific routines, data collection, and interpretation might be very different. The reader is referred to relevant reviews for use of AFM for biological applications.97–100 Discussion of these differences is beyond the scope of this review.62 We point out a few very recent results for highly compliant and viscoelastic biological materials such as dynamic AFM probing of articular cartilage with both traditional sharp AFM tip and colloidal probes by Stolz et al.,101 mechanical probing of rat hippocampus and cortex areas as a function of aging by Elkin et al.,102 or monitoring changes in brain tissue specimens after injuries with microindentation by Shafieian et al.103 In a simple but robust study, Harmon et al. studied elastic modulus changes with temperature of photo-cross-linked poly-N-isopropylacrylamide swollen in water.104 The network’s modulus varied from 4.5 to 490 kPa over a 27◦ C range. The mechanical behavior with temperature was related to the degree of cross-linking and the polymer volume fraction. Gelatin films have been studied by Braithwaite and Luckman and a complex relationship between elastic and viscous responses has been found for different separations of tip and substrate.105 They concluded that careful analysis of these relationships could result in a separate calculation of loss and storage elastic moduli. Multi-element spring-dashpot models have been applied to analyzed force–distance curves. A quasilinear viscoelastic model was tested by Triphathy and Berger for SFS studies of agarose materials in order to derive full information on the viscoelastic behavior in a swollen state.106 Discher and coworkers used SPM to probe the elasticity of cell substrates and cell stiffness providing strong evidence that substrate stiffness significantly influences the cell lineage that stem cells express.107 McConney et al. probed a signal filtering material under varying frequencies and related the frequency-dependent mechanical properties of the

Imaging and Force Spectroscopy

257

Figure 11. SFS micromapping of G3 molecular aggregates, 64 × 64 array, 400 × 400 nm area: topography (top) and concurrently obtained surface distribution of adhesive forces (middle) and elastic modulus (bottom). (Adapted from Shulha et al.95 Copyright American Chemical Society.)

258

M. E. McConney et al.

biological material to the viscous signal-filtering ability, much as those found in spider hairs.108,109 The authors also studied biohydrogel cupulae of several different fish that are essential for underwater orientation of these species; these measurements helped guide the development of a bio-inspired material.110 A particularly novel use of SFS elastic modulus measurements demonstrated by Yamada and coworkers involved measuring the elastic modulus of rabbit muscle.111 The measurements found that the modulus tranverse to the muscle fiber changed from 11.5 in the relaxed state to 84.0 kPa in the contracted state. 2.2.4. Adhesion and Chemical force Spectroscopy. Quantifying adhesion from force spectroscopy can provide information regarding the intermolecular interactions and surface energies. Usually, precise measurements aimed at understanding specific chemical interactions involve the use of chemically modified tips in a mode of so-called chemical force spectroscopy. Chemical force spectroscopy is capable of providing rich information regarding chemical interactions with lateral resolution down to a single molecular group. A full discussion of chemical force spectroscopy and other associated techniques is beyond the scope of this review and the readers are referred elsewhere for more information.112–116 Adhesion force information is collected in the retraction portion of the force curve at the snap from region and usually pull-off forces are considered to be representative of true interactions, although some issues relevant to instability in tip behavior should be always considered. Often this adhesion data is a combination of several forces, which can include contributions from capillary forces, electrostatic forces, van der Waals forces, hydrogen bonding, ionic bonding, and covalent bonds. Capillary forces tend to dominate adhesive forces in ambient air, which are commonly on the order of 1–100 nN.115,117–119 Although capillary forces can be used to gauge hydrophobic/hydrophilic forces and thereby provide contrast, most adhesion measurements are aimed at obtaining chemical interaction information, where capillary forces are considered an undesirable interference. Therefore, in order to avoid overwhelming contribution from capillary forces, SFS measurements are performed under dry nitrogen or immersed in liquid. It should be stressed that much like elastic modulus measurements, there are errors associated with calculating the work of adhesion and surface free energies and care should be taken in both the measurements and data analysis. In this case the Hertzian model is inappropriate, and instead JKR and Derjaguin-M¨uller-Toporov (DMT) contact models are commonly employed. Under the JKR theory, the work of adhesion is defined as: WSMT

R 3 = π 2 Pad

(12)

where R is the tip radius and Pad is the force required to separate the tip from the sample surface. The DMT theory results in similar relationship with 3/2 coefficient replaced with 2. Both models have shown good agreement with experimental data despite mechanical instability of the cantilevers during pulling-off event and are more applicable to different limiting cases (materials with different balance of compliances and adhesion), as discussed elsewhere.84 Measurements on various surfaces performed in various solvents are well reported. Figure 12A presents force curves in ethanol for SAMs with different tip and surface functionalities.120 Measurements immersed in aqueous environments have led to more specialized measurements, namely, measuring the pK of surfaces, which can be powerful for investigating surface confinement effects of the ionizability of functional groups on

Imaging and Force Spectroscopy

259

Figure 12. Adhesion measurements are highly effected by surface chemistry, solvent, and probe chemistry (A) Top: a force–titration curve; Bottom: a corresponding contact angle titration curve. This figure demonstrates the capability of force spectroscopy to measure interfacial energy. (B) Force–distance curves taken in different solvents and with different tip functionalities. (Obtained from Noy et al.120 with permission from the American Chemical Society.)

the surface. In fact, there are significant differences in the dissociation constants from macromolecular surfaces compared to monomers in solution, which can be attributed to a variety of factors, including decreased available degrees of freedom associated with bonding/immobilization, the effect of the dielectric permittivity from adjacent functional groups, and the electrostatic free energy of the substrate.112,121 These differences are usually measured by quantifying the surface energy through contact angle measurements taken at different pHs. Similar measurements can be performed by monitoring adhesion forces between functionalized tips and surfaces with solution pH, through the so-called force–titration measurements introduced by Lieber et al.122 Figure 12B shows a plot comparing results from force titrations and contact angle titrations. Measurements involving surface that are incapable of dissociation show no effect from pH changes. A force–titration approach has the ability to map the surface energies on the nanoscale and associate any energetic contrast with nanoscale features.123 It should be mentioned that in order to probe unknown surface pKs, it is important to use tips functionalized with hydrophilic groups that are incapable of changing ionization with pH.112 Force–titration measurements demonstrate the important role that the surrounding fluid plays in localized adhesion measurements with SFS. Variants of this technique can also be used to study the interactions between macromolecules in different chemical environments. Jiang et al. explored the interaction of hydrophobic and hydrophilic polyamines with polyethylene oxide (PEO) at various solution pHs, which has key implications for PEO’s applications as a biocompatible material and anti-biofouling material.124 Surprisingly, this work indicated that polyallylamine, a hydrophobic polyion, can have favorable adhesion to PEO in an aqueous environment.

260

M. E. McConney et al.

There are many different ways to apply the capabilities of adhesion measurements and chemical force microscopy toward polymeric materials. A major area of interest in composite material development is the interaction of functional micro-/nanofillers with polymer matrices. A novel use of force spectroscopy has been in quantifying these interactions. For example, Barber et al. studied the interaction of carbon nanotubes and polyethylene-butene with the use of probes with nanotubes attached to the tip.125 This was accomplished by repeatedly heating the polymer matrix above the glass transition temperature, pushing the nanotubes into the polymer, and cooling and then pulling the nanotubes back out of the polymer, as is shown in Figure 13A. The interactions between individual functional groups and individual groups on freshly prepared carbon nanotubes has also been studied by LeMieux et al. with functionalized tips in fluidic environment achieving single molecular group interactions conditions for a variety of important functional groups (Figures 13B and C). Intermolecular interaction histograms obtained in this study were shown to follow proper theoretical predictions based upon the variation of the electronic state for different gaps. These results can be used to

Figure 13. Studies using force spectroscopy to measure the interaction energy between soft matter and carbon nanotubes. (A) A graph showing the force to pull a carbon nanotube out of a polymer matrix. The inset is a schematic explaining the testing approach and the various regions of the graph. (B) A schematic indicating the testing method of another study involving measuring the single molecule interactions with carbon nanotubes. (C) The resulting data provide the binding force histogram of the different molecules with the wall of single-wall carbon nanotubes. ((A) obtained from Barber et al.125 with permission from The American Institute of Physics; (B) and (C) adapted from Friddle et al.126 Copyright Nature Publishing Group.)

Imaging and Force Spectroscopy

261

better design carbon nanotube–polymer interfacial chemistry and can be considered for understanding of surface defects in carbon nanotubes.126 2.3. Pulsed-Force Microscopy Pulsed-force microscopy (PFM) is a popular method used to map topography, relative stiffness, adhesion, and electrostatic properties at about 1 ms per a pixel, about 1,000 times faster than conventional force spectroscopy. Since the advent of PFM by Marti and coworkers, the popularity of this method is gaining momentum, especially since it has become commercially available.127 Simplistically stated, PFM is essentially dynamic force-curve mapping. PFM typically involves driving a z-piezoelement with amplitude from 10 to 500 nm at 100 Hz to 5 kHz, orders of magnitude less than the cantilever resonance frequency, thereby obtaining force curves on the millisecond timescale.128 The z-piezoelement is driven with a sinusoidal profile, as opposed to the triangular wave that drives typical force curves. A typical PFM force–distance curve can be seen in Figure 14A, with the piezoelement driving signal plotted as the dashed line.129 PFM uses minimized forces, which avoids plastic deformation with ease. Furthermore, PFM measures and maps adhesion force directly. This technique acquires the same information as static force spectroscopy mapping, but the sampling is much faster. The height, stiffness, and adhesion measurements come from a peak-trough picking routine, which is used to quickly process the data, thereby providing images. The system acts under a constant force mode, like static force spectroscopy mapping, and therefore sample stiffness can slightly affect the height data. Softer domains would appear to be depressed compared to stiffer domains on a soft sample. The high sampling rate that the stiffness and adhesion forces are probed have a significant effect on the apparent behavior of the material due to time–temperature superposition effects. A study comparing adhesion data from SFS and PFM for poly-4-methyl-1-pentene under varying temperature highlights differences between the methods.128 The SFS and the PFM data show an expected increase in the absolute adhesion force with temperature, but the absolute adhesion force drops much less in the PFM data than the data from SFS as shown in Figures 14B and C. The difference is due to an apparent shift in the glass transition caused by the much higher probing frequency of PFM. Another interesting effect of the fast probing times is that amorphous regions can counterintuitively appear more sticky than crystalline domains, which have a higher surface energy, because the amorphous chains are much more mobile. Therefore, it should be quite apparent that SFS and PFM are highly complementary techniques. Between SFS and PFM, one could probe polymers at frequencies from 0.01 Hz to 5 kHz probing frequency range, where conventional SFS ranges from 0.01 Hz up to frequencies approaching 100 Hz and PFM ranges from 100 Hz up to 5 kHz. 2.4. Friction Force Microscopy Friction force microscopy (FFM) is an AFM-based technique used to characterize the tribological properties of surfaces.130 Common measurements with these modes involve relating frictional forces to applied normal forces (so-called loading curves). These measurements involve the use of an AFM in contact mode controlling normal loads and monitoring the lateral cantilever deflection (torsion) signal, as shown in Figure 15, left. FFM imaging can

262

M. E. McConney et al.

Figure 14. (A) A schematic showing the resulting force data from pulsed force microscopy. The dotted line shows the relative force modulation voltage. The arrows indicate the points picked to acquire the baseline, maximum applied force, and adhesion. (B) Adhesion vs. temperature from SFS and (C) from PFS of poly(4-methyl-1-pentene) (Tg = 303 K). ((A) obtained from Krotil et al.129 with permission from Wiley-Blackwell; (B) and (C) obtained from Marti et al.128 with permission from Elsevier.)

Figure 15. Right, a schematic showing the lateral and normal forces applied in FFM and the resulting laser spot deflection in the photodetector. Center, a 6 × 6 nm friction force image of KF (001) imaged in ultra high vaccuum. Left, a friction loop, indicating the average force level for each direction and the clear hysteresis between the directions. (Obtained from Carpick and Salmeron130 with permission from the American Chemical Society.)

Imaging and Force Spectroscopy

263

be as simple as scanning a surface in contact mode and using the lateral signal to produce a friction force image. The probe is scanned back and forth across a sample surface, generating so-called friction loops, which are the base for lateral force microscopy (LFM), another term for FFM (see Figure 15, center). Therefore, each back-and-forth cycle produces a “friction loop” curve as shown in Figure 15, right. The average friction force is quantified as half of the difference between the average friction force of each direction of the friction loop. A sample region is scanned under several different normal loads, thus producing a friction force vs. normal load curve. Usually measurements are performed on homogenous areas of a sample with smooth surface. Friction force images are often obtained for heterogeneous samples and the friction force contrast is associated with adhesive properties of components with common spikes between different areas recognized as “geometrical” friction artifact.131 Typically, measurements involve conventional silicon and silicon nitride probes, but diamond-coated probes are becoming popular for friction measurements because of the reduced wear associated with these probes. It is also possible to obtain chemical-related information from friction force images, especially when obtained with chemically functionalized probes. This can be done by simply obtaining a friction force image on a chemically heterogeneous sample.114 Friction at nanoscale is still far from well understood, but FFM has indeed provided an avenue for a deeper understanding. A notable example of the impact FFM has had on the understanding of friction was work performed by Carpick and coworkers.132 FFM (or LFM) is useful for polymeric studies involving reducing material wear, enhancing lubrication, stiction in microelectromechanical systems (MEMS), and other nanotribological phenomena.16,133,134 FFM can be sensitive to chemical information, especially when functionalized tips are used, providing information on shearing behavior of surface layer, friction coefficients, wearing dynamics, and velocity-dependent shearing. Application of FFM to SAMs, adsorbed molecular layers, and Langmuir-Blodgett (LB) monolayers from amphiphilic molecules have been widely exploited to elucidate their morphology and applicability as molecular lubricants with a variety of regular and extremely sharp tips.15,133 An example of FFM imaging of heterogeneous LB film from amphiphilic stearic acid presented in Figure 16 clearly demonstrates extremely low localized friction in selected areas coated with organic monolayer.135 Concurrently, extremely high friction is observed on bare silicon surface areas on the same image. FFM has also proved to be an invaluable technique for analyzing frictional properties of biological materials, such as cartilage. A notable example by Ortiz and coworkers involved the application of FFM in analyzing cartilage aggrecan under varying ionic strengths and with varying length scales.136 They found that at low ionic strength the lateral force did not vary with the lateral displacement rate, but at physiological ionic strengths the macromolecules’ frictional force significantly increased with the lateral displacement rate. This work provided molecular-scale insight into the deformation behavior of cartilage macromolecules. Vancso and coworkers performed FFM with varying tip radii over different temperature and wide frequency range (1–107 Hz) on polymethyl methacrylate to study the effects of surfaces on molecular mobility and free volume.137 This detailed study resulted in a master curve that was used to quantify the activation energy associated with the α relaxation mode and the sub-Tg β relaxation modes. The results showed a significant decrease in activation energies associated with the surface as compared to bulk values. This study in effect put to rest the controversy related to the effect that surfaces have on molecular mobility of glassy materials.

264

M. E. McConney et al.

Figure 16. Topography (left) and friction (right) for incomplete Langmuir-Blodgett monolayer of stearic acid (1 × 1 µm) showing high friction on silicon (bright areas) and very low friction on LB monolayer (dark areas). (Adapted from Tsukruk et al.135 Copyright American Chemical Society.)

3. Thermal Characterization of Polymers 3.1. Approaches Studying the local thermal properties of materials is of fundamental importance in understanding a variety of phenomena, including photon–phonon interactions, electron–phonon interactions, molecular motion, and various phase transitions.138 Although various thermal characterization techniques based on SPM have been developed, this review will focus on methods based on AFM techniques combined with the electrical resistance thermometry, which is applicable to polymer materials and is relatively well developed. This includes scanning thermal microscopy (SThM) imaging of thermal conductivity and other local thermal analysis (L-TA) techniques. The term micro- or nano-thermal analysis (micro/nano-TA) encompasses a variety of techniques involving characterizing localized material properties on a temperature controlled sample. Control over the temperature of a sample is provided by the use of a thermally active, electrically resistive probe (thermal probe) and/or a variable temperature microscope stage (temperature stage). If the latter is being used, practically any type of probe normally available for AFM may be mounted in the microscope. A thermal probe may function as a thermometer as well as a heat source. This enables a different type of micro-/nano-TA to be carried out, in which heat is applied to the sample from an external energy source (infrared radiation, for example) and the probe is used to sense the resulting change in temperature of the material. This approach enables spectroscopy to be carried out with a spatial resolution that is, in theory, better than the diffraction limit. Initially, Wollaston wire probes (loops) were the most common. These thermal probes were developed by Dinwiddie et al.139 and first used by Balk et al.140 and Hammiche et al.141 A diagram of the construction details of this probe is shown in Figure 17, top.142 The Wollaston probe with loop diameter of several tens of a micrometer is a relatively massive structure compared with most inert probes used in other forms of AFM. The Wollaston probe, whose high and variable spring constant (5 to 20 N/m) and complexity

Imaging and Force Spectroscopy

265

Figure 17. Top: A schematic (left) of a Wollaston resistive probe and corresponding micrograph (right). Bottom: A scanning electron micrograph of higher resolution thermal probes. (Top images were obtained from Gorbunov et al.153 Copyright Sage Publications; bottom image is reproduced with permission from Anasys Instruments, Inc.)

render it unsuitable for all but the contact mode. These probes routinely perform with spatial and thermospatial resolution on the submicrometer, although nanometer resolution is occasionally reported.140,141,143–147 Temperature distribution in the contact area of the thermal tip and polymer surface is relatively sharp under fast probing even with large thermal tip (Figure 18).148 Some of the best results for robust routine local thermal analysis with high spatial resolution comes from probes based on the approach adopted by King et al.149 The spatial resolution of these probes is the same as conventional AFM tips. Unfortunately, though these probes are capable of imaging topography at nanometer scale, these probes are not suitable for high-resolution thermal imaging. The heated area is on the top of the inverted pyramidal tip; this means that the resistive element that is sensitive to temperature is relatively large, of the order of 10 µm (Figure 17, bottom). The effect of this combination is that the heater serves very well to heat the tip, but the thermal resistance thermometer function is impaired. Microfabricated bowtie probes are probes where the metal conducting layer has a bowtie shape at the tip so that electrical resistance is located at the narrow middle area.147 Elongated rectangular discontinuities in the coating are detected in thermal images down to a width of 200 nm but only when lying parallel to the raster direction. The resolution perpendicular to this direction is shown to be roughly a factor of two poorer. Nonetheless,

266

M. E. McConney et al.

Figure 18. Modeling of temperature distribution in the contact area of thermal tip. (Obtained from Tsukruk et al.151 Copyright Elsevier.)

electrical-resistance thermometry probes are an active area of development and future developments will surely lead to enhanced thermal resolution. Quantitative thermal conductivities measurements can be obtained from both SThM and local-TA measurements as well.142,150 It is important that the temperature within the thermal contact is virtually homogeneous (<3% of variation for all materials) and the temperature in the center of the heated zone, Tc , can be used for the estimation of the average temperature.151 Also taking into consideration that the thermal probing of a surface can easily satisfy a quasi-stationary case for heat flow from the thermal probe to a surface, the heat transfer can be described by a quasi-stationary equation to analyze the dynamics of the heat dissipation as well.150 SThM enables the acquisition of images of the surface of a sample constructed from spatial contrast in one or more thermal properties of the material.147 The most common form of SThM is constant temperature mode, and SThM is used here to refer to this mode. In constant-temperature SThM, the thermal probe is held at a fixed temperature by means

Imaging and Force Spectroscopy

267

of a thermal feedback loop as it is raster-scanned across the surface of the sample. The AFM force-feedback mechanism holds it at a constant contact force. The power supplied to the thermometry probe to maintain it at the selected temperature is recorded and used to construct the “thermal conductivity image.”145 Areas of relatively high thermal conductivity will result in more power being supplied to the thermal probe than neighboring areas with lower conductivity. As with many SPM-based techniques, interpreting thermal images acquired with SThM is complicated by surface topography effects caused by variations in the contact area. For Wollaston wire probes, surface roughness above 30 nm can significantly affect quantifying thermal conductivity.152 Even with a thermally homogeneous material, the surface topography gives rise to thermal image contrast. It is also possible for topographical effects to mask or interfere with contrast of thermal images of multiphase materials. Because of this effect a careful visual comparison of the topographic and thermal images must be made to determine how closely the location, size, and shape of features in one image are reflected in the other. This problem is greatly reduced when a flat sample surface can be prepared by sectioning or polishing.

3.2. Local Thermal Analysis Local thermal analysis refers to a localized thermal measurement, much like force–distance curves are localized mechanical measurements. Typically, L-TA measurements do not involve mapping. These measurements are performed by contacting the surface with probe under a set force and then running a thermal measurement.141,143,153 After the tip is exerting a predetermined downward threshold force (i.e., cantilever deflection), a temperature ramp is applied to the sample via the probe. This is usually a linear heating program or linear heating followed by linear cooling. Heating and cooling cycles may be set at different rates. Essentially, two signals are acquired simultaneously, the vertical deflection of the probe and the power required to ramp the probe temperature providing information on thermomechanical properties. Indeed, the measurement of probe deflection with temperature is the micro (or local) analogue of thermomechanical analysis used on bulk samples (micro-/nano-TMA, L-TMA). Similarly, the measurement of probe power consumption is the micro-analogue of differential thermal analysis (micro-DTA, L-DTA). Both of these techniques have been used extensively in the study of thermal transitions in polymers and other materials.148,154 Once the probe is in contact with the surface and the temperature program is initiated, the force-feedback mechanism is disabled and the fixed end of the cantilever remains at a constant height throughout the experiment. Therefore, these experiments are not constant force experiments. Thermal expansion typically results in a steady increase of the cantilever deflection (thus in an increase in applied force) and an increase in the sample penetration. For a material that undergoes no thermal transitions over the temperature range of the experiment, the probe deflection with temperature will be essentially linear and upwards as the sample beneath the probe heats and expands. Heating a sample to a phase transition results in the softening of the sample material, measured by both a dramatic increase in the sample penetration (decrease in cantilever deflection) and an increase in power required to sustain the probe temperature. Figure 19 is a graph of the resulting data from a typical micro-TMA and micro-DTA experiments on polyethylene terephthalate (PET) surface, demonstrating how well-defined glass transition and melting phenomena reveal themselves in L-TA data during the heating cycle.

268

M. E. McConney et al.

Figure 19. A typical micro-TMA and micro DSC curve for poly(ethylene terephthalate). (Adapted from Gorbunov et al.154 Copyright Sage Publications.)

The heating of the probe element itself causes some movement of the cantilever, but for the relatively massive Wollaston wire thermal probe this effect should be minimal. Providing that a baseline subtraction procedure is carried out (acquired from a run with the probe in free air), the rate of power consumption of the probe over the duration of the same experiment on a sample should remain constant. When heating polymer through phase transitions (a glass transition, cold crystallization, curing, melting, or degradation), the response of the micro-TA signals can be quite rich and distinctive, as illustrated in Figure 19.151 This behavior shows typical micro-DTA and micro-TMA results for a crystallizable but initially mostly amorphous polymer. There is a large indentation at the softening temperature (around the glass transition temperature) and then a further indentation at the melting temperature. It is important to note that, the DTA signal is not sensitive to enthalpies like differential scanning calorimetry (DSC). During data analysis and interpretation it is important to note that L-TA is highly dependent on changes in the contact area; therefore, it cannot be known just from these data. This is a disadvantage of the micro-/nano-TA approach compared to conventional calorimetry. However, the glass transition and melting events are clearly detected. Furthermore, there are still many advantages to L-TA, including mapping DTA and TMA at sub-nanometer scale. L-TA can also be used to analyze ultrathin films. L-TA can be also used to study differences between surface and bulk properties through the use of suitable sectioning techniques.155 AC heating can be applied to SThM and such approach can also be applied to LTA experiments. AC heating may be applied to the thermal probe. This produces a fixed temperature modulation in the range of ±1 to ±10◦ C, although it is usually confined

Imaging and Force Spectroscopy

269

between ±2 and ±5◦ C. This may be seen as analogous to the development by Reading and coworkers156 of modulated temperature differential scanning calorimetry (MTDSC). In micro-/nano-TA the response of the sample to the modulated and underlying heat flows can be separated using a deconvolution program. The modulated regime is sensitive to the reversible changes in the heat capacity of the material, associated with molecular vibrations, and the latter detects changes due to kinetically controlled processes that are unable to reverse at the temperature and the rate of the modulation. An obvious advantage of this technique is its ability to characterize heterogeneous samples in which different types of transitions occur over the same temperature range. Theoretically, the use of AC heating offers similar advantages to those of modulated-temperature DSC over conventional DSC. It has been shown that the AC signals may be particularly sensitive to thermal transitions that produce a relatively large change in heat capacity for a small heat input.157 3.2.1. Thermal Probe Calibrations. In order to measure the temperature of local transitions, a temperature–resistivity calibration of the thermal probe must first be carried out. The subject of temperature calibration has been addressed comprehensively by Blaine et al.158 and by Meyers et al.159 This process typically involves measuring the resistivity of the tip at the melting transition of known calibration samples. These calibration samples should be over several hundred nanometers thick to avoid substrate contributions. Furthermore, the calibration sample’s thermal conductivity should be considerably less than the probe material thermal conductivity. The L-TA technique acquires two signals (L-TMA and L-DTA), of which one should be used to indicate the transition for calibration samples. Presently, it seems better to use the L-TMA signal because this has a higher signal-to-noise ratio. The temperature calibration should be carried out on two or more substances whose melting temperature (Tm ) is well known from the literature. For this purpose, it is often more convenient to use polymer films whose melting point has previously been measured using DSC or another technique. This does make the assumption that the enthalpies measured in DSC experiments can be considered to mirror their mechanical analogues (i.e., softening). There is now a reasonable body of data in the literature that allows one to conclude that this assumption is, broadly speaking, appropriate. When using DSC data as the point of comparison there is the question of which characteristic temperature on the DSC curve should be chosen as the one corresponding to the penetration temperature. Several researchers have suggested using the extrapolated onset temperature of melting, but it is known that the leading edge of melting endotherms of polymers is not straight, so the determination of the “extrapolated onset temperature of melting” is subjective.158–160 Nonetheless, a good correlation between bulk and local measurements has been demonstrated, with the variation of the transition onset of ±3◦ C. 3.2.2. L-TA of Ultrathin Polymer Films. Analysis of the ultrathin polymeric films on a high-conductivity substrate is a significant challenge. The huge difference in the thermal conductivity of a polymeric film and a substrate results in heat dissipation mostly to the substrate through the tested film. In this case, L-TMA and L-DTA measurement procedures should be significantly modified.151 To account for the substrate, one can engage a separate reference probe on an identical substrate using a microscopic manipulator on a separate microstage under a stereo microscope. The two thermal probes should ideally have similar thermal characteristics and be independently tested prior to their selection to balance heat dissipation. With the modified measurement setup, the thermal sensitivity of the thermal

270

M. E. McConney et al.

Figure 20. Glass transition of polystyrene vs. film thickness. (Adapted from Gorbunov et al.154 Copyright Sage Publications.)

probe increases dramatically, thus allowing detection of minute heat dissipation variations (<1 µW) associated with the polymer film itself. The experimental data shown in Figure 20 indicate that the glass transition temperature decreases when the film thickness is less than 400 nm (compare this data with results for bulk PS film [1 µm thick] in Figure 20).154 For the thinnest film presented in this plot, the glass transition temperature decreases by 20◦ C from its bulk value. These results follow general trends observed for ultrathin polymeric films deposited on solid substrates with weak film–substrate interactions. These results can also be explained because of enhanced chain mobility associated with the free surface.161,162 This data demonstrates the sensitivity of the present micro-/nano-TA design to probe nanometer-thick polymer films. L-TA has also been used to characterize photodegradation and other phenomena that cause chain scission by studying changes in the glass transition after exposure to polymer degrading processes.163 It is interesting to note that force and heating rate have minimal effects on the melting temperature, but both force and heating rate have a significant effect on the glass transition temperature.

4. Electrical/Magnetic Characterization of Polymer Composites 4.1. Kelvin Probe Force Microscopy/Electrostatic Force Microscopy Kelvin probe force microscopy (KPFM) and its simpler analogue electrostatic force microscopy (EFM) enable the spatial mapping of the work function and the surface potential

Imaging and Force Spectroscopy

271

distribution with a lateral resolution of a few nanometers and a potential resolution of a few mV. Work function () is defined as the minimum energy required to remove an electron from the electronic ground state in a material.164 It is known that the work function () is a sensitive indicator of surface conditions and is affected by adsorbed layers, surface charging, surface imperfections, and surface and bulk contaminations. Though macroscopic Kelvin probes for measuring the average work function for relatively large surfaces are known for a long time, KPFM was first demonstrated in 1987 using optical heterodyne detection to track the frequency changes in a vibrating AFM cantilever resulting from the normal electric field components of surface charges and potentials. In KPFM a conductive tip scans over the surface, interacting electrostatically with the surface under investigation. The potential of the tip to that of the surface is matched using an electronic feedback. When the potential of the tip exactly equals that of the material, electrostatic interaction between the tip and the sample is nullified. The voltage applied to nullify the electrostatic interaction is the local measure of the work function or more directly the contact potential difference between the tip and the surface. If the work function of the tip is known, a quantitative two-dimensional map of the surface work function can be constructed from the applied DC feedback signal. The KPFM technique is performed the noncontact mode and does not involve the injection of any charges into the sample as in conductive AFM. Electrostatic force microscopy, on the other hand, is much simpler in that the technique does not involve in an electronic feedback. EFM is typically operated in lift mode, as previously described. An oscillating (at the resonance frequency) conductive tip biased at fixed DC voltage scans over the surface electrostatically interacting with the surface. EFM measurements are performed in two different modes: DC mode and AC mode. In the DC mode, a conductive tip oscillated near the resonance frequency (noncontact mode) with fixed DC bias (V tip ) scans over the surface at fixed height (few tens of nanometers) above the surface of the sample. The tip electrostatically interacts with the sample and the changing electrostatic force with the vertical separation distance, which causes a shift in the resonance frequency and the phase (ϕ) of the cantilever given by ϕ ∝ (d 2 C/dz2 )(Vtip + φ − VS )2

(13)

where Vs is the voltage within the sample, is the work function difference between the tip and sample, V tip is the tip voltage, and C is the tip sample capacitance. The observed phase shift is thus proportional to the square of the DC voltage difference between the tip and the sample. Thus, mapping the frequency or the phase shift as a function of the tip voltage pixel by pixel, or by calibrating the phase/frequency response of the cantilever as a function of bias and maintaining a constant value of V tip , EFM can be employed to measure potential profiles with the high resolution. KPFM has been employed for the investigation of organic solar cells comprised of poly-[2-(3,7-dimethyloctyloxy)-5-methyloxy]-para-phenylene-vinylene/1(3-methoxycarbonyl) propyl-1-phenyl [6,6]C61 (MDMO-PPV/PCBM) blends, identifying a barrier for electron transmission from the electron-rich PCBM nanoclusters to the extracting cathode.165 Figure 21 shows the topography and the KPFM mapping (under light illumination) of MDMO-PPV and PCBM blended film that was spin cast from chlorobenzene and toluene. The topography images clearly reveal that in the case of the films deposited from chlorobenzene, polymer nanospheres are distributed almost evenly throughout the bulk of the film, whereas a skin layer, incorporating polymer nanospheres, surrounds the big PCBM

272

M. E. McConney et al.

Figure 21. AFM and simultaneous KPFM images of the topography and work function of toluenecast blend film of MDMO-PPV/PCBM with a mass ratio of 1:4, measured in the dark and under 442-nm laser illumination. (Obtained from Hoppe et al.165 with permission from the American Chemical Society.)

clusters in toluene cast films. Apart from the dramatic difference in the morphology of the blend, the variation of the work function on the surface is much larger in the case of films deposited from toluene (0.1 eV) compared to that deposited from chlorobenzene (0.2 eV). In a related study, Sirringhaus and coworkers have employed KPFM to map the surface potential and the photoinduced surface photovoltage and correlate this with the topography of the polyfluorene blend–based photovoltaic devices.166 The results clearly suggest that an optimization of appropriate size of phase separation percolation of both the electron and hole transporting phases with their respective electrodes are highly essential for improving the efficiency of energy harvesting. One of the significant issues with KPFM is that the experimentally obtained potential profiles do not generally reflect the true profile in the device due to complex coupling between the tip and the sample.167 In the initial stages, it was believed that the tip–sample separation and tip radius are limiting factors of the resolution attained in KPFM measurements.168–172 However, by combining experimental and finite element analysis, Charrier et al. have quantitatively shown that the potential profiles obtained by scanning Kelvin probe microscopy do not purely reflect the electrostatic potential under the tip apex but

Imaging and Force Spectroscopy

273

Figure 22. (a) Schematic showing the microcantilever orthogonal to the channel of the field effect transistor (FET) device. (b) Experimental potential profile with the lever orthogonal to the channel (squares) and modeling for a full 3D tip (solid black line) containing the apex, cone, and lever. The dashed line shows the simulation for a probe consisting only of the cone and apex, the dotted line for a probe consisting of only an apex. (Obtained from Charrier et al.173 with permission from the American Chemical Society.)

are strongly affected by the electrostatic coupling between the entire probe and the entire device, even for small tip–sample separations.173 Figure 22A shows the experimental potential profile in the geometry shown in Figure 22B and three different modeling curves. The solid line is calculated for the full 3D probe consisting of the apex, cone, and lever. Calculations for a probe consisting of the cone + apex (dashed line) and only a single apex (dotted line) are also shown. It can be observed that removing the lever and leaving only the apex + cone, the full potential difference at the electrodes becomes 20% higher. Removing the cone and leaving the apex results in further deviation from the experimentally observed potential profile, clearly underscoring the importance of taking the entire probe and device geometry into account for reliable quantitative potential profiles. 4.2. Conductive Atomic Force Microscopy Conductive atomic force microscopy (c-AFM) enables the simultaneous mapping of the topographical and electrical conductivity of the sample using a conductive AFM tip. This technique involves using the electrically conducting tip as one electrode and a conductive substrate or a metal electrode on the surface of the sample as second electrode. The measurement can be performed by either by applying a constant voltage between the tip and the metal electrode, with simultaneous recording images that can be used as a measure of the local conductivity, or collecting local I-V curves by sweeping the voltage between the tip and the other electrode, which can be mapped. In a different kind of measurement, I-Z curves can be obtained by holding the voltage constant while the Z-piezoelement is moved perpendicular to the sample surface, thereby changing the tip–surface contact area to determine the force where the conductivity saturates. This force is then used to image the conductivity in constant force. Conductive AFM can be operated in two different configurations, namely, horizontal and vertical.174–176 Vertical and horizontal modes are similar in that they both use a conductive AFM tip as an electrode, but the modes differ in the substrate electrode configuration. In the horizontal mode, the material under investigation is deposited on an insulting surface

274

M. E. McConney et al.

and is directly connected to an electrode. On the other hand, in the vertical configuration, the material is deposited on a conductive surface that acts a second electrode. Typically, c-AFM measurements are performed in the contact mode. However, this mode of operation results in the damage of soft polymeric and biological (e.g., DNA) samples. To avoid the potential damage of the sample surface, some groups have adapted an alternate approach that involves in acquiring the topography using a dynamic method such as tapping mode followed by point contact I-V measurements at predefined regions.177 c-AFM has been extensively employed to probe the morphology, conductivity, and carrier mobility of polymer thin-film devices.178–182 One of the extensively studied system is the PEDOT:PSS blend commonly employed as a interface layer between the anode (ITO) and the organic semiconductor layer in various optoelectronic devices. A vertical c-AFM configuration was applied to study the effect of processing conditions (such as annealing, PSS content, solvent treatment) on the vertical charge transport of PEDOT:PSS.183 Though the topography images do not show any significant change with annealing, from the conductive maps it was observed that most of the current passes through the film surface via small conductive hot spots in a relatively insulating matrix. Figure 23 shows the topography and the c-AFM images of PEDOT:PSS film annealed at 140◦ C for different periods of time. The increase in macroscopic conductivity observed following the annealing of PEDOT:PSS films results from an increase in both the number of the conductive hot spots and current carrying capacity of the PEDOT domains observed on the film surface. Apart from obtaining simultaneous topography and conductive maps, vertical configuration of c-AFM has been extensively employed to obtain current density voltage curves (J-V curves), which in turn were used to extract local hole mobilities, using space charge limited current (SCLC) model.180,184 Carrier mobility is extracted by fitting the J-V data to

Figure 23. 1 µm2 area AFM topography and c-AFM images of the PEDOT:PSS films annealed at 140◦ C: (a and e) 0 min, (b and f) 10 min, (c and g) 30 min, (d and h) 100 min. The topography (top row) exhibits very little change through the course of annealing, whereas the c-AFM images (bottom row) clearly show that the number of conductive pathways (bright spots) increase as function of the annealing time. (Obtained from Pingree et al.183 with permission from the American Chemical Society.)

Imaging and Force Spectroscopy

275

the Mott-Gurney law. J =

V2 9 εε0 µ 3 8 L

(14)

where J is the current density, ε is the relative dielectric constant of the active layer, ε0 is the permittivity of free space, µ is the charge carrier mobility, V is the applied voltage, and L is the thickness of the device. However, the technique usually results in carrier mobilities higher compared to that observed in planar electrode configuration as the current spreads out under the AFM tip, enabling a larger space charge limited current density than is expected in the plane-parallel case as shown in Figure 24. Ginger et al. have recently demonstrated that the primary cause of this observation is the fundamental difference in geometry between the two configurations, namely, planar electrodes and the vertical c-AFM.185 Conventional Mott-Gurney law is not applicable for c-AFM measurements because SCLC measurements performed in this geometry deviate from the J ∝ L−3 dependence. Taking the tip sample geometry into account and using finite

Figure 24. Top: Schematics showing the geometry of the c-AFM and planar macroscopic device measurement respective geometry. Current spreading laterally beneath the AFM tip results in a larger space charge limited current density than is expected in the plane-parallel case. Bottom: J-V curves measured using c-AFM (circles) and macroscopic devices (diamonds) on P3HT showing the apparently higher current density in C-AFM measurements. The grey dotted line shows the fit using classical Mott-Gurney law to each of the curves to extract the mobility. c-AFM measurement was made using a 50-nm-diameter platinum-coated tip. (Obtained from Reid et al.185 with permission from the American Chemical Society.)

276

M. E. McConney et al.

element analysis (FEA) simulations they have suggested a semi-empirical equation for the current density as J = αεε0 µ0 e

0.89γ (V /L)1/2

V2 δ L3

L D

1.6±0.1 (15)

which is valid for common tip diameters and sample thicknesses, using a scaling factor based on the ratio of tip diameter, D, to sample thickness, L. This scaling factor enables the extraction of quantitative values of charge carrier mobility from J-V curves collected by c-AFM for samples. 4.3. Magnetic Force Microscopy Magnetic force microscopy (MFM) operates in the noncontact mode in which a tip coated with a ferromagnetic material (such as Ni, Co, Fe) detects the stray magnetostatic field of the magnetic dipoles of the sample. Because the magnetic interactions are long range (similar to the electrostatic interactions), the magnetic imaging is performed in lift mode with a set distance between the probe and surface of typically 20–50 nm. In MFM, during the second line scan the cantilever deflection is monitored and used to create the MFM image. MFM has been widely employed to probe magnetic recording media and imaging and magnetization of Co, Ni, and iron magnetic nanodots.186–190 MFM has also been employed to image and estimate the magnetic moment of magnetotactic bacteria.191 MFM has been employed to study the self-assembly of polyethylene glycol (PEG) and polystyrene-coated Fe2 O3 nanospheres (magnetic nanospheres) under an external magnetic field.192 Sun et al. have described the polymer-mediated assembly of FePt nanoparticles using polyvinylpyrrolidone (PVP) and polyethyleneimine (PEI) polymers.193 The assembly process involved the exchange of oleic acid/oleyl amine around the magnetic nanoparticles with a functional polymer that is previously deposited on a substrate. Figure 25 shows the

Figure 25. (a) AFM topography and (b) MFM image of a three-layer 4-nm Fe58 -Pt42 nanoparticle assembly annealed at 530◦ C. Whereas the AFM reveals the smooth surface topography of the assembly, the MFM image reveals the assembled particles. (Obtained from Sun et al.193 with permission from the American Chemical Society.)

Imaging and Force Spectroscopy

277

topography and the corresponding MFM image of a three-layer 4-nm Fe58 Pt42 assembly treated with a pulsed laser under a perpendicular magnetic field (2.5 kOe). The AFM image shows that the smooth FePt nanoparticle assembly is intact after the laser treatment. The dark spots in MFM image indicate the magnetization pointing to the out of the particle assembly plane. MFM has also been used for mapping the dispersion of carbon nanotubes in a polymer matrix.194 From the MFM phase images, the carbon nanotubes were often found as agglomerates throughout the film. Furthermore, MFM also clearly revealed individual nanotube bundles and areas with high localization of nanotubes, which could not be observed in topographic images. The contrast between the nanotubes and the background was strongly dependent on the distance between the tip and the surface (lift height) with lift heights greater than 15 nm exhibiting diminished contrast.

5. Conclusions This review summarized characterization techniques for soft matter that are based on close proximity probes. Mechanical, thermal, electrical, and magnetic characterization techniques were also presented. The review included a discussion of techniques, calibration procedures, and common pitfalls regarding force spectroscopy measurements and the characterization of elastic modulus and adhesion forces. Examples presented here are not comprehensive but rather selected to demonstrate the most important capabilities of the presented techniques and their applicability to polymeric materials. Key examples of the method are presented to communicate the capabilities and impact that probe-based characterization techniques have had on the mechanical, thermal, magnetic, and electrical characterization of polymeric and composite materials. Main attention is paid on how measurements are conducted from practical viewpoint, how data should be processed, and several examples of corresponding recent results from application of a particular operation mode are briefly presented and discussed.

Acknowledgements The authors thank S. L. Youth for technical assistance and the following agencies for continuous support in SPM studies of soft materials in the SEMA lab: NSF-DMR, NSFCMMI, NSF-CBET, AFOSR, AFRL, ARO, DARPA, and DOE.

References 1. Binnig, G.; Rohrer, H. “Scanning tunneling microscopy,” Helvetica Physica Acta, 1982, 55, 726–735. 2. Binnig, G.; Rohrer, H.; Gerber, C.; Weibel, E. “Tunneling through a controllable vacuum gap,” Applied Physics Letters, 1982, 40, 178–180. 3. Binnig, G.; Rohrer, H.; Gerber, C.; Weibel, E. “Surface studies by scanning tunneling microscopy,” Physical Review Letters, 1982, 49, 57–61. 4. Gerber, C.; Lang, H.P. “How the doors to the nanoworld were opened,” Nature Nanotechnology, 2006, 1, 3–5. 5. Binning, G.; Rohrer, H. “Scanning tunneling microscopy—from birth to adolescence,” Reviews of Modern Physics, 1987, 59, 615–625. 6. Magonov, S. N.; Whangbo, M.-H. Surface Analysis with STM and AFM: Experimental and Theoretical Aspects of Image Analysis; VCH: New York, 1996. 7. Binning, G.; Quate, C. F.; Gerber, C. H. “Atomic force microscope,” Physical Review Letters, 1986, 56, 930–933.

278

M. E. McConney et al.

8. Singamaneni, S.; LeMieux, M. C.; Lang, H. P.; Gerber, C.; Lam, Y.; Zauscher, S.; Datskos, P. G.; Lavrik, N. V.; Jiang, H.; Naik, R. R.; Bunning, T. J.; Tsukruk, V. V. “Bimaterial microcantilevers as a hybrid sensing platform,” Advanced Materials, 2008, 20, 653–680. 9. Raiteri, R.; Nelles, G.; Butt, H.-J.; Knoll, W.; Skladal, P. “Sensing of biological substances based on the bending of microfabricated cantilevers,” Sensors and Actuators B: Chemical, 1999, 61, 213–217. 10. Lang, H. P.; Berger, R.; Andreoli, C.; Brugger, J.; Despont, M.; Vettiger, P.; Gerber, C.; Gimzewski, J. K.; Ramseyer, J.-P.; Meyer, E.; G¨untherodt, H.-J. “Sequential position readout from arrays of micromechanical cantilever sensors,” Applied Physics Letters, 1998, 72, 383–385. 11. Dutta, P.; Chapman, P.; Datskos, P. G.; Sepaniak, M. J. “Characterization of ligandfunctionalized microcantilevers for metal ion sensing,” Analytical Chemistry, 2005, 77, 6601–6608. 12. Fritz, J.; Baller, M. J.; Lang, H.-P.; Rothuizen, H.; Vettiger, P.; Meyer, E.; G¨untherodt, H.-J.; Gerber, C.; Gimzewski, J. K. “Translating biomolecular recognition into nano-mechanics,” Science, 2000, 288, 316–318. 13. Singamaneni, S.; McConney, M. E.; LeMieux, M. C.; Jiang, H.; Enlow, J. O.; Bunning, T. J.; Naik, R. R.; Tsukruk, V. V. “Polymer-silicon flexible structures for fast chemical vapor detection,” Advanced Materials, 2007, 19, 4248–4255. 14. Tsukruk, V. V.; Reneker, D. H. “Scanning probe microscopy of polymeric and organic molecular films: from self-assembled monolayers to composite multilayers,” Polymer, 1995, 36, 1791–1808. 15. Tsukruk, V. V.; Bliznyuk, V. N.; Hazel, J.; Visser, D.; Everson, M. P. “Organic molecular films under shear forces: Fluid and solid langmuir monolayers,” Langmuir, 1996, 12, 4840–4849. 16. Bliznyuk, V. N.; Everson, M. P.; Tsukruk, V. V. “Nanotribological properties of organic boundary lubricants: Langmuir films versus self-assembled monolayers,” Journal of Tribology, 1998, 120, 489–495. 17. Garc´ıa, R.; P´erez, R. “Dynamic atomic force microscopy methods,” Surface Science Reports, 2002, 47, 197–301. 18. Sheiko, S. S. “Imaging of Polymers Using Scanning Force Microscopy: From Superstructures to Individual Molecules”, in New Developments in Polymer Analytics II, Advances in Polymer Science Vol. 151; Schmidt, M., Ed.; Springer-Verlag: Berlin, 2000; 61–174. 19. Martin, Y.; Williams, C. C.; Wickramasinghe, H. K. “Atomic force microscope—force mapping and profiling on a sub 100-Å scale,” Journal of Applied Physics, 1987, 61, 4723–4729. 20. Zhong, Q.; Innis, D.; Kjoller, K.; Elings, V. B. “Fractured polymer/silica fiber surface studied by tapping mode atomic force microscopy,” Surface Science Letters, 1993, 290, L688–L692. 21. Magonov, S. N.; Elings, V.; Whangbo, M.-H. “Phase imaging and stiffness in tapping-mode atomic force microscopy,” Surface Science, 1997, 375, L385–L391. 22. Magonov, S. N.; Cleveland, J.; Elings, V.; Denley, D.; Whangbo, M.-H. “Tapping-mode atomic force microscopy study of the near-surface composition of a styrene-butadiene-styrene triblock copolymer film,” Surface Science, 1997, 389, 201–211. 23. Sarid, D. Oxford Series in Optical and Imaging Sciences: Scanning Force Microscopy; Oxford University Press: New York, 1991. 24. Ratner, B.; Tsukruk, V. V., Eds. Scanning Probe Microscopy of Polymers, ACS Symposium Series Vol. 694; American Chemical Society: Washington, DC, 1998. 25. Tsukruk, V. V.; Spencer, N. D., Eds. Advances in Scanning Probe Microscopy of Polymers, Macromolecular Symposium Vol. 167; Wiley-VCH: Weinheim, 2001. 26. Tsukruk, V. V. “Scanning probe microscopy of polymer surfaces,” Rubber Chemistry and Technology, 1997, 70, 430–467. 27. Magonov, S.; Reneker, D. H. “Characterization of polymer surfaces with atomic force microscopy,” Annual Review of Materials Science, 1997, 27, 175–222. 28. Giessibl, F. J. “Advances in atomic force microscopy,” Reviews of Modern Physics, 2003, 75, 949–983.

Imaging and Force Spectroscopy

279

29. Roiter, Y.; Minko, S. “AFM single molecule experiments at the solid-liquid interface: In situ conformation of adsorbed flexible polyelectrolyte chains,” Journal of the American Chemical Society, 2005, 127, 15688–15689. 30. Julthongpiput, D.; Lin, Y.-H.; Teng, J.; Zubarev, E. R.; Tsukruk, V. V. “Y-Shaped amphiphilic brushes with switchable micellar surface structures,” Journal of the American Chemical Society, 2003, 125, 15912–15921. 31. Julthongpiput, D.; Lin, Y.-H.; Teng, J.; Zubarev, E. R.; Tsukruk, V. V. “Y-Shaped polymer brushes: nanoscale switchable surfaces,” Langmuir, 2003, 19, 7832–7836. 32. Lin, Y.-H.; Teng, J.; Zubarev, E. R.; Shulha, H.; Tsukruk, V. V. “In situ observation of switchable nanoscale topography for Y-Shaped binary brushes in fluids,” Nano Letters, 2005, 5, 491–495. 33. Sheiko, S. S.; Moller, M. “Visualization of macromolecules—a first step to manipulation and controlled response,” Chemistry Reviews, 2001, 101, 4099–4123. 34. Xu, H.; Shirvanyants, D.; Beers, K. L.; Matyjaszewski, K.; Dobrynin, A. V.; Rubinstein, M.; Sheiko, S. S. “Molecular visualization of conformation-triggered flow instability,” Physical Review Letters, 2005, 94, 237801. 35. Scheuring, S.; Seguin, J.; Marco, S.; L´evy, D.; Robert, B.; Rigaud, J.-L. “Nanodissection and high-resolution imaging of the Rhodopseudomonas viridis photosynthetic core complex in native membranes by AFM,” Proceedings of the National Academy of Sciences U. S. A., 2003, 100, 1690–1693. 36. Engel, A.; M¨uller, D. J. “Observing single biomolecules at work with the atomic force microscope,” Nature Structural Biology, 2000, 7, 715–718. 37. Fotiadis, D.; Scheuring, S.; M¨uller, S. A.; Engel, A.; M¨uller, D. J. “Imaging and manipulation of biological structures with the AFM,” Micron, 2002, 33, 385–397. 38. Kumaki, J.; Kawauchi, T.; Yashima, E. “Reptational movements of single synthetic polymer chains on substrate observed by in situ atomic force microscopy,” Macromolecules, 2006, 39, 1209–1215. 39. Kolosov, O.; Yamanaka, K. “Nonlinear detection of ultrasonic vibrations in an atomic force microscope,” Japanese Journal of Applied Physics, 1993, 32, L1095–L1098. 40. Rabe, U.; Arnold, W. “Acoustic microscopy by atomic force microscopy,” Applied Physics Letters, 1994, 64, 1493–1495. 41. Dinelli, F.; Biswas, S. K.; Briggs, G. A. D.; Kolosov, O. V. “Measurements of stiff-material compliance on the nanoscale using ultrasonic force microscopy,” Physical Review B, 2000, 61, 13995–14006. 42. Rabe, U.; Amelio, S.; Kopycinska, M.; Hirsekorn, S.; Kempf, M.; G¨oken, M.; Arnold, W. “Imaging and measurement of local mechanical material properties by atomic force acoustic microscopy,” Surface and Interface Analysis, 2002, 33, 65–70. 43. Shekhawat, G. S.; Dravid, V. P. “Nanoscale imaging of buried structures via scanning near-field ultrasound holography,” Science, 2005, 310, 89–92. 44. Smith, D. P. E. “Limits of force microscopy,” Review of Scientific Instruments, 1995, 66, 3191–3195. 45. Tsukruk, V. V.; Gorbunov, V. V. “Nanomechanical analysis of polymer surfaces,” Probe Microscopy, 2002, 241, 3–4. 46. Ducker, W. A.; Senden, T. J.; Pashley, R. M. “Direct measurement of colloidal forces using an atomic force microscope,” Nature, 1991, 353, 239–241. 47. Butt, H.-J. “Measuring electrostatic, van der Waals, and Hydration forces in electrolyte solutions with an atomic force microscope,” Biophysical Journal, 1991, 60, 1438–1444. 48. McConney, M. E.; Anderson, K. D.; Brott, L. L.; Naik, R. R.; Tsukruk, V. V. “Bio-inspired material approaches to sensing,” Advanced Functional Materials, 2009, 19, 2527–2544. 49. Shulha, H.; Wong, C.; Kaplan, D. L.; Tsukruk, V. V. “Unfolding the multi-length scale domain structure of silk fibroin protein,” Polymer, 2006, 47, 5821–5830. 50. Tsukruk, V. V.; Bliznyuk, V. N. “Adhesive and friction forces between chemically modified silicon and silicon nitride surfaces,” Langmuir, 1998, 14, 446–455.

280

M. E. McConney et al.

51. Zhang, W.; Zhang, X. “Single molecule mechanochemistry of macromolecules,” Progress in Polymer Science, 2003, 28, 1271–1295. 52. K¨uhner, F.; Gaub, E. H. “Modelling cantilever-based force spectroscopy with polymers,” Polymer, 2006, 47, 2555–2563. 53. Zou, S.; Korczagin, I.; Hempenius, M. A.; Sch¨oherr, H.; Vancso, G. J. “Single molecule force spectroscopy of smart poly(ferrocenylsilane) macromolecules: Towards highly controlled redox-driven single chain motors,” Polymer, 2006, 47, 2483–2892. 54. Al-Maawali, S.; Bemis, J. E.; Akhremitchev, B. B.; Liu, H.; Walker, G. C. “Single-molecule afm study of polystyrene grafted at gold surfaces,” The Journal of Adhesion, 2005, 81, 999–1016. 55. Stroh, C.; Wang, H.; Bash, R.; Ashcroft, B.; Nelson, J.; Gruber, H.; Lohr, D.; Lindsay, S. M.; Hinterdorfer, P. “Single molecule recognition imaging microscopy,” Proceedings of the National Academy of Sciences U. S. A., 2004, 101, 12503–12507. 56. Zlatanova, J.; Lindsay, S. M.; Leuba, S. H. “Single molecule force spectroscopy in biology using the atomic force microscope,” Progress in Biophysics and Molecular Biology, 2000, 74, 37–61. 57. Clausen-Schauman, H.; Seitz, M.; Krautbauer, R.; Gaub, H. E. “Force spectroscopy with single bio-molecules,” Current Opinion in Chemical Biology, 2000, 4, 524–530. 58. VanLandingham, M. R.; Villarrubia, J. S.; Guthrie, W. F.; Meyers, G. F. “Nanoindentation of polymers: An overview,” Macromolecular Symposia, 2001, 167, 15–43. 59. Vanlandingham, M. R.; McKnight, S. H.; Palmese, G. R.; Elings, J. R.; Huang, X.; Bogetti, T. A.; Eduljee, R. F.; Gillespie, J. W. “Nanoscale indentation of polymer systems using the atomic force microscope,” The Journal of Adhesion, 1997, 64, 31–59. 60. Li, X.; Bhushan, B. “A review of nanoindentation continuous stiffness measurement technique and its applications,” Materials Characterization, 2002, 48, 11–36. 61. Moeller, G. “AFM nanoindetation of viscoelastic materials with large end-radius probes,” Journal of Polymer Science Part B: Polymer Physics, 2009, 47, 1573–1587. 62. Butt, H.-J.; Cappella, B.; Kappl, M. “Force measurements with the atomic force microscope: technique, interpretation, and applications,” Surface Science Reports, 2005, 59, 1–152. 63. Cappella, B.; Dietler, G. “Force-distance curves by atomic force microscopy,” Surface Science Reports, 1999, 34, 1–104. 64. Cleveland, J. P.; Manne, S.; Bocek, D.; Hansma, P. K. “A nondestructive method for determining the spring constant of cantilevers for scanning force microscopy,” Review of Scientific Instruments, 1993, 64, 403–405. 65. Sader, J. E.; Chon, J. W. M.; Mulvaney, P. “Calibration of rectangular atomic force microscope cantilevers,” Review of Scientific Instruments, 1999, 70, 3967–3969. 66. Sader, J. E. “Parallel beam approximation for v-shaped atomic force microscope cantilevers,” Review of Scientific Instruments, 1995, 66, 4583–4587. 67. Hazel, J. L.; Tsukruk, V. V. “Spring constants of composite ceramic/gold cantilevers for scanning probe microscopy,” Thin Solid Films, 1999, 339, 249–257. 68. Sader, J. E.; Larson, I.; Mulvaney, P.; White, L. R. “Method for the calibration of atomic force microscope cantilevers,” Review of Scientific Instruments, 1995, 66, 3789–3798. 69. Gibson, C. T.; Watson, G. S.; Myhra, S. “Determination of the spring constants of probes for force microscopy/spectroscopy,” Nanotechnology, 1996, 7, 259–262. 70. Hutter, J. L.; Bechhoefer, J. “Calibration of atomic force microscope tips,” Review of Scientific Instruments, 1993, 64, 1868–1873. 71. Hazel, J.; Tsukruk, V. V. “Friction force microscopy measurements: Normal and torsional spring constants for V-shaped cantilevers,” Journal of Tribology, 1998, 120, 814–819. 72. Matei, G. A.; Thoreson, E. J.; Pratt, J. R.; Newell, D. B.; Burnham, N. A. “Precision and accuracy of thermal calibration of atomic force microscopy cantilevers,” Review of Scientific Instruments, 2006, 77, 083703. 73. Gibson, C. T.; Smith, D. A.; Roberts, C. J. “Calibration of silicon atomic force microscope cantilevers,” Nanotechnology, 2005, 16, 234–238.

Imaging and Force Spectroscopy

281

74. Machleidt, T.; K¨astner, R.; Franke, K.-H. “Reconstruction and Geometric Assessment of AFM Tips”, in Nanoscale Calibration Standards and Methods: Dimensional and Related Measurements in Micro- and Nanometer Range; Wilkening, G., Koenders, L., Eds.; Wiley-VCH: Weinheim, 2005; 297–310. 75. Czerkas, S.; Dziomba, T.; Bosse, H. “Comparison of Different Methods of SFM Tip Shape Determination for Various Characterization Structures and Types of Tip”, in Nanoscale Calibration Standards and Methods: Dimensional and Related Measurements in Micro- and Nanometer Range; Wilkening, G., Koenders, L., Eds.; Wiley-VCH: Weinheim, 2005; 311–320. 76. Chizhik, S. A.; Huang, Z.; Gorbunov, V. V.; Myshkin, N. K.; Tsukruk, V. V. “Micro-mechanical properties of elastic polymeric materials as probed by scanning force microscopy,” Langmuir, 1998, 14, 2606–2609. 77. Tsukruk, V. V.; Sidorenko, Y.; Gorbunov, V. V.; Chizhik, S. A. “Surface nanomechanical properties of polymer nanocomposite layers,” Langmuir, 2001, 17, 6715–6719. 78. Stafford, C. M.; Harrison, C.; Beers, K. L.; Karim, A.; Amis, E. J.; Vanlandingham, M. R.; Kim, H.-C.; Volksen, W.; Miller, R. D.; Simony, E. E. “A Buckling-based metrology for measuring the elastic moduli of polymeric thin films,” Nature Materials, 2004, 3, 545–550. 79. Jiang, C.; Singamaneni, S.; Merrick, E.; Tsukruk, V. V. “Complex buckling instability patterns of nanomembranes with encapsulated gold nanoparticle arrays,” Nano Letters, 2006, 6, 2254–2259. 80. Markutsya, S.; Jiang, C.; Pikus, Y.; Tsukruk, V. V. “Free-standing multilayered nanocomposites films as highly sensitive nanomembranes,” Advanced Functional Materials, 2005, 15, 771– 780. 81. Higgins, M. J.; Proksch, R.; Sader, J. E.; Polcik, M.; Mc Endoo, S.; Cleveland, J. P.; Jarvis, S. P. “Noninvasive determination of optical lever sensitivity in atomic force micrscopy,” Review of Scientific Instruments, 2006, 77, 013701. 82. Lin, D. C.; Horkay, F. “Nanomechanics of polymer gels and biological tissues: A critical review of analytical approaches in the hertzian regime and beyond,” Soft Matter, 2008, 4, 669–682. 83. Lin, D. C.; Shreiber, D. I.; Dimitriadis, E. K.; Horkay, F. “Spherical indentation of soft matter beyond the hertzian regime: Numerical validation of hyperelastic models,” Biomechics and Modeling in Mechanobiology, 2009, 8, 345–358. 84. Marti, O.; Hild, S. “Temperature-Dependent Surface Properties of Thin Polystyrene Films Determined by Scanning force Microscopy”, in Microstructure and Microtribology of Polymer Surface ACS Symposium Series 741; Tsukruk, V.V., Wahl, K., Eds.; American Chemical Society: Washington, DC, 2000, pp. 212–226. 85. Aim´e, J. P.; Elkaakour, Z.; Odin, C.; Bouhacina, T.; Michel, D.; Curely, J.; Dautant, A. “Comments on the use of the force mode in atomic force microscopy for polymer films,” Journal of Applied Physics, 1994, 76, 754–762. 86. Choi, T.; Jang, J.-H.; Ullal, C. K.; LeMieux, M. C.; Tsukruk, V. V.; Thomas, E. L. “The elastic properties and plastic behavior of two-dimensional polymer structures fabricated by laser interference lithography,” Advanced Functional Materials, 2006, 16, 1324–1330. 87. Jang, J.-H.; Ullal, C. K.; Choi, T.; LeMieux, M. C.; Tsukruk, V. V.; Thomas, E. L. “3D polymer microframes that exploit length-scale-dependent mechanical behavior,” Advanced Materials, 2006, 18, 2123–2127. 88. Singamaneni, S.; Chang, S.; Jang, J.-H.; Davis, W.; Thomas, E. L.; Tsukruk, V. V. “Mechanical properties of composite polymer microstructures fabricated by interference lithography,” Physical Chemistry Chemical Physics, 2008, 10, 4093–4105. 89. Julthongpiput, D.; LeMieux, M.; Tsukruk, V. V. “Micromechanical properties of glassy and rubbery polymer brush layers as probed by atomic force microscopy,” Polymer, 2003, 44, 4557–4562. 90. Tsukruk, V. V.; Gorbunov, V. V.; Huang, Z.; Chizhik, S. A. “Dynamic microprobing of viscoelastic polymer properties,” Polymer International, 2000, 49, 441–444. 91. Tsukruk, V. V.; Huang, Z. “Micro-thermomechanical properties of heterogeneous polymer films,” Polymer, 2000, 41, 5541–5545.

282

M. E. McConney et al.

92. Luzinov, I.; Julthongpiput, D.; Tsukruk, V. V. “Stability of microdomain morphology in tethered block copolymer monolayers,” Polymer, 2001, 42, 2267–2273. 93. Kovalev, A.; Shulha, H.; LeMieux, M.; Myshkin, N.; Tsukruk, V. V. “Nanomechanical probing of layered nanoscale polymer films with atomic force microscopy,” Journal of Materials Research, 2004, 19, 716–728. 94. Shulha, H.; Kovalev, A.; Myshkin, N.; Tsukruk, V. V. “Some aspects of AFM nanomechancal probing of surface polymer films,” European Polymer Journal, 2004, 40, 949–956. 95. Shulha, H.; Zhai, X.; Tsukruk, V. V. “Molecular stiffness of individual hyperbranched macromolecules at solid surfaces,” Macromolecules, 2003, 36, 2825–2831. 96. Tsukruk, V. V.; Shulha, H.; Zhai, X. “Nanoscale stiffness of individual dendritic molecules and their aggregates,” Applied Physics Letters, 2003, 82, 907–909. 97. Kienberger, F.; Ebner, A.; Gruber, H. J.; Hinterdorfer, P. “Molecular recognition imaging and force spectroscopy of single biomolecules”, Accounts of Chemical Research, 2006, 39, 29–36. 98. Engel, A.; Muller, D. J. “Observing single biomolecules at work with the atomic force microscope,” Nature Structural Biology, 2000, 7, 715–718. 99. Hansma, H. G.; Kim, K. J.; Laney, D. E. “Properties of biomolecules measured from atomic force microscope images: A review,” Journal of Structural Biology, 1997, 119, 99–108. 100. Fotiadis, D.; Scheuring, S.; Muller, S. A.; Engel, A.; Muller, D. J. “Imaging and manipulation of biological structures with the AFM”, Micron, 2002, 33, 385–397. 101. Stolz, M.; Raiteri, R.; Daniels, A. U.; VanLandingham, M. R.; Baschong, W.; Aebi, U. “Dynamic elastic modulus of porcine articular cartilage determined at two different levels of tissue organization by indentation-type atomic force microscopy,” Biophysical Journal, 2004, 86, 3269–3283. 102. Elkin, B. S.; Ilankovan, A.; Morrison, B. “Age dependent regional mechanical properties of the rat hippocampus and cortex,” Journal of Biomedical Engineering, 2010, 132, 011010-1-10. 103. Shafieian, M.; Darcish, K. K.; Stone, J. R. “Changes to the viscoelastic properties of brain tissue after traumatic axonal injury,” Journal of Biomechanics, 2009, 42, 2136–2142. 104. Harmon, M. E.; Kuckling, D.; Frank, C. W. “Photo-crosslinkable pnipaam copolymers. 5. Mechanical properties of hydrogel layers,” Langmuir, 2003, 19, 10660–10665. 105. Braithwaite, G. J. C.; Luckham, P. F. “The simultaneous determination of the forces and viscoelastic properties of adsorbed polymer layers,” Journal of Colloid and Interface Science, 1999, 218, 97–111. 106. Tripathy, S.; Berger, E. J. “Measuring viscoelasticity of soft samples using atomic force microscopy,” Journal of Biomechanical Engineering, 2009, 131, 094507–094513. 107. Engler, A. J.; Sen, S.; Sweeney, H. L.; Discher, D. E. “Matrix elasticity directs stem cell lineage specification,” Cell, 2006, 126, 677–689. 108. McConney, M. E.; Schaber, C. F.; Julian, M. D.; Barth, F. G.; Tsukruk, V. V. “Viscoelastic nanoscale properties of cuticle contribute to the high-pass properties of spider vibration receptor (Cupoennius salei keys),” Journal of the Royal Society Interface, 2007, 4, 1135–1143. 109. McConney, M. E.; Schaber, C. F.; Julian, M. D.; Eberhardt, W. C.; Humphrey, J. A. C.; Barth, F. G.; Tsukruk, V. V. “Surface force spectroscopic point load measurements and viscoelastic modeling of the micromechanical properties of air flow sensitive hairs of a spider (Cupiennius salei),” Journal of the Royal Society Interface, 2009, 6, 681–694. 110. Peleshanko, S.; Julian, M. D.; Ornatska, M.; McConney, M. E.; LeMieux, M. C.; Chen, N.; Tucker, C.; Yang, Y.; Liu, C.; Humphrey, J. A. C.; Tsukruk, V. V. “Hydrogel-encapsulated microfabricated haircells mimicking fish cupula neuromasts,” Advanced Materials, 2007, 19, 2903–2909. 111. Yoshikawa, Y.; Yasuike, T.; Yagi, A.; Yamada, T. “Transverse elasticity of myofibrils of rabbit skeletal muscle studied by atomic force microscopy,” Biochemical and Biophysical Research Communications, 1999, 256, 13–19. 112. Noy, A.; Vezenov, D. V.; Lieber, C. M. “Chemical force spectroscopy,” Annual Review of Materials Science, 1997, 27, 381–421.

Imaging and Force Spectroscopy

283

113. Noy, A. “Interaction at Solid-Fluid Interfaces”, In Nanoscale Structure and Assembly at SolidFluid Interfaces; Liu, X. Y., De Yoreo, J. J., Eds.; Kluwer Academic Publishers: Norwell, MA, 2004; 57–82. 114. Takano, H.; Kenseth, J. R.; Wong, S.-S.; O’Brien, J. C.; Porter, M. D. “Chemical and biochemical analysis using scanning force microscopy,” Chemical Reviews, 1999, 99, 2845–2890. 115. Janshoff, A.; Neitzert, M.; Oberd¨orfer, Y.; Fuchs, H. “Force-spectroscopy of molecular systems—single molecule spectroscopy of polymers and biomolecules,” Angewandte Chemie International Edition, 2000, 39, 3212–3237. 116. Noy, A., Ed. Handbook of Molecular Force Spectroscopy; Springer Science + Business Media, LLC: New York, 2008. 117. Thundat, T.; Zheng, X.-Y.; Chen, G. Y.; Warmack, R. J. “Role of relative humidity in atomic force microscopy imaging,” Surface Science Letters, 1993, 294, L939–L943. 118. Binggeli, M.; Mate, C. M. “Influence of capillary condensation on water on nanotribology studied by force microscopy,” Applied Physics Letters, 1994, 65, 415–417. 119. Fujihira, M.; Aoki, D.; Okabe, Y.; Takano, H.; Hokari, H.; Frommer, J.; Nagatani, Y.; Sakai, F. “Effect of capillary force on friction force microscopy: A scanning hydrophilicity microscope,” Chemistry Letters, 1996, 25, 499–502. 120. Noy, A.; Frisbie, C. D.; Rosznyai, L. F.; Wrighton, M. S.; Lieber, C. M. “Chemical force microscopy: Exploiting chemically modified yips to quantify adhesion, friction and functional group distributions in molecular assemblies,” Journal of the American Chemical Society, 1995, 117, 7943–7951. 121. Zhmud, B. V.; Golub, A. A. “Protolytic equilibria ligands immobilized at rigid matrix surfaces: A theoretical study,” Colloid and Interface Science, 1994, 167, 186–192. 122. Vezenov, D. V.; Noy, A.; Rozsnyai, L. F.; Lieber, C. M. “Force titrations and ionization state sensitive imaging of functional groups in aqueous solutions by chemical force microscopy,” Journal of the American Chemical Society, 1997, 119, 2006–2015. 123. Sch¨onherr, H.; Hruska, Z.; Vancso, J. “Toward high resolution mapping of functional group distribution at surface treated polymers by AFM using modified tips,” Macromolecules, 2000, 33, 4532–4537. 124. Jiang, X.; Ortiz, C.; Hammond, P. T. “Exploring the rules for selective deposition: interactions of model polyamines on acid and oligioethylene oxide surfaces,” Langmuir, 2002, 18, 1131–1143. 125. Barber, A. H.; Cohen, S. R.; Wagner, H. D. “Measurement of carbon nanotube-polymer interfacial strength,” Applied Physics Letters, 2003, 82, 4140–4142. 126. Friddle, R. W.; LeMieux, M. C.; Cicero, G.; Artyukhin, A. B.; Tsukruk, V. V.; Grossman, J. C.; Galli, G.; Noy, A. “Single functional group interactions with individual carbon nanotubes,” Nature Nanotechnology, 2007, 2, 692–697. 127. Rosa-Zeiser, A.; Weilandt, E.; Hild, S.; Marti, O. “The simultaneous measurement of elastic, electrostatic, and adhesive properties by scanning force microscopy: Pulsed-force mode operation,” Measurement Science and Technology, 1997, 8, 1333–1338. 128. Marti, O.; Stifner, T.; Waschipky, H.; Quintus, M.; Hild, S. “Scanning probe microscopy of heterogenous polymers,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, 1999, 154, 65–73. 129. Krotil, H.-U.; Stifter, T.; Waschipky, H.; Weishaupt, K.; Hild, S.; Marti, O. “Pulsed force mode: A new method for the investigation of surface properties,” Surface and Interface Analysis, 1999, 27, 336–340. 130. Carpick, R. W.; Salmeron, M. “Scratching the surface: Fundamental investigations of tribology with atomic force microscopy,” Chemical Reviews, 1997, 97, 1163–1194. 131. Bhushan, B. Nanotribology and Nanomechanics: An Introduction, 2nd ed.; Springer-Verlag: Berlin, 2008. 132. Cannara, R. J.; Brukman, M. J.; Cimatu, K.; Sumant, A. V.; Baldelli, S.; Carpick, R. W. “Nanoscale friction varied by isotopic shifting of surface vibrational frequencies,” Science, 2007, 318, 780–783.

284

M. E. McConney et al.

133. Tsukruk, V. V. “Molecular lubricants and glues for micro- and nanodevices,” Advanced Materials, 2001, 13, 95–108. 134. Tsukruk, V. V.; Everson, M. P.; Lander, L. M.; Brittain, W. J. “Nanotribological properties of composite molecular films: C60 anchored to a self-assembled monolayer,” Langmuir, 1996, 12, 3905–3911. 135. Tsukruk, V. V.; Bliznyuk, V. N.; Hazel, J.; Visser, D.; Everson, M. P. “Organic molecular films under shear forces: Fluid and solid langmuir monolayers,” Langmuir, 1996, 12, 4840– 4850. 136. Han, L.; Dean, D.; Ortiz, C.; Grodzinsky, A. J. “Lateral nanomechanics of cartilage aggrecan macromolecules,” Biophysical. Journal, 2007, 92, 1384–1398. 137. Tocha, E.; Sch¨onher, H.; Vancso, G. J. “Surface relaxations of poly(methyl methacrylate) assesed by friction force microscopy on the nanoscale,” Soft Matter, 2009, 5, 1489–1495. 138. Majumdar, A. “Scanning thermal microscopy,” Annual Review of Materials Science, 1999, 29, 505–585. 139. Dinwiddie, R. B.; Pylkki, R. J.; West, P. E. “Thermal Conductivity Contrast Imaging with a Scanning Thermal Microscope”, in Thermal Conductivity 22; Tong, T. W., Ed.; Technomics Inc.: Lancaster, PA, 1994; 668–677. 140. Balk, L. J.; Maywald, M.; Pylkki, R. J. “Nanoscopic Detection of the Thermal Conductivity of Compound Semiconductor Materials by Enhanced Scanning Thermal Microscopy”, in 9th Conference on Microscopy of Semiconducting Materials (Inst. Phys. Conf. Ser. 146); Cullis, A. G., Staton-Bevan, A. E., Eds., IOP Publishing Ltd.: Bristol, England, 1995; 655–658. 141. Hammiche, A.; Hourston, D. J.; Pollock, H. M.; Reading, M.; Song, M. “Scanning thermal microscopy: Subsurface imaging, thermal mapping of polymer blends, and localized calorimetry,” Journal of Vacuum Science and Technology B, 1996, 14, 1486–1491. 142. Gorbunov, V. V.; Fuchigami, N.; Tsukruk, V. V. “Microthermal analysis with scanning thermal microscopy. I. Methodology and experimental,” Probe Microscopy, 2000, 2, 53–63. 143. Hammiche, A.; Pollock, H. M.; Song, M.; Hourston, D. J. “Sub-surface imaging by scanning thermal microscopy,” Meas. Sci. Technol., 1996, 7, 142–150. 144. Pollock, H. M.; Hammiche, A.; Song, M.; Hourston, D. J.; Reading, M. “Interfaces in polymeric systems as studied by C.A.S.M.—a new combination of localised calorimetric analysis with scanning microscopy,” The Journal of Adhesion, 1998, 67, 217–234. 145. Price, D. M.; Reading, M.; Hammiche, A.; Pollock, H. M. “Micro-thermal analysis: Scanning thermal microscopy and localized thermal analysis,” International Journal of Pharmaceutics, 1999, 192, 85–96. 146. Price, D. M.; Reading, M.; Hammiche, A.; Pollock, H. M. “New adventures in thermal anaylsis,” Journal of Thermal and Analytical Calorimetry, 2000, 60, 723–733. 147. Pollock, H. M.; Hammiche, A. “Micro-thermal analysis: Techniques and applications,” Journal of Physics D: Applied Physics, 2001, 34, R23–R53. 148. Gorbunov, V. V.; Fuchigami, N.; Hazel, J. L.; Tsukruk, V. V. “Probing surface micro-thermal properties by scanning thermal microscopy,” Langmuir, 1999, 15, 8340–8343. 149. King, W. P.; Saxena, S.; Nelson, B. A.; Weeks, B. L.; Pitchimani, R. “Nanoscale thermal analysis of an energetic material,” Nano Letters, 2006, 6, 2145–2149. 150. Gorbunov, V. V.; Fuchigami, N.; Tsukruk, V. V. “Microthermal analysis with scanning thermal microscopy. II. Calibration, modeling, and interpretation,” Probe Microscopy, 2000, 2, 65–75. 151. Tsukruk, V.; Gorbunov, V.; Fuchigami, N. “Microthermal analysis of polymeric materials,” Thermochimica Acta, 2003, 395, 151–158. 152. Fischer, H. “Quantitative determination of heat conductivities by scanning thermal microscopy” Thermochimica Acta, 2005, 425, 69–74. 153. Lawson, N. S.; Ion, R. H.; Pollock, H. M.; Hourston, D. J.; Reading, M. “Characterising polymer surfaces—nanoindentation, surface force data, calorimetric microscopy,” Physica Scripta, 1994, T55, 199–205. 154. Gorbunov, V. V.; Fuchigami, N.; Tsukruk, V. V. “Microthermal probing of ultrathin polymer films,” High Performance Polymers, 2000, 12, 603–610.

Imaging and Force Spectroscopy

285

155. Chui, B. W.; Stowe, T. D.; Ju, Y. S.; Goodson, K. E.; Kenny, T. W.; Mamin, H. J.; Terris, B. D.; Ried, R. P.; Rugar, D. “Low-stiffness silicon cantilevers with integrated heaters and piezoresistive sensors for high-density AFM thermomechanical data storage,” Journal of Microelectromechanical Systems, 1998, 7, 69–78. 156. Jones, K. J.; Kinshott, I.; Reading, M.; Lacey, A. A.; Nikolopoulos, C.; Pollock, H. M. “The origin and interpretation of the signals of MTDSC,” Thermochimica Acta, 1997, 305, 187–199. 157. H¨aßler, R.; zur M¨uehlen, E. “An introduction to µTATMand its application to the study of interfaces,” Thermochimica Acta, 2000, 361, 113–120. 158. Blaine, R. L.; Slough, C. G.; Price, D. M. “Microthermal Analysis Calibration, Repeatability and Reproducibility”, in Proceedings of the 27th NATAS, Omnipress: Madison, WI, 1999; 691–696. 159. Meyers, G.; Pastzor, A.; Kjoller, K. “Localized thermal analysis: From the micro- to the nanoscale,” American Laboratory, 2007, 39, 9–14. 160. Moon, I.; Androsch, R.; Chen, W.; Wunderlich, B. “The principles of micro-thermal analysis and its application to the study of macromolecules,” Journal of Thermal Analysis and Calorimetry, 2000, 59, 187–203. 161. Forrest, J. A.; Dalnoki-Veress, K. “The glass transition in thin polymer films,” Advances in Colloid and Interface Science, 2001, 94, 167–196. 162. Alcoutlabi, M.; McKenna, G. B. “Effects of confinement on material behavior at the nanometre size scale,” Journal of Physics: Condensed Matter, 2005, 17, R461–R524. 163. Menczel, J. D.; Prime, R. B., Eds. Thermal Analysis of Polymers: Fundamentals and Applications; John Wiley & Sons: Hoboken, NJ, 2009. 164. Pruton, M. Introduction to Surface Physics; Oxford University Press: New York, 1994. 165. Hoppe, H.; Glatzel, T.; Niggemann, M.; Hinsch, A.; Lux-Steiner, M. C.; Sariciftci, N. S. “Kelvin probe force microscopy study on conjugated polymer/fullerene bulk heterojunction organic solar cells,” Nano Letters, 2005, 5, 269–274. 166. Chiesa, M.; B¨urgi, L.; Kim, J.-S.; Shikler, R.; Friend, R. H.; Sirringhaus, H. “Correlation between surface photovoltage and blend morphology in polyfluorene-based photodiodes,” Nano Letters, 2005, 5, 559–563. 167. Zerweck, U.; Loppacher, C.; Otto, T.; Grafstr¨om, S.; Eng, L. M. “Accuracy and resolution limits of kelvin probe force microscopy,” Physical Review B, 2005, 71, 125424. 168. Koley, G.; Spencer, G.; Bhangale, H. R. “Cantilever effects on the measurement of electrostatic potentials by scanning kelvin probe microscopy,” Applied Physics Letters, 2001, 79, 545–547. 169. Gil, A.; Colchero G´omez-Herrero, J.; Bar´o, A. M. “Electrostatic force gradient signal: Resolution enhancement in electrostatic force microscopy and improved kelvin probe microscopy,” Nanotechnology, 2003, 14, 332–340. 170. Takahashi, T.; Ono, S. “Tip-to-sample distance dependence of an electrostatic force in KFM measurements,” Ultramicroscopy, 2004, 100, 287–292. 171. Sacha, G. M.; Verdaguer, A.; Martinez, J.; S´aenz, J. J.; Ogletree, D. F.; Salmeron, M. “Effective tip radius in electrostatic force microscopy,” Applied Physics Letters, 2005, 86, 123101. 172. Argento, C.; French, R. H. “Parametric tip model and force-distance relation for hamaker constant determination from atomic force microscopy,” Journal of Applied Physics, 1996, 80, 6081–6090. 173. Charrier, D. S. H.; Kemerink, M.; Smalbrugge, B. E.; de Vries, T.; Janssen, R. A. J. “Real versus measured surface potentials in scanning kelvin probe microscopy,” ACS Nano, 2008, 2, 622–626. 174. Dai, H.; Wong, E.; Lieber, C. M. “Probing electrical transport in nanomaterials: Conductivity of individual carbon nanotubes,” Science, 1996, 272, 523–526. 175. Klein, D. L.; McEuen, P. L. “Conducting atomic force microscopy of alkane layers on graphite,” Applied Physics Letters, 1995, 66, 2478–2480. 176. O’Shea, S. J.; Atta, R. M.; Murrell, M. P.; Welland, M. E. “Conducting atomic force microscopy study of silicon dioxide breakdown,” Journal of Vacuum Science and Technology B, 1995, 13, 1945–1952.

286

M. E. McConney et al.

177. Kelley, T. W.; Granstrom, E. L.; Frisbie, C. D. “Conducting probe atomic force microscopy: a characterization tool for molecular electronics,” Advanced Materials, 1999, 11, 261–264. 178. Pingree, L. S. C.; Hersam, M. C.; Kern, M. M.; Scott, B. J.; Marks, T. J. “Spatially-resolved electroluminescence of operating organic light-emitting diodes using conductive atomic force microscopy,” Applied Physics Letters, 2004, 85, 344–346. 179. Ionescu-Zanetti, C.; Mechler, A.; Carter, S. A.; Lal, R. “Semiconductive polymer blends: correlating structure with transport properties at the nanoscale,” Advanced Materials, 2004, 16, 385–389. 180. Douheret, O.; Lutsen, L.; Swinnen, A.; Breselge, M.; Vandewal, K.; Goris, L.; Manca, J. “Nanoscale electrical characterization of organic photovoltaic blends by conductive atomic force microscopy,” Applied Physics Letters, 2006, 89, 032107. 181. Coffey, D. C.; Ginger, D. S. “Patterning phase separation in polymer films with Dip-Pen nanolithography,” Journal of the American Chemical Society, 2005, 127, 4564–4565. 182. Coffey, D. C.; Reid, O. G.; Rodovsky, D. B.; Bartholomew, G. P.; Ginger, D. S. “Mapping local photocurrents in polymer/fullerene solar cells with photoconductive atomic force microscopy,” Nano Letters, 2007, 7, 738–744. 183. Pingree, L. S. C.; MacLeod, B. A.; Ginger, D. S. “The changing face of PEDOT:PSS films: Substrate, bias, and processing effects on vertical charge transport,” Journal of Physical Chemistry C, 2008, 112, 7922–7927. 184. Frenette, M.; MacLean, P. D.; Barclay, L. R. C.; Scaiano, J. C. “Radically different antioxidants: Thermally generated carbon-centered radicals as chain-breaking anti-oxidants,” Journal of the American Chemical Society, 2006, 128, 16532–16539. 185. Reid, O. G.; Munechika, K.; Ginger, D. S. “Space charge limited current measurements on conjugated polymer films using conductive atomic force microscopy,” Nano Letters, 2008, 8, 1602–1609. 186. Porthun, S.; Abelmann, L.; Lodder, C. “Magnetic force microscopy of thin film media for high-density magnetic recording,” Journal of Magnetism and Magnetic Materials, 1998, 182, 238–273. 187. Folks, L.; Woodward, R. C. “The use of MFM for investigating domain structures in modern permanent magnet materials,” Journal of Magnetism and Magnetic Materials, 1998, 190, 28–41. 188. Kleiber, M.; K¨ummerlen, F.; L¨ohndorf, M.; Wadas, A.; Wiesendanger, R. “Magnetization switching of submicrometer Co dots induced by a magnetic force microscope tip,” Physical Review B, 1998, 58, 5563–5567. 189. Gider, S.; Shi, J.; Awschalom, D. D.; Hopkins, P. F.; Campman, K. F.; Gossard, A. C.; Kent, A. D.; von Molnar, S. “Imaging and magnetometry of switching in nanometer-scale iron particles,” Applied Physics Letters, 1996, 69, 3269–3271. 190. Bliznyuk, V.; Singamaneni, S.; Sahoo, S.; Polisetty, S.; He, X.; Binek, C. “Self-assembly of magnetic ni nanoparticles into 1D arrays with antiferromagnetic order,” Nanotechnology, 2009, 20, 105606. 191. Proksch, R. B.; Schaffer, T. E.; Moskowitz, B. M.; Dahlberg, E. D.; Bazylinski, D. A.; Frankel, R. B. “MFM of the submicron magnetic assembly in magnetotactic bacterium,” Applied Physics Letters, 1995, 66, 2582–2584. 192. Yan, L.; Shen, S.; Li, W.; Wang, X. “Dipolar chains and 2D aligned stripes of polymer-coated magnetic iron colloid,” Journal of Applied Polymer Science, 2006, 101, 4211–4215. 193. Sun, S.; Anders, S.; Hamann, H. F.; Thiele, J.-U.; Baglin, J. E. E.; Thomson, T.; Fullerton, E. E.; Murray, C. B.; Terris, B. D. “Polymer mediated self-assembly of magnetic nanoparticles,” Journal of the American Chemical Society, 2002, 124, 2884–2885. 194. Lillehei, P. T.; Park, C.; Rouse, J. H.; Siochi, E. J. “Imaging carbon nanotubes in high performance polymer composites via magnetic force microscopy,” Nano Letters, 2002, 2, 827–829.

Polymer Reviews, 50:287–320, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.493254

Cryogenic Transmission Electron Microscopy for Direct Observation of Polymer and Small-Molecule Materials and Structures in Solution SHENG ZHONG AND DARRIN J. POCHAN Department of Materials Science and Engineering and Delaware Biotechnology Institute, University of Delaware, Newark, Delaware The application of cryogenic transmission electron microscopy (cryo-TEM) in the study of molecular self-assembly of amphiphilic macromolecules, lipid/surfactants, peptides, and other hybrid material systems is quickly growing in popularity as a standard characterization technique. Cryo-TEM allows the direct visualization of nanostructures and microstructures embedded in a thin film of vitrified solvent at liquid nitrogen temperature. This direct observation technique provides precise measurements on particle size and shape on the nanoscale and can serve as a complementary method for unambiguous interpretation of structural information obtained by model-dependent X-ray and neutron scattering and light scattering. This review covers research results over the past several years using cryo-TEM to image interesting solution-state structures in molecular and materials research areas except biology where cryo-TEM has been extensively used to image biological assemblies and particles such as membranes and viruses, respectively. Some results that discuss particularly effective use of cryo-TEM for characterization of solute-state materials are highlighted. Keywords cryo-TEM, self-assembly, nanostructure, micelles, vesicles, solution

1. Introduction Research in material science and engineering, chemical engineering, bioengineering, and chemistry has advanced to where it is possible to make complex nanostructures by simply designing the self-assembly of natural and synthetic molecules. Cryogenic transmission electron microscopy (cryo-TEM) has become a critical technique in the characterization of assembled nanostructures as it can directly reveal the size and shape of complex structures from nanometer to micrometer length scales in the solution state. Examples include toroids,1 helices,2,3 multicompartment micelles,4–6 sponge phases,7 and cubosomes8 that are formed through the self-assembly of polymers, lipids and surfactants, peptides and DNA, as well as hybrid nanoparticle mixtures. It is nearly impossible to unambiguously characterize these complex structures via scattering techniques. Importantly, cryo-TEM has been effective in identification of new morphologies and phases in the solution state. Another reason for the growing importance of cryo-TEM in soft matter is the need to characterize the interplay between different objects in solution, because it could directly impact theoretical and Received July 2, 2009; accepted December 16, 2009. Address correspondence to Darrin J. Pochan, Department of Materials Science and Engineering, University of Delaware, 201 DuPont Hall, Newark, DE 19716. E-mail: [email protected]

287

288

S. Zhong and D. J. Pochan

practical considerations of new systems. For example, if possible it is critical to visualize gel network components and nanostructure in order to more accurately and precisely interpret rheological data.9–11 All these areas require the investigation of solute-state nanostructures and microstructures in as artifact free conditions that could be made possible using cryoTEM techniques. Conventional transmission electron microscopy (TEM) or atomic force microscopy (AFM) could possibly introduce artifacts during sample preparation. Sample preparation of these techniques typically involves solution evaporation or chemical fixation. These processes may cause distortion in the arrangement of subunits and structural deviation due to the collapse of swollen phases. Moreover, heavy metal staining has been widely used in sample preparation for those materials only containing low-Z elements in order to enhance the contrast. Distinguishing real, experimental structure among drying and staining artifacts demands experience and is not always successful. Structural modification during sample preparation is also a serious concern in AFM because soft structures are in contact with a substrate. Solute-state AFM can only rule out the effects of solution evaporation but not the interaction between materials/structures of interest and a substrate. Often, the resolution of the image is compromised. Scattering techniques such as neutron, X-ray, and light scattering are also powerful methods for probing the global structure of assembled molecules in solution. Successful interpretation of scattering data relies on the modeling of the scattering data. Cryo-TEM should be considered as a complement to scattering techniques because successful, real-space imaging of structure in solution via cryo-TEM could improve the modeling of scattering data. The subsequent clear matching of cryoTEM determined structure with scattering data is just as illuminating as a clear discrepancy because discrepancies between the locally observed, in situ structure via cryo-TEM and the globally observed structure via scattering allow one to discern the polydispersity in size and shape of assembled structure. Another important application of cryo-TEM is its ability to allow characterization of intermediate structures at different stages of a kinetic pathway that can be easily trapped during vitrification but are difficult to probe precisely via scattering when one is averaging information globally throughout an entire sample/beam path. In addition, the intermediate states of structure evolution can contain very complicated structural features that are not always suitable for description from existing models. Cryo-TEM can provide an in situ characterization of the structure trapped in vitrified solvents in a scale ranging from nanometers to almost micrometers. Moreover, the sample preparation procedure is carried out in an enclosed vitrification chamber such as FEI’s Vitrobot (FEI Worldwide Corporate, Oregon, USA), in which temperature and humidity can be conveniently controlled. This design for cryo-TEM is unique in that it can preserve structures that are vulnerable to assembly environment, such as solution composition, pH, ionic strength, temperature, and even mechanical operations. Preparation of a cryo-TEM sample includes three steps: (1) applying a certain amount of solution onto a TEM sample grid with small perforations, usually present due to the presence of a lacey carbon thin film supported by the TEM sample grid; (2) blotting excessive solvent in the droplet using solvent-absorbing filter paper; and (3) plunging the ultrathin film of solution supported on the grid into a low-temperature reservoir of hydrocarbon condensed to liquid nitrogen temperature. There are two ways to deposit solutions onto support grids. One way is to use pipettes with control of droplet volume and the other way is to dip the grid into the solution or viscous gel directly for desired film thickness. The latter one is often used for solutions or gels with high viscosity. Blotting is vital to form a liquid film with a thickness (usually less than 300 nm12) that is thin enough to allow enough electron beam through for imaging but is not too thin to exclude materials/structures from the film. Many condensed gases are

Cryogenic Transmission Electron Microscopy

289

used as low-temperature reservoirs, including liquid ethane at its freezing point (most used), liquid nitrogen (used for specific solvents such as toluene,13 oleic acid,14 tetrahydrofuran,5 and ionic liquids15–17), and liquid propane18 (less popular). Cryogen reservoirs should be immiscible to sample solvents and be able to absorb significant energy during vitrification so that solvents can be vitrified in the amorphous state in a short time. The higher the cooling rate determined by the cryogen, the faster and more effective the vitrification. Liquid ethane, the most popular cryogen, which can provide a cooling rate of 105 K s−1, is able to vitrify the solvent in a very short time.19 A relatively low cooling rate will result in crystallization of the solvent. Crystallization of the solvent molecules will cause Bragg scattering of electrons, resulting in failure in imaging.12 Fast liquid vitrification enables the preservation of the structures in their original form. In this review, we will cover the cryo-TEM studies in the assembly of not only polymers but lipid/surfactants, peptides, and nanoparticles in solution. The systems of lipid/surfactants are presented first because this is the area in which most work has been done over the past several years, although the area of polymers is growing rapidly. The knowledge of self-assembly of lipid/surfactant systems is also valuable in the understanding of assembly phenomena in other systems.

2. Lipids and Surfactants 2.1. Vesicles and Their Applications Cryo-TEM has been most successfully and frequently applied toward the characterization of the nanostructures in lipid/surfactant systems. Vesicles are the most extensively studied assembled structures formed by lipid/surfactants due to their wide applications in the pharmaceutical industry, cosmetics, and personal care products.20,21 Using cryo-TEM combined with scattering experiments, one can accurately characterize the size and shape of vesicles at the nanoscale range. Many assembly factors are used to control the self-association of lipid/surfactants in order to obtain different, final assembled structures. The type and concentration of lipid/surfactants22–29 strongly affects the size of vesicles formed. For example, Yoshimura et al.23 have obtained uniform vesicles with diameter of 7 nm by replacing a conventional surfactant by an amino acid–based Gemini surfactant that had two pairs of hydrophobic–hydrophilic groups covalently connected by a spacer. Mixing different surfactants or surfactant/polymer, such as catanionic systems consisting of oppositely charged surfactants29–37 and surfactants/protein hybrids,38–40 can create stable, functional vesicles of narrow size distribution. For instance, Franses and colleagues40 have demonstrated that when a lipid, dipalmitoylphosphatidylcholine (DPPC), was mixed with a protein, fibrinogen (FB), in two different buffer solutions, stable particles with size of 200–300 nm of lipid/protein complexes were formed in the buffer solution in which FB was stable and precipitates of lipid/protein complexes were found when FB was not stable. These aggregates were the results of the interactions of the lipid and protein that are important in biological applications. Lipophilic photoluminescent agents have been commonly utilized in biological studies because they can be embedded into hydrophobic, biological systems.18 Lo et al.18 synthesized a series of luminescent organoiridium polypyridine complexes and these agents were incorporated in vesicles and exhibited shorter emission wavelengths and longer emission lifetime than those of the samples in fluid solutions. Morphological control over vesicles is also clearly subject to conditions of the bulk solution, such as salt type and concentration,35,41–43 counterions,44–46 metal ions complexes,47

290

S. Zhong and D. J. Pochan

Figure 1. (a) 0.27% Extruded DODAC dispersion obtained by dilution with a hyperosmotic CaCl2 solution at 23◦ C.42 (permission needed from Langmuir ACS). (b) 1.0 mM DODAB in extruded vesicles prepared with d = 100 nm.22 Arrow points to a lens-like vesicle (Reproduced with permission from Elsevier).

temperature,42,48,49 and pH50–52 as well as sample preparation method.22,53–55 Generally, applying these factors in one system can generate interesting results. For example, extruded vesicles are presented in Figure 1A,42 where vesicles were originally formed by dispersing dioctadecyldimethylammonium chloride (DODAC) in a hyperosmotic CaCl2 solution at 23◦ C.42 This shape distortion observed both above and below the gel-to-liquid crystalline transition temperature was due to the osmotic stress similar to the osmoregulation mechanism popular in biological systems where the cells sense and respond to changes in the osmotic environment. Interestingly, similar angular and lens-like vesicles have been made through a totally different strategy, as shown in Figure 1B.22 In this system, cationic vesicles of dioctadecyldimethylammonium chloride and bromide (DODAC and DODAB, respectively) in aqueous solution were prepared via the extrusion method. In this case, the external mechanical force22 arising from forcing vesicles to pass through an extrusion filter of a certain diameter (d) was responsible for vesicle deformation. Vesicles grew larger as filter diameter (d) increased. Aside from the pore size, addition of monovalent sodium chloride/bromide led to expansion of vesicle size due to the fact that binding of Cl−/Br− on a positively charged lipid head led to a decrease in the repulsion between lipid molecules and, thus, a decrease in the interfacial curvature. Other sample preparation procedures, like dissolution in CO2 -containing medium53 or sonication and freezing–thawing,54 have been proven to be effective in managing the morphology of vesicles. Using vesicles for drug delivery requires the system to be physiologically benign and robust enough to endure prolonged circulation times in vivo before releasing drugs at the targeted site. However, vesicular aggregations of surfactants tend to fuse or close upon themselves during circulation and are easily cleared from the body. In order to increase the stability of vesicles and their circulation time, S¨oderman and colleagues56,57 synthesized a series of poly(ethylene glycol) (PEG)-conjugated lipid PEG-12-acyloxystearates as a novel class of pharmaceutical solubilizers. PEG is a biologically inert polymer. It is well

Cryogenic Transmission Electron Microscopy

291

known that vesicles/micelles with a PEG-modified surface have an extended blood circulation time because they are protected by PEG from being attacked by opsonins, proteins found in the reticuloendothelial system.58 Their work also showed that hemolytic activity could be hampered with addition of PEG-conjugated surfactants. The circulation time was predictably increased when PEG59 or PEG-conjugated lipids60–62 was mixed within surfactant layer of vesicles. Cryo-TEM revealed, however, that vesicular structures would transform to wormlike micelles,57,61 elongated vesicles,56,59 or a mixture60 with additives. Therefore, it is important in future studies to establish a PEG-protected liposome system of uniform size and shape. Instead of using PEG protection, robust vesicles were achieved via polymerization of terminal, unsaturated hydrocarbon functionalities on the alkyl tail of assembled surfactants.63,64 In this system, the hydrophobic parts were buried within the formed bilayer. When using PEG-coated or covalently locked vesicles for pharmaceutical application, hampered drug release due to the surface/layer modification is always a concern. Although ultrasound or microwave treatment has proved to be an effective way to trigger drug release,65 self-induced release systems are advantageous when the ultimate application is targeted for in vivo delivery. Romberg et al.66 have designed a poly(amino acid)-lipid conjugate exhibiting a prolonged liposome circulation time. Furthermore, the hydrophilic poly(amino acid) was intentionally selected and could be triggered at the specific target site. Cryo-TEM has been applied in other systems effectively to show that liposome capsulated drugs67 and could be ruptured by melittin68 (a peptide well studied in its interaction with liposome) or a cluster compound.69 Another important liposome application is gene delivery. DNA maintains extended conformations in aqueous solution due to its anionic, polyelectrolyte character. Cationic lipids in the form of vesicles in aqueous solution have the ability to complex strongly with DNA to form lipid-DNA complexes called lipoplexes.7,70–77 For medical applications, lipoplexes themselves must be nontoxic under the physiological environment and ideally should not induce any immune responses. However, many lipids, especially catanionic species, are often toxic. Rosa et al.70 reported the formation of catanionic vesicles composed of excess cationic surfactant combined with an anionic surfactant. The cationic amphiphile was amino-derivative arginine-N-lauroyl amide dihydrochloride (ALA) and the anionic one was sodium cetylsulfate. Representative cryo-TEM images for catanionic vesicles before and after binding with DNA are shown in Figure 2.70 At low charge ratio (R) of DNA to net charge of catanionic vesicles (<0.4), unperturbed, small vesicular complexes were seen (Figure 2A). When R = 0.4, some of the assemblies started collapsing, resulting in a lamellar structure (Figure 2B). The d-spacing of lamella was measured directly from cryo-TEM micrographs and revealed that the repeat distance was increased from 4.7 to 5.8 nm as sodium octylsulfate was replaced by sodium cetylsulfate. Amino acid–based cationic surfactants are proving to be a promising candidate for DNA encapsulation, able to exhibit a strong association with DNA while remaining physiologically benign.71 Other systems, such as cationic–zwitterionic lipidsomes,72,73 multivalent cations,74 bioassemblies with ligands for specific cell targeting,75 and dendritic lipids,76 have been examined for their complexation with DNA. Cryo-TEM micrographs across these disparate systems always reveal some type of lamellar bundles. Others have been exploring complexation with RNA. For example, Pitard’s group made an siRNA delivery system by using lipidic aminoglycoside derivatives that complex with RNA, as characterized by cryo-TEM,78 indicating the universal electrostatic association effect in DNA/RNA delivery systems. New spongelike and cubic phases for lipid/DNA complexes were observed by Talmon and coworkers. In this system, a redox-active lipid formed lipoplexes in which

292

S. Zhong and D. J. Pochan

Figure 2. Cryo-TEM micrographs of the supernatant for the mixture of vesicles and DNA at R = 0.4. (a) Unperturbed DNA-loaded vesicles and (b) perturbed DNA-loaded complexes.70 (Reproduced with permission from Elsevier).

morphology depended on the lipid oxidation state. Regular, multilamellar nanostructures were found when lipids were in a reduced state. These lamellar structures transformed into irregular morphologies accompanied with individual spongelike and cubic phases when lipids were oxidized.7

2.2. Wormlike Micelles and Networks Wormlike micelles and elongated micellar networks also are extensively studied by cryoTEM. Unambiguous visualization of networks in cryo-TEM is critical for the identification of the topological transitions of micelles and networks especially when the structure changes correspond to a change in mechanical properties. One nice example is a study on a viscoelastic lipid aqueous solution conducted by Danino and coworkers.85 Cationic surfactants in this study underwent a spherical micelle–wormlike micelle-interconnected network phase transition with an increase in salt concentration. Starting from a fluid solution only having spherical micelles (Figure 3A), the viscosity of the aqueous solution accordingly increased with the appearance of short wormlike micelles (Figure 3B) and then reached a maximum point due to the micellar entanglement of very long, threadlike micelles (Figure 3C). When very long, wormlike micelles in water were triggered to transform to spherical micelles, the reverse rheological response from a gel to a more typical viscous response could occur.80 The stiffness of the wormlike micelle hydrogel could be enhanced, other than by an increase in the length of wormlike micelles, through the addition of vesicles to a hydrogel network in which the fibers were able to associate with the vesicles due to hydrophobic association.81 The rheological properties of cationic surfactant systems mixed with counterions are also subject to charge association/dissociation.82 A tightly packed, interconnected structure could be triggered upon cooling, involving a phase transition from prolate micelles to final rodlike aggregates.83 In a different bolaamphiphile system, formation of wormlike micelles also was directly related to temperature as confirmed by cryo-TEM.84

Cryogenic Transmission Electron Microscopy

293

Figure 3. 0.15 M hexadecyltrimethylammonium bromide (CTAB) in aqueous solution at 303 K. (a) Absence of NaNO3 ; (b) presence of 0.02 M NaNO3 ; and (c) presence of 0.04 M NaNO3 .79 (Reproduced with permission from Elsevier).

Danino and coworkers85 used cryo-TEM to show that wormlike micelles formed by erucyl bis(hydroxyethyl)methyl ammonium chloride and sodium hydroxynaphthoate in aqueous solution exhibited the tendency to align when induced by a shear force. This alignment was stabilized by intermicellar attractive interactions (such as π -π and cation-π ). Berti et al.86 have reported the observation of helical fibers formed by nucleobase-derived lipids that merged the responsiveness of conventional ionic assemblies to the additional tunability provided by molecular recognition between complementary bases due to their DNA-like polar head groups.

2.3. Phase Transformation of Lipid/Surfactant Assemblies Cryo-TEM has improved the study of the kinetics of structure evolution due to the ability of the user to collect data containing both structures present during structural transitions as well as the coexistence of nanostructures. Analyzing mixtures of morphologically distinctive structures remains a difficult task for scattering methods because the scattering profile of morphological mixtures is difficult for unambiguous modeling. In the lipid/surfactant research arena, certain catanionic systems display rich intermediate structures as captured via cryo-TEM.31,32 Kaler et al. recently reported33 the phase behavior of a catanionic system consisting of cationic surfactant and a hybrid nonionic–anionic surfactant. The originally spherical assembly transformed via uniaxial growth into ribbon-like micelles with either time or variation in molar fractions of these two surfactants, as seen in other systems.87,88 Other papers showed that by adding salts into a catanionic system, a wealth of assembled phases; including disks, ribbons, vesicles,34,35,46 and ladder shape structures,43 could be captured via cryo-TEM. Recently, Talmon’s group89 demonstrated that a polyelectrolyte was able to associate with a mixture of cationic surfactant to lead to the formation of liquid-crystal particles in the solution. In a liquid mesophase, Danino’s group revealed the kinetic route for the formation of a cubic phase by showing a series of representative cryoTEM micrographs of the cubosomes at each transition stage ranging from initial fusion of unilamellar vesicles to final cubic aggregates with regular internal order packing8 as shown in Figure 4.8 Similar, unambiguous cryo-TEM images for cubosomes are also represented in other papers.90,91 Many other reports showed that vesicles can elongate to tubes92–94 or that tubes coexist with spherical micelles.95

294

S. Zhong and D. J. Pochan

Figure 4. Suggested path of the growth of cubosomes, starting from unilamellar vesicles.8 Scale bar is not presented in original graph (reproduced with permission from The American Chemical Society).

2.4. Spherical Micelles Finally, micelles, spherical in particular, have been studied for funtionalization in solution. The applications are wide ranging with examples such as spherical micelle solutions containing Fe(III) complexed with surfactant that contained a peptide head group,96 metallosurfactants,97 fluorescence dye within the lipid micelles,98 spherical aggregations of fluorescent amphicalixarenes,99 micelles for magnetic resonance imaging,100 or medical application.101 Micro-/nano-emulsions, for which cryo-TEM is an excellent method for nanostructure through microstructure characterization, are also used in various applications such as drug delivery systems, cosmetics, and personal care products.102,103

3. Polymers and Amphiphilic Block Copolymers The study in self-assembly of amphiphilic block copolymers has attracted immense attention in the past several decades due to the recognition of merits specific to a polymeric system, such as a low critical micelle concentration, versatility in polymer chemistry, selectivity of solvents, and slow kinetics in polymer chain exchange.12 Therefore, polymeric assembly systems provide an aggregation behavior rich in complexity with regard to possible structures, kinetic pathways, and function. Assembled polymeric microstructures investigated by cryo-TEM include conventional assemblies such as spherical micelles,104–111 vesicles,112–118 cylindrical/wormlike micelles,119–121 and spherical colloids14,122–126 and unconventional structures such as multicompartment micelles,4–6 nanoparticles with internal

Cryogenic Transmission Electron Microscopy

295

microstructures,127 Janus particles,128–130 “spherocylinders,”131 toroids,1,132 helical structures,2,3,133 disks,134 micelle superlattices,17 cubic mesoporous aggregates,135 and dumbbellshaped double spheres.136 Unlike lipids and surfactants, amphiphilic block copolymers are composed of chemically distinctive polymeric blocks with molecular weights of 1–100 kDalton. When they assemble in aqueous solutions, hydrophilic blocks will be highly swollen by the solvent and consequently form a corona covering the dense core resulting from collapse of hydrophobic blocks, keeping the core from contacting the solvent. These characteristics of polymer architecture and assembly behavior bestow polymeric assemblies many advantages for successful cryo-TEM characterization. Firstly, the dense core of collapsed hydrophobic chains can provide stronger mass contrast than that of lipid/surfactants in cryo-TEM images. Secondly, higher mass assemblies are able to endure longer electron beam exposure time without causing any damage and thus allowing the acquisition of high-quality images. Thirdly, the polymeric assemblies are generally larger because they result from the selfassociation of long polymer chains, therefore making it easier to examine the subtle micro/nanostructures at high magnification. 3.1. Multicompartment Micelles and Janus Particles Microdomains formed via the segregation of chemically distinctive chains in one micelle are able to be distinguished in cryo-TEM images, mainly due to the difference in electron density of polymer chains. There are two types of multicompartment micelles. Lodge and Hillmyer’s group5,6,137 has designed a miktoarm star triblock copolymer covalently combining three mutually immiscible polymeric components into one molecule. One block is hydrophilic poly(ethylene oxide) and the other two are hydrophobic polyethylethylene and poly(perfluoropropylene oxide).137 When the polymer was dispersed in water, the unlike hydrophobic blocks were enclosed within the same core but segregated into distinctive nanodomains. In cryo-TEM images, fluorine-containing microdomains were darker because of their higher electron density relative to that of hydrocarbon domains. Layered cylindrical micelles, star micelles, and patchy spherical micelles were determined using cryo-TEM. By applying this strategy, new miktoarm triblock copolymers were introduced and their segregation behaviors in aqueous solution and organo/water mixtures were examined via cryo-TEM.5,6 Another multicompartment micellar system was introduced by Pochan, Wooley and coworkers.4 Instead of using one polymer, two block copolymers sharing the same hydrophilic poly(acrylic acid) (PAA) block but immiscible hydrophobic blocks poly(methyl acrylate)-b-polystyrene (PMA-b-PS) and poly(methyl acrylate)b-poly(2,3,4,5,6-pentafluorostyrene) (PMA-b-PPFS) were employed. These two block copolymers were physically bound via electrostatic association between the common PAA corona blocks and the added diamine molecules in tetrahydrofuran (THF), a good solvent for both hydrophilic and hydrophobic blocks. As soon as water was added, the unlike hydrophobic blocks were trapped into the same aggregate cores and locked there by the hydrophilic corona where positively charged amine molecules complexed with negatively charged acidic polymer corona blocks. Immiscibility between fluorinated and hydrocarbon chains, as well as different interfacial energies between the unlike hydrophobic blocks and the common PMA midblock, resulted in undulations, as shown in Figure 5.4 Those discrete bulbs with darker contrast in cylindrical micelles (Figure 5B) were ascribed to PPFS chain-rich phase. Cryo-TEM has also been extensively employed to investigate Janus particles. M¨uller’s group has demonstrated the formation of Janus disks128 and cylinders.130 The cross-section analysis on cryo-TEM images of the aggregates128 suggested that the particles were indeed

296

S. Zhong and D. J. Pochan

Figure 5. Cryo-TEM image of cylindrical micelles with internal phase-separated cores at 67% water/THF solution.4 (Reproduced with permission from Science).

divided into two parts as determined by grayscale intensity. Other Janus systems129 have been examined using cryo-TEM as well. 3.2. Self-Assemblies in Organic Solvents and Ionic Liquids Making cryo-TEM samples with organic or mixed solvents will encounter a vitrification problem because most organic solvents are soluble in liquid ethane. One way to solve it is to use liquid nitrogen instead as a vitrifying medium. Liquid nitrogen is used as a cryogen in different solvent systems, such as organic solvents5,13,14,112 and ionic liquids.15–17 Organic solvents that are suitable for cryo-TEM and their corresponding cryogen reservoirs are listed in a previous review paper.12 Lodge’s group112 showed that the structure of poly(styrene-b-dimethylsiloxane) diblock copolymer could be manipulated when samples were prepared in a series of styrene-selective dialkyl phthalates. With increase of selectivity of the mixed solvent at room temperature, the thermodynamic stable micellar morphology changed from spheres to cylinders to vesicles. A reverse transition was observed as temperature increased due to the reduced selectivity of the solvent at elevated temperature. All morphologies were exclusively characterized by cryo-TEM using liquid nitrogen as cryogen. Recently, Lodge’s group15,138,139 has studied the equilibrium assembly of various block copolymers in different ionic liquids. Many conventional nanostructures, such as spherical micelles by polybutadiene-b-polyethylene oxide) (PB-PEO) in 1-ethyl3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMI][TFSI]),138 vesicles by PBPEO in 1-butyl-3-methylimidazolium hexafluo-rophosphate ([BMI][PF6 ]),15 and cylindrical micelles by polystyrene-b-poly(methyl methacrylate) (PS-PMMA) in [BMI][PF6 ],139 have been revealed via cryo-TEM. Some large aggregates formed by lipids in ionic liquids have been reported using cryo-TEM as well.140 In these systems the vitrified ionic liquid solvents have more electron density than the aggregates do, so the contrast in these systems is reversed compared to those of aqueous or most organic/water mixture systems, as shown in Figure 6. Lodge’s group has chosen liquid nitrogen as cryogen for pure THF due to the negligible solubility of THF in liquid nitrogen.5 THF/water mixture can be effectively vitrified in

Cryogenic Transmission Electron Microscopy

297

Figure 6. Vesicles (a) in aqueous solution showing positive contrast116 and (b) in ionic liquid showing negative contrast.15 (Reproduced with permission from The American Chemical Society).

the liquid ethane.1,2,4,127 Pochan’s group127 has shown that the internal nanostructure of polymeric nanoparticles was controllable through the THF/water volume ratio of the solvent mixture. The overall size of the nanoparticles remained constant, but the internal structure changed from a porous structure to a channel-like phase. The cryo-TEM images were compared with those taken from conventional TEM images. Using uranyl acetate aqueous solution to negatively stain polymeric samples is a well-established technique to enhance the contrast of polymer assemblies or biological materials for TEM imaging. In Hales et al.’s study,127 the hydrophilic blocks showed stronger contrast in conventional TEM images due to the deposition of the heavy metal. In contrast, the corona exhibited less mass thickness contrast compared to those densely packed hydrophobic blocks in cryo-TEM images. Combining this heavy metal staining method and cryo-TEM characterization is demonstrated as a convenient way to determine the composition of polymeric phases. 3.3. Thermoresponsive Structures Changing the type of solvents or composition of mixtures is not the only way to control polymeric structures assembled in solution.127,141 Temperature-responsive assembly systems based on block copolymers with poly(N-isopropylacrylamide) (PNIPAM) are well studied. PNIPAM undergoes a reversible coil-globule transition at temperature of ∼32◦ C. A thermally responsive block copolymer gel formed via PS-b-PNIPAM-b-PS was examined via cryo-TEM at different temperatures. Cryo-TEM successfully revealed shrinkage in gel volume upon heating.142 Direct size measurement of PS-PNIPAM core-shell structure106,107 was also conducted using cryo-TEM images. Similar size variation of spherical micelles composed of thermoresponsive polymers with temperature also was reported.108 However, Plamper et al.109 have also observed the morphological thermostability of star-like, miktoarm, thermoresponsive block copolymers at intermediate temperatures. 3.4. Self-Assemblies of Polyelectrolytes Ionic strength and pH play an important role in modification of the morphology of polyelectrolyte assemblies. M¨uller’s group has done extensive work on effect of conditions

298

S. Zhong and D. J. Pochan

of bulk solution on the assembly of polyelectrolyte block copolymers.143–145 Poly(n-butyl acrylate)-b-poly(acrylic acid) (PnBA-b-PAA) can form spherical micelles consisting of a liquid hydrophobic core and a pH- and ionic strength–response hydrophilic corona. The size of the micelles did not change noticeably upon the modification of pH and salinity after the micellization in salt-free solution. If the polymer was dissolved initially with the presence of salt, those micelles had a strong tendency for cluster formation as examined via cryo-TEM and neutron scattering. They argued that in a salt-free environment, unimers preferred to travel between micelles due to coulombic repulsion within PAA chains, thus leading to a smaller size of spherical micelles.143 Their work also indicated that the average aggregation number of micelles of ionic PAA containing block copolymer was subject to pH and ionic strength. The average aggregation number of micelles decreased with pH but increased with ionic strength.144 Other work did not observe a change in micellar aggregation number but did observe the collapse of the charged corona with the addition of lipid.146 Similar phenomenon of charged corona collapse with the presence of salt was also observed in dumbbell-shaped polyelectrolyte brushes.136 Phase transitions in polyelectrolyte systems have been characterized using cryo-TEM as well. Strandman et al.111 showed that with an increase in pH wormlike micelles of the polyelectrolyte star block copolymer (PMMA-b-PAA)4 would disintegrate into spherical micelles due to the increased repulsion force within PAA corona. In contrast, Mendes et al.147 observed a reversed sphere-to-rod transition by adding ethanol and KCl to an aqueous micellar solution. This uniaxial micellar growth was due to reduced solubility of the polymers after the addition. Many other papers about the phase transitions caused by a change of conditions of bulk solutions148,149 are available. As an interesting analogy to lipid/surfactant catanionic system, one can create polymeric systems containing both cationic and anionic polymers, block copolymers, or mixtures of polymer molecules and small-molecule additives. The technological and scientific importance of these systems has been recognized due to their versatility in phase behavior and convenience in manufacture. Fabrication of a new microstructure or nanostructure requires a simple mixture of interactive molecules and shows a convenient route to rich phase behavior and nanostructure with enormous control in a simple manner. One good example of systems rich in assembly behavior through simple mixing is the chargeable triblock copolymer PAA-b-PMA-b-PS with multiamine molecules studied in the Pochan and Wooley collaboration. The underlying goal for studying self-assembly in this system is to be able to construct increasingly complex nanostructures in a simple manner.2 Exotic nanostructures rarely found in polymeric systems; for example, toroids1 and multimicrometer-long helices,2 have been successfully achieved. Cryo-TEM is a critical characterization technique to analyze these fluid nanostructures in solution. For example, helical cylinders, swollen by solvents, will collapse when the solvent is evaporated. Therefore, one can only observe its fluid, expanded morphology in a vitrified solvent thin film. As depicted in Figure 7,2 the natural solute-state helix had a regular pitch distance along the entire multimicrometer-long micelles that was tunable with the type and amount of amine molecules. These results were realized by measuring a statistically large enough number of individual helical micelles imaged in the same grid and from different grids prepared from the same batch of solution. The narrow size distribution indicates the reliability of cryo-TEM as a morphological characterization technique as well as the stability of the fluid microstructures. Cryo-TEM still does not reflect the average pitch distance of helix globally throughout the solution. Small-angle X-ray scattering showed a very strong peak representing the regular helical pitch distance (unpublished data150). Therefore, it is still necessary to use scattering characterization methods for global, quantitative analysis. Other literature examples of a similar morphology are short helices of tens of nanometers in length that

Cryogenic Transmission Electron Microscopy

299

Figure 7. Cryo-TEM for (a) with triethylenetetramine (TETR2), NH:COOH = 10, pitch distance is 45 ± 3 nm; (b) with TETR2, NH:COOH = 15, pitch distance is 58 ± 5 nm; (c) with diethylenetetramine (TRI2), NH:COOH = 10 : 1, pitch distance is 64 ± 3 nm.2 (Reproduced with permission from The Royal Society of Chemistry).

have been found when linear polyelectrolytes were condensed with multivalent inorganic salts.3,133 After investigation on aggregates of positively charged polyelectrolyte block copolymer with a negatively charged homopolymer, Hofs et al.151 proved that the number of polymer chains staying in aggregations depends on the mass ratio of the hydrophobic portion in a block copolymer and the ionic strength and pH of the solution. However, other research contradicts this result, showing stability of the complex of polyelectrolyte even with the variation of salt concentration.152 Polyelectrolytes are of great interest due to the possibility of complexation with biomacromolecules for possible medical applications. Talmon and coworkers123 designed a novel cationic pentablock copolymer that could condense plasmid DNA. The diameters of nanoplexes ranged from 100 to 150 nm measured from cryo-TEM images. The conventional spherical shape was not the only morphology adopted by polyelectrolyte/DNA complexes.153,154 Another biocompatible polyelectrolyte was synthesized by Won and colleagues154 to incorporate DNA for delivery, and cryo-TEM showed that DNA was twisted into compact structures, whereas polymers collapsed into spherical micelles attached on the surface of DNA aggregates. Proteins have also been encapsulated into spherical nanoparticles of block copolymers. Hammond and coworkers14 were able to control polymer/protein capsules to achieve a steady protein release in aqueous solution. Drug delivery has been implemented using polymer systems as well. Vesicles of biodegradable block copolymers were loaded with antibody in water for delivery to the brain.155 Drug-loaded, negatively charged catanionic vesicles were mixed with a positively charged, polymeric network to form a drug release gel.156 Lipids including dyes mixed with polymers were also studied for their phase behavior157 and possible medical applications.158,159 3.5. Assemblies of Polymers with Modified Blocks Chemistry of polymers allows implementation of interactions between polymer chains in order to design and observe new assembly phenomenon. Maeda et al.120 introduced a new block copolymer containing a block undergoing π -π stacking in aqueous solution, and a network of polymeric assembly was derived from this π -conjugated association. Similar work was done by Lee et al.,132 showing the formation of toroids via π -π stacking of polymers, which further grew into two-dimensional (2D) networks.132 Very recently, Lin et al.160 reported a new coil–rod diblock and coil–rod–coil triblock copolymers whose

300

S. Zhong and D. J. Pochan

“rod” block was a conjugated poly[2,7-(9,9-dihexylfluorene)]. In different organic/water mixture, cylindrical bundles were formed through π -π stacking in a step-wise growth route. Metallo-supramolecular A-b-B-b-A triblock copolymers containing a B block formed from a metal–ligand complex were able to aggregate into spherical micelles that were subsequently treated as a new candidate for stimuli-responsive micelles.161–163

4. Peptide and Peptide-Conjugated Macromolecules 4.1. Peptide-Based Hydrogels Most de novo designed peptides with the preference for β-sheet formation exhibit selfassembly progressing in the direction perpendicular to peptide backbone through specific hydrogen bonding. By varying the amino acid sequence, biodegradable hydrogels are made in many peptide systems.10,164 Applications for hydrogels, for example, as an artificial extracellular matrix for tissue regeneration, require desired mechanical properties that are dependent on the nanostructure and microstructures of the gel. Pochan, Schneider and coworkers9 have used small angle neutron scattering (SANS) and cryo-TEM to quantitatively investigate designed peptide nanofibrillar hydrogel networks. They argued, for example, that an increase in peptide concentration resulted in a denser fibrillar network as revealed in SANS and directly visualized in cryo-TEM. The imaging of gel self-assembly kinetics at early stages for β-hairpin peptide selfassembly is a key to understanding the mechanisms of network formation and its recovery after shear thinning.10 For example, MAX1 peptide, (VK)4 -VDPPT-(KV)4 -NH2 , was designed to undergo an intramolecular folding into β-hairpin structure and subsequently self-assemble into long fibers. According to Pochan, Schneider and coworkers,10 the kinetics of gelation was separated in two distinct timescales: (1) early stage fibril cluster formation and intercluster overlap through dangling fibrils and (2) percolation of nanofibrillar clusters into a hydrogel network. This kinetic process could be directly observed in cryo-TEM micrographs shown in Figure 8 acquired through observing vitrified peptide solution during the self-assembly process. As shown in cryo-TEM prepared at different, subsequent aging times, the dark lines on the light background (vitrified water) were assembled peptide fibrils. After aging 10 min, the earliest discernible fibrils were several hundred nanometers long and many already formed branch junctions. Prolonged aging of the solution allowed the fibrils to extend further into the solution, reaching the point where dangling fibrils from different clusters occurred. The dense percolation of fibrillar clusters then led to the formation of hydrogels. Formation of fibular clusters before gelation is of interest because these clusters can survive the large shear stress under which the gel network can be shear thin delivered, subsequently recovering the gel network mechanical properties after cessation of shear by percolation. Nanofibers as a scaffold for dental stem cells have been constructed by Galler et al.164 Repeat units of the peptides can be divided into four parts: alkyl tail, enzyme-cleavable site, glutamic acid for calcium binding, and cell adhesion motif arginine-glycine-aspartic acid (RGD). Their assembly produced a nanostructured, cell-responsive matrix that was so soft that it could be easily filled into small defects. Variation in peptide sequence resulted in two types of gel: one was good for soft tissue regeneration and the other was suitable for mineralized tissues. Hartgerink and coworkers introduced other peptide designs that could form fibers, such as coil–coil peptide with blunt-ends165 and multidomain peptides.166 Naturally obtained egg white lysozyme is found to form wormlike micelles and, subsequently, hydrogel in aqueous solution with the presence of 20 mM dithiothreitol (DTT) at room temperature.167

Cryogenic Transmission Electron Microscopy

301

Figure 8. Cryo-TEM images showing different steps in formation of the hydrogel network: (a–d) t = 10 min, (e, f) t = 26 min, and (g) t = 45 min.10 (Reproduced with permission from The American Chemical Society).

4.2. Assemblies of Peptide-Conjugated Amphiphiles Peptide-conjugated macromolecules, a molecular combination of peptide sequences covalently linked to alkyl chains or polymeric blocks, show a great diversity in self-assembly

302

S. Zhong and D. J. Pochan

Figure 9. Cryo-TEM for the twisted ribbon assembled from RGD-peptide.167 (Reproduced with permission from The American Chemical Society).

behaviors. Stupp’s group168 has shown that hydrophilic peptide-alkyl chain amphiphiles can self-assemble into cylindrical nanofibers and ribbons as a result of hydrogen bonding among peptide segments and hydrophobic collapse of alkyl tails. In order to facilitate biological applications, they introduced the bioactive cell adhesion epitope RGD at the terminus of the peptide segment. Cryo-TEM examination revealed nanobelt morphologies with a left-handed twist (Figure 9168) for this RGD-bearing peptide.168 Different from Stupp’s hydrophilic peptide–hydrophobic alkyl arrangement,168 a hydrophobic peptide segment was conjugated to water-soluble PEG.169 Extended nanofibers with a hydrophobic peptide core and a swollen PEG corona in aqueous solution were characterized using cryo-TEM. Nematic liquid crystals and then hexagonal columnar phases on increasing concentration were formed. The repulsion between fibrils was mediated due to the attractive interactions that resulted from interpenetration of PEG chains.169 Applying a similar idea, B¨orner et al.170 were able to use cryo-TEM to verify PEG-peptide fiber bundles having the high tendency to pack into well-defined aggregates (Figure 10170). These nematic bundles appeared to be suitable for fiber-reinforced materials because layer formation was not allowed. Another system of well-characterized fiber bundles was achieved in an oligo(p-phenylenevinylene)peptide π -conjugated peptide conjugate system due to π -conjugation within the peptide segment.171 With the implementation of a binding attraction between peptide segments, a thermally reversible rodlike micelle could be noncovalently constructed through a peptide coiled-coil motif. A pair of peptide sequences G(EIAALEK)3 (E) and (KIAALKE)3 G (K) capable of associating into heterocoiled coils were chosen to conjugate to PEG and PS, respectively.

Cryogenic Transmission Electron Microscopy

303

Figure 10. Cryo-TEM image of bundles of the anisometric aggregates of peptide–PEG conjugates.169 (Reproduced with permission from The American Chemical Society).

Mixing PS-E and PEG-K in solution resulted in the physical triblock copolymer PS-E/KPEG. This triblock copolymer then assembled into rodlike micelles but separated into spherical micelles with elevated temperature.172 Other than a fibril morphology, spherical micelles and sheets,173 liquid crystals,174 double emulsions,175 vesicles,176 and nanotubes177 have been reported in peptide or peptide conjugate systems, as observed via cryo-TEM.

5. Nanoparticles, Carbon Nanotubes, and DNA Motifs Dispersed in Solution 5.1. Nanoparticles in Solution Nanoparticles (NPs) of controlled size and shape made from semiconductors, metals, and metallic oxides have been of great interest for a number of possible applications in drug delivery, cosmetics, clinical treatments, electronic or optical materials, and catalysis. However, these NPs tend to aggregate into clusters in solution. Grafting water-soluble polymers or lipids on the surface of the NPs can stabilize the suspension of NPs and prevent aggregation. A mixture of gold NP and block copolymer Pluronics resulted in the formation of NP-polymer hybrid materials and the discrete distribution of NPs in the solution was examined via cryo-TEM.178 Biomacromolecules, like proteins, have also been used to cover gold NPs with a consequent partial conformational change and aggregation of the protein–gold NP hybrids with an increase in concentration.179 By coating NPs with amphiphilic copolymers containing PEG, Yu et al.180 could make biocompatible semiconductor quantum dots (QDs) and iron oxide NP for bioimaging. Dispersion of C60 in aqueous solution was achieved by encapsulating them in vesicles.181 Colloids,182–186 vesicles,65,187 and lipid capsulations185,188 were suitable to protect inorganic NPs as proved in many studies. Ballauff’s group189 has been devoted to the development of an in situ synthesis of NPs among PAA polyelectrolyte brushes. In this polyelectrolyte system, metal ions can be confined as counterions within the brush layer and subsequent reduction of the metal

304

S. Zhong and D. J. Pochan

Figure 11. Cryo-TEM images of magnetite dispersion (a) in zero field and (b) in a homogeneous magnetic field (0.2 T).198 (Reproduced with permission from The American Physical Society).

salts led to the formation of NPs that decorated the brushes. Following this manufacturing protocol, formation of Ag NPs,189 Au NPs,190 and Pt nanocrystals191 was successful. Silver carboxylate crystal was formed via redox reaction of silver salts with the presence of vesicle forming lipids, leading to final budlike structures.192 Au and Ag nanowires could be grown on thin surfactant film templates, as reported by Krichevski et al.193 Electrostatic association between oppositely charged soft matter and surface-modified NPs is widely used as an effective means to make hybrid materials. M¨uller’s group194,195 have made tert-amino-hydroxy-functionalized silsesquioxane NPs that were capable of attaching to PAA chains in poly(n-butyl acrylate)-b-poly(acrylic acid) (PnBA-b-PAA) block copolymers over a wide range of pH condition. Similar aggregations, such as positively charged polymer with a metallo-supramolecular polymer196 or polyelectrolyte with an NP-coated lipid complexes,197 are discussed in the literature as well. Magnetic properties of some NPs allow the construction of nano-/microstructures from solute suspensions that are governed by external stimuli such as magnetic field. CryoTEM images (Figure 11198) proved that magnetic Fe2 O4 could be aligned with an external magnetic field and reversibly returned to anisotropic dispersion in natural state.198 An irreversible linear alignment of gold NP was acquired by covalently linking NPs through polymerization of alkanedithiols grafted on the NP surface.199 PbSe and CdSe QDs could form dipolar linear and hexagonal packing depending on their dipole moment alone without any external field.200 5.2. Dispersion of Carbon Nanotubes in Solution Single-walled carbon nanotubes (SWCNT) have been regarded as exceptional near-infrared (near-IR) fluorophores for photostability, high optical anisotropy, and large Stokes shifts, making them promising for applications in optical materials.201–203 SWCNTs are highly hydrophobic and aggregate into nonemissive bundles, so they must be separated for further applications. Cryo-TEM micrographs of SWCNT solutions indicated204 efficient dispersion of single SWCNTs using peptides noncovalently coated onto the carbon nanotubes to enhance the CNT solubility and dispersion in aqueous solution. This noncovalent interaction between CNT and peptides is preferred because it will not change the photoemission of the SWCNTs. Backes et al.205 used water-soluble perylene derivatives to mix with SWCNT in

Cryogenic Transmission Electron Microscopy

305

Figure 12. Cryo-TEM of 0.1 wt% SWCNTs in 10 wt% cetyl trimethyl ammonium bromide (CTAB),208 showing demixing into two phases under shear. (Reproduced with permission from The Royal Society of Chemistry).

aqueous solution. SWCNTs could be fully dispersed at very low concentration of perylene due to the π -π interaction and the electronic interaction between the perylene unit and the nanotube surface. Individually dispersed SWCNTs have been obtained using pH-responsive polyelectrolyte polymer as well.206 Dispersed CNTs are subject to shear force applied on the mixture and thus behave like suspended objects. Hong and Kim207 have proved that CNT clumps in solution were effectively dispersed after being subjected to a continuous extensional flow for a prolonged time. Regev’s group has studied the phase behavior of CNT with lipids under shear via rheology.208,209 They found that CNTs were lined with lipid aggregates in two distinctive phases, as presented in cryo-TEM images (Figure 12209). Cryo-TEM is also used to analyze the heterostructure of complexes of magnetic iron oxide nanoparticles and SWCNT,210 clay in solution,211 and clay with polymers,212 indicative of the broad fields that are impacted by the cryo-TEM technique. 5.3. Assemblies of DNA Motifs in Solution Double-stranded DNAs are of great interest in nanotechnology due to the highly selective hydrogen bonding between complementary DNA strands by Watson-Crick base pairing.213,214 The design of multiarm junctions, structural analogues to natural Holliday junctions, allows the fabrication of complex 2D and three-dimensional (3D) nanostructures for an expanding list of applications, including nanomechanical devices, computing systems, and programmable molecular machines.213 Among the wide range of DNA assemblies, 3D DNA structures are most characterized by cryo-TEM combined with AFM, the most employed characterization method in DNA assembly. Cryo-TEM is able to reveal the prominent edges of DNA boxes215 and cages and their spatial arrangement,216–218 providing critical information for structural computational simulation. Polygonal cages are good example for 3D DNA assemblies. Mao et al.217,218 have designed a three-point-star tile that contains an elongated loop in the center. The increase in the loop length provided it with a significant flexibility and thus allowed it to bend, serving as a 3D point joint for 3D cage structures, such as tetrahedra, dodecahedra, and buckyballs, illustrated in cryo-TEM. A 3D icosahedra and a large molecular cage

306

S. Zhong and D. J. Pochan

were also produced by using the five-point-star tiles.218 This strategy reduced the need for massive DNA building sequences and could be used to make more complex 3D structures.

6. Cryo-TEM Tomography Techniques The combination of fast computer availability, high-resolution CCD cameras, and sophisticated imaging analysis software has enabled the development of cryo-electron tomography (cryo-ET).219 Cryo-ET employs a powerful imaging analysis software that is able to construct 3D images from 2D data observation and collection.220 Cryo-ET allows the visualization of complex assembly morphologies221 and cellular structures under close-to-life conditions in three dimensions with nanoscale resolution.222–224 Though currently critical for biological applications, cryo-ET will become a force in soft materials characterization in the near future. However, the defects caused by electron beam radiation damage still need to be carefully considered. There are two principles for 3D reconstruction: single-particle analysis (SPA) and electron tomography (ET).220 When assemblies or biomacromolecules can be examined in isolation, different random orientations of them can be trapped in a vitrified thin film at the same time and all necessary information is recorded in a 2D fashion. SPA is then applied to construct 3D structural architecture using all these 2D projections. SPA is an effective method to produce 3D images. However, it has some intrinsic drawbacks. First, each collected 2D image is actually a 2D projection of a 3D object. Second, slight structural deviations from particle to particle would be combined to give rise to possible dramatic artifacts in the final 3D structure. ET is designed to minimize this artifact. ET focuses on the thorough examination of a single object. The data for ET 3D reconstruction are obtained by taking multiple 2D images of the same sample being gradually rotated around one/several axis/axes perpendicular to the electron beam,220 as shown in Figure 13.220 There are three ways for 3D reconstruction in tomography, including the Fourier direct method, back-projection method, and series expansion method.225 Cryo-ET is widely applied in biology. By employing the cryo-ET technique, artifacts notorious to chemical fixation and dehydration procedures or from heavy metal staining223 are eliminated. Also, the sample is directly observed under protection of vitrified solvent without any heavy metal staining, avoiding problems in interpretation caused by accumulation of staining materials.223 Easily finding and identifying the structure of interest in tomography recorded at low-dose conditions and having a high signal-to-noise ratio is a key interest in developing effective cryo-ET. Using cryo-ET, the cell wall222 and virus–antibody complexes224 have been thoroughly visualized. Other than biomacromolecules, cryo-ET has been used for characterization of a helical/concentric circular mesostructure,226 revealing its real internal structure for the first time. Cryo-ET also showed that gold NPs were embedded inside a lipid bilayer, leaving the surface and the internal volume free for further funtionalization for drug delivery applications.221 The spatial arrangement of filaments227 or polystyrene particles attaching to one silica core228 as well as the morphology of block copolymer aggregates229 have been clearly revealed. These examples have indicated the possible applications of cryo-ET. The in situ characterization of complex 3D nanostructures and 3D structural arrangement of subunits is an area in which cryo-ET will have huge impact. In summary, cryo-TEM has been quickly becoming a standard characterization technique to be combined with scattering and other techniques—for instance, rheology

Cryogenic Transmission Electron Microscopy

307

Figure 13. Principle of electron tomography and an example. (A) Single-axis tilt geometry for data collection and (B–D) using of electron tomography to analyze the mammalian Golgi from the pancreatic beta cell line.220 (Reproduced with permission from Wiley-Blackwell, Inc.).

measurements—for characterization of self-assembled structures and materials. The thorough and proper interpretation of aggregates and their association mechanisms is essential in the design of new materials and the construction of functional nanostructures.

References 1. Cui, H. G.; Chen, Z. Y.; Wooley, K. L.; Pochan, D. J. “Origins of toroidal micelle formation through charged triblock copolymer self-assembly,” Soft Matter, 2009, 5, 1269–1278. 2. Zhong, S.; Cui, H. G.; Chen, Z. Y.; Wooley, K. L.; Pochan, D. J. “Helix self-assembly through the coiling of cylindrical micelles,” Soft Matter, 2008, 4, 90–93. 3. Xu, Y. Y.; Bolisetty, S.; Drechsler, M.; Fang, B.; Yuan, J. Y.; Harnau, L.; Ballauff, M.; M¨uller, A. H. E. “Manipulating cylindrical polyelectrolyte brushes on the nanoscale by counterions: Collapse transition to helical structures,” Soft Matter, 2009, 5, 379–384. 4. Cui, H. G.; Chen, Z. Y.; Zhong, S.; Wooley, K. L.; Pochan, D. J. “Block copolymer assembly via kinetic control,” Science, 2007, 317, 647–650. 5. Liu, C.; Hillmyer, M. A.; Lodge, T. P. “Evolution of multicompartment micelles to mixed corona micelles using solvent mixtures,” Langmuir, 2008, 24, 12001–12009. 6. Saito, N.; Liu, C.; Lodge, T. P.; Hillmyer, M. A. “Multicompartment micelles from polyestercontaining abc miktoarm star terpolymers,” Macromolecules, 2008, 41, 8815–8822. 7. Pizzey, C. L.; Jewell, C. M.; Hays, M. E.; Lynn, D. M.; Abbott, N. L.; Kondo, Y.; Golan, S.; Talmon, Y. “Characterization of the nanostructure of complexes formed by a redox-active cationic lipid and dna,” Journal of Physical Chemistry B, 2008, 112, 5849–5857. 8. Efrat, R.; Kesselman, E.; Aserin, A.; Garti, N.; Danino, D. “Solubilization of hydrophobic guest molecules in the monoolein discontinuous Q(L) cubic mesophase and its soft nanoparticles,” Langmuir, 2009, 25, 1316–1326.

308

S. Zhong and D. J. Pochan

9. Hule, R. A.; Nagarkar, R. P.; Altunbas, A.; Ramay, H. R.; Branco, M. C.; Schneider, J. P.; Pochan, D. J. “Correlations between structure, material properties and bioproperties in self-assembled beta-hairpin peptide hydrogels,” Faraday Discussions, 2008, 139, 251–264. 10. Yucel, T.; Micklitsch, C. M.; Schneider, J. P.; Pochan, D. J. “Direct observation of early-time hydrogelation in beta-hairpin peptide self-assembly,” Macromolecules, 2008, 41, 5763–5772. 11. Hule, R. A.; Nagarkar, R. P.; Hammouda, B.; Schneider, J. P.; Pochan, D. J. “Dependence of selfassembled peptide hydrogel network structure on local fibril nanostructure,” Macromolecules, 2009, 42, 7137–7145. 12. Cui, H.; Hodgdon, T. K.; Kaler, E. W.; Abezgauz, L.; Danino, D.; Lubovsky, M.; Talmon, Y.; Pochan, D. J. “Elucidating the assembled structure of amphiphiles in solution via cryogenic transmission electron microscopy,” Soft Matter, 2007, 3, 945–955. 13. Esselink, F. J.; Dormidontova, E.; Hadziioannou, G. “Evolution of block copolymer micellar size and structure evidenced with cryo electron microscopy,” Macromolecules, 1998, 31, 2925–2932. 14. Miller, A. C.; Bershteyn, A.; Tan, W.; Hammond, P. T.; Cohen, R. E.; Irvine, D. J. “Block copolymer micelles as nanocontainers for controlled release of proteins from biocompatible oil phases,” Biomacromolecules, 2009, 10, 732–741. 15. Simone, P. M.; Lodge, T. P. “Lyotropic phase behavior of polybutadiene-poly(ethylene oxide) diblock copolymers in ionic liquids,” Macromolecules, 2008, 41, 1753–1759. 16. He, Y. Y.; Li, Z. B.; Simone, P.; Lodge, T. P. “Self-assembly of block copolymer micelles in an ionic liquid,” Journal of the American Chemical Society, 2006, 128, 2745–2750. 17. Abbas, S.; Lodge, T. P. “Superlattice formation in binary mixtures of block copolymer micelles,” Langmuir, 2008, 24, 6247–6253. 18. Lo, K. K. W.; Lee, P. K.; Lau, J. S. Y. “Synthesis, characterization, and properties of luminescent organoiridium(iii) polypyridine complexes appended with an alkyl chain and their interactions with lipid bilayers, surfactants, and living cells,” Organometallics, 2008, 27, 2998–3006. 19. Siegel, D. P.; Green, W. J.; Talmon, Y. “The mechanism of lamellar-to-inverted hexagonal phasetransitions—a study using temperature-jump cryoelectron microscopy,” Biophysical Journal, 1994, 66, 402–414. 20. Svenson, S. “Controlling surfactant self-assembly,” Current Opinion in Colloid & Interface Science, 2004, 9, 201–212. 21. Hak, S.; Sanders, H.; Agrawal, P.; Langereis, S.; Gr¨ull, H.; Keizer, H. M.; Arena, F.; Terreno, E.; Strijkers, G. J.; Nicolay, K. “A high relaxivity GD(III)DOTA-DSPE-based liposomal contrast agent for magnetic resonance imaging,” European Journal of Pharmaceutics and Biopharmaceutics, 2009, 72, 397–404. 22. Lopes, A.; Edwards, K.; Feitosa, E. “Extruded vesicles of dioctadecyldimethylammonium bromide and chloride investigated by light scattering and cryogenic transmission electron microscopy,” Journal of Colloid and Interface Science, 2008, 322, 582–588. 23. Yoshimura, T.; Sakato, A.; Tsuchiya, K.; Ohkubo, T.; Sakai, H.; Abe, M.; Esumi, K. “Adsorption and aggregation properties of amino acid-based n-alkyl cysteine monomeric and N,N -dialkyl cystine gemini surfactants,” Journal of Colloid and Interface Science, 2007, 308, 466–473. 24. Ciani, L.; Ristori, S.; Bonechi, C.; Rossi, C.; Martin, G. “Effect of the preparation procedure on the structural properties of oligonucleotide/cationic liposome complexes (lipoplexes) studied by electron spin resonance and zeta potential,” Biophysical Chemistry, 2007, 131, 80–87. 25. Tucker, I.; Penfold, J.; Thomas, R. K.; Grillo, I.; Barker, J. G.; Mildner, D. F. R. “The surface and solution properties of dihexadecyl dimethylammonium bromide,” Langmuir, 2008, 24, 6509–6520. 26. Marques, E. F.; Brito, R. O.; Silva, S. G.; Rodr´ıguez-Borges, J. E.; do Vale, M. L.; Gomes, P.; Ara´ujo, M. J.; S¨oderman, O. “Spontaneous vesicle formation in catanionic mixtures of amino acid-based surfactants: Chain length symmetry effects,” Langmuir, 2008, 24, 11009–11017. 27. Sakai, K.; Kaji, M.; Takamatsu, Y.; Tsuchiya, K.; Torigoe, K.; Tsubone, K.; Yoshimura, T.; Esumi, K.; Sakai, H.; Abe, M. “Fluorocarbon-hydrocarbon gemini surfactant mixtures in aqueous solution,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2009, 333, 26–31.

Cryogenic Transmission Electron Microscopy

309

28. Blanco, E.; Olsson, U.; Ruso, J. M.; Schulz, P. C.; Prieto, G.; Sarmiento, F. “Phase behavior of semifluorinated catanionic mixtures: Head group dependence and spontaneous formation of vesicles,” Journal of Colloid and Interface Science, 2009, 331, 522–531. 29. Li, X.; Dong, S. L.; Jia, X. F.; Song, A. X.; Hao, J. C. “Vesicles of a new salt-free cat-anionic fluoro/hydrocarbon surfactant system,” Chemistry—A European Journal, 2007, 13, 9495–9502. 30. Tsuchiya, K.; Ishikake, J.; Kim, T. S.; Ohkubo, T.; Sakai, H.; Abe, M. “Phase behavior of mixed solution of a glycerin-modified cationic surfactant and an anionic surfactant,” Journal of Colloid and Interface Science, 2007, 312, 139–145. 31. Vlachy, N.; Renoncourt, A.; Drechsler, A.; Verbavatz, J. M.; Tourauld, D.; Kunz, W. “Blastulae aggregates: New intermediate structures in the micelle-to-vesicle transition of catanionic systems,” Journal of Colloid and Interface Science, 2008, 320, 360–363. 32. Li, X.; Dong, S. L.; Hao, J. C. “Phase behavior, rheological properties, and vesicles of alkyldimethylamine oxide and fluorinated acidic surfactant mixtures,” Soft Matter, 2009, 5, 990–998. 33. Hassan, P. A.; Hodgdon, T. K.; Sagasaki, M.; Fritz-Popovski, G.; Kaler, E. W. “Phase behavior and microstructure evolution in aqueous mixtures of cetyltrimethylammonium bromide and sodium dodecyl tri-oxyethylene sulfate,” Comptes Rendus Chimie, 2009, 12, 18–29. 34. Kakehashi, R.; Karlsson, G.; Almgren, M. “Stomatosomes, blastula vesicles and bilayer disks: Morphological richness of structures formed in dilute aqueous mixtures of a cationic and an anionic surfactant,” Journal of Colloid and Interface Science, 2009, 331, 484–493. 35. C´ardenas, M.; Schill´en, K.; Alfredsson, V.; Duan, R. D.; Nyberg, L.; Amebrant, T. “Solubilization of sphingornyelin vesicles by addition of a bile salt,” Chemistry and Physics of Lipids, 2008, 151, 10–17. 36. Thomas, G. B.; Rader, L. H.; Park, J.; Abezgauz, L.; Danino, D.; DeShong, P.; English, D. S. “Carbohydrate modified catanionic vesicles: Probing multivalent binding at the bilayer interface,” Journal of the American Chemical Society, 2009, 131, 5471–5477. 37. Lozano, N.; Pinazo, A., La Mesa, C.; Perez, L.; Andreozzi, P.; Pons, R. “Catanionic vesicles formed with arginine-based surfactants and 1,2-dipalmitoyl-sn-glycero-3-phosphate monosodium salt,” Journal of Physical Chemistry B, 2009, 113, 6321–6327. 38. Lee, K. E.; Kim, H. M.; Lee, J. O.; Jeon, H.; Han, S. S. “Regulation of CD40 reconstitution into a liposome using different ratios of solubilized LDAO to lipids,” Colloids and Surfaces B: Biointerfaces, 2008, 62, 51–57. 39. Manojlovic, V.; Winkler, K.; Bunjes, V.; Neub, A.; Schubert, R.; Bugarski, B.; Leneweit, G. “Membrane interactions of ternary phospholipid/cholesterol bilayers and encapsulation efficiencies of a rip ii protein,” Colloids and Surfaces B: Biointerfaces, 2008, 64, 284– 296. 40. Kim, S. H.; Haimovich-Caspi, L.; Omer, L.; Yu, C. M.; Talmon, Y.; Wang, N. H. L.; Franses, E. I. “Stability and state of aggregation of aqueous fibrinogen and dipalmitoylphosphatidylcholine lipid vesicles,” Langmuir, 2007, 23, 5657–5664. 41. Rold´an-Vargas, S.; Barnadas-Rodr´ıguez, R.; Quesada-P´erez, M.; Estelrich, J.; CallejasFern´andez, J. “Surface fractals in liposome aggregation,” Physical Review E, 2009, 79, 14. 42. Saveyn, P.; Cocquyt, J.; Bomans, P.; Frederik, P.; De Cuyper, M.; Van Der Meeren, P. “Osmotically induced morphological changes of extruded dioctadecyldimethylammonium chloride (DODAC) dispersions,” Langmuir, 2007, 23, 4775–4781. 43. Fatouros, D. G.; Bergenstahl, B.; Mullertz, A. “Morphological observations on a lipid-based drug delivery system during in vitro digestion,” European Journal of Pharmaceutical Sciences, 2007, 31, 85–94. 44. Jha, A. K.; Bose, A. “Surfactant microstructure and particle aggregation control using amphiphile adsorption on surface-functionalized polystyrene spheres,” Langmuir, 2009, 25, 123–126. 45. Bordes, R.; Vedrenne, M.; Coppel, Y.; Franceschi, S.; Perez, E.; Rico-Lattes, I. “Micelle-vesicle transition of fatty acid based ion-pair surfactants: Interfacial evidence and influence of the ammonium counterion structure,” ChemPhysChem, 2007, 8, 2013–2018.

310

S. Zhong and D. J. Pochan

46. Aratono, M.; Mori, A.; Koga, I.; Shigehisa, M.; Onimaru, N.; Tsuchiya, K.; Takiue, T.; Matsubara, H. “Spontaneous vesicle formation of mixtures of double-chain cationic surfactants with a different counterion,” Journal of Physical Chemistry B, 2008, 112, 12304–12311. 47. Brandel, J.; Torelli, S.; Gellon, G.; Serratrice, G.; Putaux, J. L.; Pierre, J. L. “From molecular to nanostructured iron complexes of amphiphilic chelators based on 8-hydroxyquinoline subunits—evidence of self-assembled edifices mimicking siderophores from marine bacteria,” European Journal of Inorganic Chemistry, 2009, 1, 86–92. 48. Feitosa, E. “Spontaneous vesicles of sodium dihexadecylphosphate in HEPES buffer,” Journal of Colloid and Interface Science, 2008, 320, 608–610. 49. Silva, B. F. B.; Marques, E. F.; Olsson, U. “Unusual vesicle-micelle transitions in a saltfree catanionic surfactant: Temperature and concentration effects,” Langmuir, 2008, 24, 10746–10754. 50. Shin, K. H.; Kim, K. W.; Kim, J. Y.; Lee, K. E.; Han, S. S. “Rhamnolipid morphology and phenanthrene solubility at different pH values,” Journal of Environmental Quality, 2008, 37, 509–514. 51. Capone, S.; Walde, P.; Seebach, D.; Ishikawa, T.; Caputo, R. “pH-sensitive vesicles containing a lipidic beta-amino acid with two hydrophobic chains,” Chemistry and Biodiversity, 2008, 5, 16–30. 52. Guillot-Nieckowski, M.; Joester, D.; St¨ohr, M.; Losson, M.; Adrian, M.; Wagner, B.; Kansy, M.; Heinzelmann, H.; Pugin, R.; Diederich, F.; Gallani, J. L. “Self-assembly, DNA complexation, and pH response of amphiphilic dendrimers for gene transfection,” Langmuir, 2007, 23, 737–746. 53. Cano-Sarabia, M.; Ventosa, N.; Sala, S.; Pati˜no, C.; Arranz, R.; Veciana, J. “Preparation of uniform rich cholesterol unilamellar nanovesicles using CO2 -expanded solvents,” Langmuir, 2008, 24, 2433–2437. 54. Kim, S. H.; Haimovich-Caspi, L.; Omer, L.; Talmon, Y.; Franses, E. I. “Effect of sonication and freezing-thawing on the aggregate size and dynamic surface tension of aqueous dppc dispersions,” Journal of Colloid and Interface Science, 2007, 311, 217–227. 55. Gopalakrishnan, G.; Rouiller, I.; Colman, D. R.; Lennox, R. B. “Supported bilayers formed from different phospholipids on spherical silica substrates,” Langmuir, 2009, 25, 5455–5458. 56. Thor´en, P. E. G.; S¨oderman, O.; Engstr¨om, S.; von Corswant, C. “Interactions of novel, nonhemolytic surfactants with phospholipid vesicles,” Langmuir, 2007, 23, 6956–6965. 57. Vaccaro, M.; von Corswant, C.; S¨oderman, O. “Investigation of the adsorption of PEG1500-12acyloxystearate surfactants onto phospholipid bilayers: An ellipsometry and cryo-tem study,” Biophysical Journal, 2007, 93, 4300–4306. ˇ 58. Ceh, B.; Winterhalter, M.; Frederik, P. M.; Vallner, J. J.; Lasic, D. D. “Stealth(r) liposomes: From theory to product,” Advanced Drug Delivery Reviews, 1997, 24, 165–177. 59. Vel´azquez, M. M.; Valero, M.; Ortega, F.; Gonz´alez, J. B. R. “Structure and size of spontaneously mixtures: Effects of polymer size formed aggregates in aerosol OT/PEG and composition,” Journal of Colloid and Interface Science, 2007, 316, 762–770. 60. Dos Santos, N.; Allen, C.; Doppen, A. M.; Anantha, M.; Cox, K. A. K.; Gallagher, R. C.; Karlsson, G.; Edwards, K.; Kenner, G.; Samuels, L.; Webb, M. S.; Bally, M. B. “Influence of poly(ethylene glycol) grafting density and polymer length on liposomes: Relating plasma circulation lifetimes to protein binding,” Biochimica et Biophysica Acta—Biomembranes, 2007, 1768, 1367–1377. 61. Sandstr¨om, M. C.; Johansson, E.; Edwards, K. “Structure of mixed micelles formed in PEGlipid/lipid dispersions,” Langmuir, 2007, 23, 4192–4198. 62. Johansson, E.; Sandstr¨om, M. C.; Bergstr¨om, M.; Edwards, K. “On the formation of discoidal versus threadlike micelles in dilute aqueous surfactant/lipid systems,” Langmuir, 2008, 24, 1731–1739. 63. Lee, J. H.; Danino, D.; Raghavan, S. R. “Polymerizable vesicles based on a single-tailed fatty acid surfactant: A simple route to robust nanocontainers,” Langmuir, 2009, 25, 1566– 1571.

Cryogenic Transmission Electron Microscopy

311

64. Abe, M.; Koike, T.; Nishiyama, H.; Sharma, S. C.; Tsubone, K.; Tsuchiya, K.; Sakai, K.; Sakai, H.; Shchipunov, Y. A.; Schmidt, J.; Talmon, Y. “Polymerized assemblies of cationic gemini surfactants in aqueous solution,” Journal of Colloid and Interface Science, 2009, 330, 250–253. 65. Steinberg, Y.; Schroeder, A.; Talmon, Y.; Schmidt, J.; Khalfin, R. L.; Cohen, Y.; Devoisselle, J. M.; Begu, S.; Avnir, D. “Triggered release of aqueous content from liposome-derived sol-gel nanocapsules,” Langmuir, 2007, 23, 12024–12031. 66. Romberg, B.; Flesch, F. M.; Hennink, W. E.; Storm, G. “Enzyme-induced shedding of a poly(amino acid)-coating triggers contents release from dioleoyl phosphatidylethanolamine liposomes,” International Journal of Pharmaceutics, 2008, 355, 108–113. 67. Changsan, N.; Chan, H. K.; Separovic, F.; Srichana, T. “Physicochemical characterization and stability of rifampicin liposome dry powder formulations for inhalation,” Journal of Pharmaceutical Sciences, 2009, 98, 628–639. 68. Str¨omstedt, A. A.; Wessman, P.; Ringstad, L.; Edwards, K.; Malmsten, M. “Effect of lipid headgroup composition on the interaction between melittin and lipid bilayers,” Journal of Colloid and Interface Science, 2007, 311, 59–69. 69. Awad, D.; Damian, L.; Winterhalter, M.; Karlsson, G.; Edwards, K.; Gabel, D. “Interaction of Na2B12H11SH with dimyristoyl phosphatidylcholine liposomes,” Chemistry and Physics of Lipids, 2009, 157, 78–85. 70. Rosa, M.; Miguel, M. D.; Lindman, B. “DNA encapsulation by biocompatible catanionic vesicles,” Journal of Colloid and Interface Science, 2007, 312, 87–97. 71. Rosa, M.; Mor´an, M. D.; Miguel, M. D.; Lindman, B. “The association of DNA and stable catanionic amino acid-based vesicles,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2007, 301, 361–375. 72. Rodr´ıguez-Pulido, A.; Aicart, E.; Llorca, O.; Junquera, E. “Compaction process of calf thymus DNA by mixed cationic-zwitterionic liposomes: A physicochemical study,” Journal of Physical Chemistry B, 2008, 112, 2187–2197. 73. Rodr´ıguez-Pulido, A.; Ortega, F.; Llorca, O.; Aicart, E.; Junquera, E. “A physicochemical characterization of the interaction between DC-Chol/DOPE cationic liposomes and DNA,” Journal of Physical Chemistry B, 2008, 112, 12555–12565. 74. Kuvichkin, V. V.; Danev, R. S.; Shigematsu, H.; Nagayama, K. “DNA-induced aggregation and fusion of phosphatidylcholine liposomes in the presence of multivalent cations observed by the cryo-TEM technique,” Journal of Membrane Biology, 2009, 227, 95–103. 75. Letrou-Bonneval, E.; Ch`evre, R.; Lambert, O.; Costet, P.; Andr´e, C.; Tellier, C.; Pitard, B. “Galactosylated multimodular lipoplexes for specific gene transfer into primary hepatocytes,” Journal of Gene Medicine, 2008, 10, 1198–1209. 76. Zidovska, A.; Evans, H. M.; Ewert, K. K.; Quispe, J.; Carragher, B.; Potter, C. S.; Safinya, C. R. “Liquid crystalline phases of dendritic lipid-DNA self-assemblies: Lamellar, hexagonal, and DNA bundles,” Journal of Physical Chemistry B, 2009, 113, 3694–3703. 77. Liu, Y. C.; Le Ny, A. L. M.; Schmidt, J.; Talmon, Y.; Chmelka, B. F.; Lee, C. T. “Photo-assisted gene delivery using light-responsive catanionic vesicles,” Langmuir, 2009, 25, 5713–5724. 78. Desigaux, L.; Sainlos, M.; Lambert, O.; Chevre, R.; Letrou-Bonneval, E.; Vigneron, J. P.; Lehn, P.; Lehn, J. M.; Pitard, B. “Self-assembled lamellar complexes of siRNA with lipidic aminoglycoside derivatives promote efficient sirna delivery and interference,” Proceedings of the National Academy of Sciences U. S. A., 2007, 104, 16534–16539. 79. Kuperkar, K.; Abezgauz, L.; Danino, D.; Verma, G.; Hassan, P. A.; Aswal, V. K.; Varade, D.; Bahadur, P. “Viscoelastic micellar water/CTAB/NaNO3 solutions: Rheology, SANS and cryo-TEM analysis,” Journal of Colloid and Interface Science, 2008, 323, 403–409. 80. Kumar, R.; Kalur, G. C.; Ziserman, L.; Danino, D.; Raghavan, S. R. “Wormlike micelles of a C22-tailed zwitterionic betaine surfactant: From viscoelastic solutions to elastic gels,” Langmuir, 2007, 23, 12849–12856. 81. Antunes, F. E.; Coppola, L.; Rossi, C. O.; Ranieri, G. A. “Gelation of charged bionanocompartments induced by associative and non-associative polysaccharides,” Colloids and Surfaces B: Biointerfaces, 2008, 66, 134–140.

312

S. Zhong and D. J. Pochan

82. Ge, W.; Kesselman, E.; Talmon, Y.; Hart, D. J.; Zakin, J. L. “Effects of chemical structures of para-halobenzoates on micelle nanostructure, drag reduction and rheological behaviors of dilute CTAC solutions,” Journal of Non-Newtonian Fluid Mechanics, 2008, 154, 1–12. 83. Lo Nostro, P.; Murgia, S.; Lagi, M.; Fratini, E.; Karlsson, G.; Almgren, M.; Monduzzi, M.; Ninham, B. W.; Baglioni, P. “Interconnected networks: Structural and dynamic characterization of aqueous dispersions of dioctanoylphosphatidylcholine,” Journal of Physical Chemistry B, 2008, 112, 12625–12634. 84. Meister, A.; Drescher, S.; Garamus, V. M.; Karlsson, G.; Graf, G.; Dobner, B.; Blume, A. “Temperature-dependent self-assembly and mixing behavior of symmetrical single-chain bolaamphiphiles,” Langmuir, 2008, 24, 6238–6246. 85. Frounfelker, B. D.; Kalur, G. C.; Cipriano, B. H.; Danino, D.; Raghavan, S. R. “Persistence of birefringence in sheared solutions of wormlike micelles,” Langmuir, 2009, 25, 167–172. 86. Berti, D.; Bombelli, F. B.; Fortini, M.; Baglioni, P. “Amphiphilic self-assemblies decorated by nucleobases,” Journal of Physical Chemistry B, 2007, 111, 11734–11744. 87. Vucic, E.; Sanders, H.; Arena, F.; Terreno, E.; Aimo, S.; Nicolay, K.; Leupold, E.; Dathe, M.; Sommerdijk, N.; Fayad, Z. A.; Mulder, W. J. M. “Well-defined, multifunctional nanostructures of a paramagnetic lipid and a lipopeptide for macrophage imaging,” Journal of the American Chemical Society, 2009, 131, 406–407. 88. Drescher, S.; Meister, A.; Blume, A.; Karlsson, G.; Almgren, M.; Dobner, B. “General synthesis and aggregation behaviour of a series of single-chain 1,omega-bis(phosphocholines),” Chemistry—A European Journal, 2007, 13, 5300–5307. 89. Nizri, G.; Makarsky, A.; Magdassi, S.; Talmon, Y. “Nanostructures formed by self-assembly of negatively charged polymer and cationic surfactants,” Langmuir, 2009, 25, 1980–1985. 90. Almgren, M.; Born´e, J.; Feitosa, E.; Khan, A.; Lindman, B. “Dispersed lipid liquid crystalline phases stabilized by a hydrophobically modified cellulose,” Langmuir, 2007, 23, 2768– 2777. 91. Efrat, R.; Aserin, A.; Kesselman, E.; Danino, D.; Wachtel, E. J.; Garti, N. “Liquid micellar discontinuous cubic mesophase from ternary monoolein/ethanol/water mixtures,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2007, 299, 133–145. 92. Yuan, Z. W.; Yin, Z. L.; Sun, S. X.; Hao, J. C. “Densely stacked multilamellar and oligovesicular vesicles, bilayer cylinders, and tubes joining with vesicles of a salt-free catanionic extractant and surfactant system,” Journal of Physical Chemistry B, 2008, 112, 1414–1419. 93. von Berlepsch, H.; Kirstein, S.; Hania, R.; Pugˇzlys, A.; B¨ottcher, C. “Modification of the nanoscale structure of the j-aggregate of a sulfonate-substituted amphiphilic carbocyanine dye through incorporation of surface-active additives,” Journal of Physical Chemistry B, 2007, 111, 1701–1711. 94. Gr¨abner, D.; Zhai, L.; Talmon, Y.; Schmidt, J.; Freiberger, N.; Glatter, O.; Herzog, B.; Hoffmann, H. “Phase behavior of aqueous mixtures of 2-phenylbenzimidazole-5-sulfonic acid and cetyltrimethylammonium bromide: Hydrogels, vesicles, tubules, and ribbons,” Journal of Physical Chemistry B, 2008, 112, 2901–2908. 95. Zidovska, A.; Ewert, K. K.; Quispe, J.; Carragher, B.; Potter, C. S.; Safinya, C. R. “Block liposomes from curvature-stabilizing lipids: Connected nanotubes, -rods, or -spheres,” Langmuir, 2009, 25, 2979–2985. 96. Owen, T.; Pynn, R.; Hammouda, B.; Butler, A. “Metal-dependent self-assembly of a microbial surfactant,” Langmuir, 2007, 23, 9393–9400. 97. Koutsantonis, G. A.; Nealon, G. L.; Buckley, C. E.; Paskevicius, M.; Douce, L.; Harrowfield, J. M.; McDowall, A. W. “Wormlike micelles from a cage amine metallosurfactant,” Langmuir, 2007, 23, 11986–11990. 98. Kirchherr, A. K.; Briel, A.; M¨ader, K. “Stabilization of indocyanine green by encapsulation within micellar systems,” Molecular Pharmaceutics, 2009, 6, 480–491. 99. Becherer, M. S.; Schade, B.; B¨ottcher, C.; Hirsch, A. “Supramolecular assembly of self-labeled amphicalixarenes,” Chemistry—A European Journal, 2009, 15, 1637–1648.

Cryogenic Transmission Electron Microscopy

313

100. Sch¨uhle, D. T.; Schatz, J.; Laurent, S.; Elst, L. V.; Muller, R. N.; Stuart, M. C. A.; Peters, J. A. “Calix[4]arenes as molecular platforms for magnetic resonance imaging (MRI) contrast agents,” Chemistry—A European Journal, 2009, 15, 3290–3296. 101. Esposito, E.; Fantin, M.; Marti, M.; Drechsler, M.; Paccamiccio, L.; Mariani, P.; Sivieri, E.; Lain, F.; Menegatti, E.; Morari, M.; Cortesi, R. “Solid lipid nanoparticles as delivery systems for bromocriptine,” Pharmaceutical Research, 2008, 25, 1521–1530. 102. Spernath, L.; Regev, O.; Levi-Kalisman, Y.; Magdassi, S. “Phase transitions in O/W lauryl acrylate emulsions during phase inversion, studied by light microscopy and cryo-TEM,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2009, 332, 19–25. 103. Alb, A. M.; Reed, W. F. “Simultaneous monitoring of polymer and particle characteristics during emulsion polymerization,” Macromolecules, 2008, 41, 2406–2414. 104. Fairley, N.; Hoang, B.; Allen, C. “Morphological control of poly(ethylene glycol)-blockpoly(epsilon-caprolactone) copolymer aggregates in aqueous solution,” Biomacromolecules, 2008, 9, 2283–2291. 105. Rieger, J.; Stoffelbach, F.; Cui, D.; Imberty, A.; Lameignere, E.; Putaux, J. L.; J´erˆome, R.; Auz´ely-Velty, R. “Mannosylated poly(ethylene oxide)-b-poly(is an element of-caprolactone) diblock copolymers: Synthesis, characterization, and interaction with a bacterial lectin,” Biomacromolecules, 2007, 8, 2717–2725. 106. Crassous, J. J.; Wittemann, A.; Siebenb¨urger, M.; Schrinner, M.; Drechsler, M.; Ballauff, M. “Direct imaging of temperature-sensitive core-shell latexes by cryogenic transmission electron microscopy,” Colloid & Polymer Science, 2008, 286, 805–812. 107. Kjoniksen, A. L.; Zhu, K.; Karlsson, G.; Nystr¨om, B. “Novel transition behavior in aqueous solutions of a charged thermoresponsive triblock copolymer,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2009, 333, 32–45. 108. Rakhmatullina, E.; Braun, T.; Chami, M.; Malinova, V.; Meier, W. “Self-organization behavior of methacrylate-based amphiphilic di- and triblock copolymers,” Langmuir, 2007, 23, 12371–12379. 109. Plamper, F. A.; McKee, J. R.; Laukkanen, A.; Nyk¨anen, A.; Walther, A.; Ruokolainen, J.; Aseyev, V.; Tenhu, H. “Miktoarm stars of poly(ethylene oxide) and poly(dimethylaminoethyl methacrylate): Manipulation of micellization by temperature and light,” Soft Matter, 2009, 5, 1812–1821. 110. Mincheva, R.; Bougard, F.; Paneva, D.; Vachaudez, M.; Manolova, N.; Rashkov, L.; Dubois, P. “Natural polyampholyte-based core-shell nanoparticles with n-carboxyethylchitosancontaining core and poly(ethylene oxide) shell,” Biomacromolecules, 2009, 10, 838–844. 111. Strandman, S.; Zarembo, A.; Darinskii, A. A.; L¨oflund, B.; Butcher, S. J.; Tenhu, H. “Selfassembling of star-like amphiphilic block copolymers with polyelectrolyte blocks. effect of pH,” Polymer, 2007, 48, 7008–7016. 112. Abbas, S.; Li, Z. B.; Hassan, H.; Lodge, T. P. “Thermoreversible morphology transitions of poly(styrene-b-dimethylsiloxane) diblock copolymer micelles in dilute solution,” Macromolecules, 2007, 40, 4048–4052. 113. Petrov, P. D.; Drechsler, M.; M¨uller, A. H. E. “Self-assembly of asymmetric poly(ethylene oxide)-block-poly (n-butyl acrylate) diblock copolymers in aqueous media to unexpected morphologies,” Journal of Physical Chemistry B, 2009, 113, 4218–4225. 114. Fenimore, S. G.; Abezgauz, L.; Danino, D.; Ho, C. C.; Co, C. C. “Spontaneous alternating copolymer vesicles of alkylmaleimides and vinyl gluconamide,” Macromolecules, 2009, 42, 2702–2707. 115. Wu, D.; Abezgauz, L.; Danino, D.; Ho, C. C.; Co, C. C. “Alternating polymer vesicles,” Soft Matter, 2008, 4, 1066–1071. 116. Davis, K. P.; Lodge, T. P.; Bates, F. S. “Vesicle membrane thickness in aqueous dispersions of block copolymer blends,” Macromolecules, 2008, 41, 8289–8291. 117. Zupancich, J. A.; Bates, F. S.; Hillmyer, M. A. “Synthesis and self-assembly of RGDfunctionalized PEO-PB amphiphiles,” Biomacromolecules, 2009, 10, 1554–1563.

314

S. Zhong and D. J. Pochan

118. Battaglia, G.; Ryan, A. J. “Bilayers and interdigitation in block copolymer vesicles,” Journal of the American Chemical Society, 2005, 127, 8757–8764. 119. Ruthard, C.; Maskos, M.; Kolb, U.; Gr¨ohn, F. “Finite-size networks from cylindrical polyelectrolyte brushes and porphyrins,” Macromolecules, 2009, 42, 830–840. 120. Maeda, H.; Ito, Y.; Haketa, Y.; Eifuku, N.; Lee, E.; Lee, M.; Hashishin, T.; Kaneko, K. “Solventassisted organized structures based on amphiphilic anion-responsive Pi-conjugated systems,” Chemistry—A European Journal, 2009, 15, 3706–3719. 121. Burrows, H. D.; Tapia, M. J.; Fonseca, S. M.; Pradhan, S.; Scherf, U.; Silva, C. L.; Pais, A.; Valente, A. J. M.; Schill´en, K.; Alfredsson, V.; Carnerup, A. M.; Tomˇsiˇc, M.; Jamnik, A. “Solubilization of poly{1,4-phenylene-[9,9-bis(4-phenoxy-butylsulfonate)]fluorene-2,7-diyl} in water by nonionic amphiphiles,” Langmuir, 2009, 25, 5545–5556. 122. Rangelov, S.; Almgren, M.; Halacheva, S.; Tsvetanov, C. “Polyglycidol-based analogues of pluronic block copolymers. light scattering and cryogenic transmission electron microscopy studies,” Journal of Physical Chemistry C, 2007, 111, 13185–13191. 123. Agarwal, A.; Vilensky, R.; Stockdale, A.; Talmon, Y.; Unfer, R. C.; Mallapragada, S. K. “Colloidally stable novel copolymeric system for gene delivery in complete growth media,” Journal of Controlled Release, 2007, 121, 28–37. 124. Filippov, S. K.; Lezov, A. V.; Sergeeva, O. Y.; Olifirenko, A. S.; Lesnichin, S. B.; Domnina, ˇ N. S.; Komarova, E. A.; Almgren, M.; Karlsson, G.; Stepanek, P. “Aggregation of dextran hydrophobically modified by sterically-hindered phenols in aqueous solutions: Aggregates vs. single molecules,” European Polymer Journal, 2008, 44, 3361–3369. 125. Hermans, T. M.; Broeren, M. A. C.; Gomopoulos, N.; Smeijers, A. F.; Mezari, B.; Van Leeuwen, E. N. M.; Vos, M. R. J.; Magusin, P.; Hilbers, P. A. J.; Van Genderen, M. H. P.; Sommerdijk, N.; Fytas, G.; Meijer, E. W. “Stepwise noncovalent synthesis leading to dendrimerbased assemblies in water,” Journal of the American Chemical Society, 2007, 129, 15631– 15638. 126. Pressly, E. D.; Rossin, R.; Hagooly, A.; Fukukawa, K. I.; Messmore, B. W.; Welch, M. J.; Wooley, K. L.; Lamm, M. S.; Hule, R. A.; Pochan, D. J.; Hawker, C. J. “Structural effects on the biodistribution and positron emission tomography (PET) imaging of well-defined Cu-64labeled nanoparticles comprised of amphiphilic block graft copolymers,” Biomacromolecules, 2007, 8, 3126–3134. 127. Hales, K.; Chen, Z. Y.; Wooley, K. L.; Pochan, D. J. “Nanoparticles with tunable internal structure from triblock copolymers of PAA-b-PMA-b-PS,” Nano Letters, 2008, 8, 2023–2026. 128. Walther, A.; Andr´e, X.; Drechsler, M.; Abetz, V.; M¨uller, A. H. E. “Janus discs,” Journal of the American Chemical Society, 2007, 129, 6187–6198. 129. Voets, I. K.; Fokkink, R.; Hellweg, T.; King, S. M.; de Waard, P.; de Keizer, A.; Stuart, M. A. C. “Spontaneous symmetry breaking: Formation of janus micelles,” Soft Matter, 2009, 5, 999–1005. 130. Walther, A.; Drechsler, M.; Rosenfeldt, S.; Harnau, L.; Ballauff, M.; Abetz, V.; M¨uller, A. H. E. “Self-assembly of janus cylinders into hierarchical superstructures,” Journal of the American Chemical Society, 2009, 131, 4720–4728. 131. Yan, Y.; Harnau, L.; Besseling, N. A. M.; de Keizer, A.; Ballauff, M.; Rosenfeldt, S.; Stuart, M. A. C. “Spherocylindrical coacervate core micelles formed by a supramolecular coordination polymer and a diblock copolymer,” Soft Matter, 2008, 4, 2207–2212. 132. Lee, E.; Jeong, Y. H.; Kim, J. K.; Lee, M. “Controlled self-assembly of asymmetric dumbbellshaped rod amphiphiles: Transition from toroids to planar nets,” Macromolecules, 2007, 40, 8355–8360. 133. Xu, Y. Y.; Bolisetty, S.; Drechsler, M.; Fang, B.; Yuan, J. Y.; Ballauff, M.; M¨uller, A. H. E. “pH and salt responsive poly(N,N-dimethylaminoethyl methacrylate) cylindrical brushes and their quaternized derivatives,” Polymer, 2008, 49, 3957–3964. 134. Weber, C. H. M.; Chiche, A.; Krausch, G.; Rosenfeldt, S.; Ballauff, M.; Harnau, L.; G¨ottkerSchnetmann, I.; Tong, Q.; Mecking, S. “Single lamella nanoparticles of polyethylene,” Nano Letters, 2007, 7, 2024–2029.

Cryogenic Transmission Electron Microscopy

315

135. Ruthstein, S.; Schmidt, J.; Kesselman, E.; Popovitz-Biro, R.; Omer, L.; Frydman, V.; Talmon, Y.; Goldfarb, D. “Molecular level processes and nanostructure evolution during the formation of the cubic mesoporous material KIT-6,” Chemistry of Materials, 2008, 20, 2779–2792. 136. Hoffmann, M.; Lu, Y.; Schrinner, M.; Ballauff, M.; Harnau, L. “Dumbbell-shaped polyelectrolyte brushes studied by depolarized dynamic light scattering,” Journal of Physical Chemistry B, 2008, 112, 14843–14850. 137. Li, Z. B.; Kesselman, E.; Talmon, Y.; Hillmyer, M. A.; Lodge, T. P. “Multicompartment micelles from ABC miktoarm stars in water,” Science, 2004, 306, 98–101. 138. Meli, L.; Lodge, T. P. “Equilibrium vs metastability: High-temperature annealing of spherical block copolymer micelles in an ionic liquid,” Macromolecules, 2009, 42, 580–583. 139. Simone, P. M.; Lodge, T. P. “Micellization of ps-pmma diblock copolymers in an ionic liquid,” Macromolecular Chemistry and Physics, 2007, 208, 339–348. 140. Sakai, H.; Saitoh, T.; Endo, T.; Tsuchiya, K.; Sakai, K.; Abe, M. “Phytosterol ethoxylates in room-temperature ionic liquids: Excellent interfacial properties and gel formation,” Langmuir, 2009, 25, 2601–2603. 141. Cui, H. G.; Chen, Z. Y.; Wooley, K. L.; Pochan, D. J. “Controlling micellar structure of amphiphilic charged triblock copolymers in dilute solution via coassembly with organic counterions of different spacer lengths,” Macromolecules, 2006, 39, 6599–6607. 142. Nyk¨aenen, A.; Nuopponen, M.; Hiekkataipale, P.; Hirvonen, S. P.; Soininen, A.; Tenhu, H.; Ikkala, O.; Mezzenga, R.; Ruokolainen, J. “Direct imaging of nanoscopic plastic deformation below bulk T-g and chain stretching in temperature-responsive block copolymer hydrogels by cryo-TEM,” Macromolecules, 2008, 41, 3243–3249. 143. Colombani, O.; Ruppel, M.; Burkhardt, M.; Drechsler, M.; Schumacher, M.; Gradzielski, M.; Schweins, R.; M¨uller, A. H. E. “Structure of micelles of poly(n-butyl acrylate)-block-poly (acrylic acid) diblock copolymers in aqueous solution,” Macromolecules, 2007, 40, 4351– 4362. 144. Burkhardt, M.; Martinez-Castro, N.; Tea, S.; Drechsler, M.; Babin, I.; Grishagin, I.; Schweins, R.; Pergushov, D. V.; Gradzielski, M.; Zezin, A. B.; M¨uller, A. H. E. “Polyisobutylene-blockpoly(methacrylic acid) diblock copolymers: Self-assembly in aqueous media,” Langmuir, 2007, 23, 12864–12874. 145. Burkhardt, M.; Ruppel, M.; Tea, S.; Drechsler, M.; Schweins, R.; Pergushov, D. V.; Gradzielski, M.; Zezin, A. B.; M¨uller, A. H. E. “Water-soluble interpolyelectrolyte complexes of polyisobutylene-block-poly(methacrylic acid) micelles: Formation and properties,” Langmuir, 2008, 24, 1769–1777. 146. Schmidt, V., Di Cola, E.; Giacomelli, C.; Brisson, A. R.; Narayanan, T.; Borsali, R. “Polyelectrolyte behavior of diblock copolymer micelles having phosphonic diacid groups at the corona,” Macromolecules, 2008, 41, 2195–2202. 147. Denkova, A. G.; Mendes, E.; Coppens, M. O. “Kinetics and mechanism of the sphere-to-rod transition of triblock copolymer micelles in aqueous solutions,” Journal of Physical Chemistry B, 2009, 113, 989–996. 148. Yan, Y.; Hoffmann, H.; Makarsky, A.; Richter, W.; Talmon, Y. “Swelling of l-alpha-phases by matching the refractive index of the water-glycerol mixed solvent and that of the bilayers in the block copolymer system of (EO)(15)-(PDMS)(15)-(EO)(15),” Journal of Physical Chemistry B, 2007, 111, 6374–6382. 149. Rangelov, S.; Halacheva, S.; Garamus, V. M.; Almgren, M. “Structural polymorphism exhibited by polyglycidol-based analogues to pluronic copolymers in aqueous solution,” Macromolecules, 2008, 41, 8885–8894. 150. Zhong, S.; Zhang, K.; Wooley, K. L.; Pochan, D. J. “Helical cylindrical micelles via kinetic self-assembly of block copolymer in solution,” to be submitted. 151. Hofs, B.; de Keizer, A.; Stuart, M. A. C. “On the stability of (highly aggregated) polyelectrolyte complexes containing a charged-block-neutral diblock copolymer,” Journal of Physical Chemistry B, 2007, 111, 5621–5627.

316

S. Zhong and D. J. Pochan

152. G¨ardlund, L.; W˚agberg, L.; Norgren, M. “New insights into the structure of polyelectrolyte complexes,” Journal of Colloid and Interface Science, 2007, 312, 237–246. 153. Ainalem, M. L.; Carnerup, A. M.; Janiak, J.; Alfredsson, V.; Nylander, T.; Schill´en, K. “Condensing DNA with poly(amido amine) dendrimers of different generations: Means of controlling aggregate morphology,” Soft Matter, 2009, 5, 2310–2320. 154. Sharma, R.; Lee, J. S.; Bettencourt, R. C.; Xiao, C.; Konieczny, S. F.; Won, Y. Y. “Effects of the incorporation of a hydrophobic middle block into a PEG-polycation diblock copolymer on the physicochemical and cell interaction properties of the polymer-DNA complexes,” Biomacromolecules, 2008, 9, 3294–3307. 155. Pang, Z. Q.; Lu, W.; Gao, H. L.; Hu, K. L.; Chen, J.; Zhang, C. L.; Gao, X. L.; Jiang, X. G.; Zhu, C. Q. “Preparation and brain delivery property of biodegradable polymersomes conjugated with OX26,” Journal of Controlled Release, 2008, 128, 120–127. 156. Dew, N.; Edwards, K.; Edsman, K. “Gel formation in systems composed of drug containing catanionic vesicles and oppositely charged hydrophobically modified polymer,” Colloids and Surfaces B: Biointerfaces, 2009, 70, 187–197. 157. Shimoni, K.; Danino, D. “Imperfect dissolution in nonionic block copolymer and surfactant mixtures,” Langmuir, 2009, 25, 2736–2742. 158. Sandstr¨om, M. C.; Johansson, E.; Edwards, K. “Influence of preparation path on the formation of discs and threadlike micelles in DSPE-PEG(2000)/lipid systems,” Biophysical Chemistry, 2008, 132, 97–103. 159. Mueller, W.; Koynov, K.; Fischer, K.; Hartmann, S.; Pierrat, S.; Basch´e, T.; Maskos, M. “Hydrophobic shell loading of PB-b-PEO vesicles,” Macromolecules, 2009, 42, 357–361. 160. Lin, C. H.; Tung, Y. C.; Ruokolainen, J.; Mezzenga, R.; Chen, W. C. “Poly [2,7-(9,9dihexylfluorene)]-block-poly(2-vinylpyridine) rod-coil and coil-rod-coil block copolymers: Synthesis, morphology and photophysical properties in methanol/THF mixed solvents,” Macromolecules, 2008, 41, 8759–8769. 161. Chiper, M.; Meier, M. A. R.; Wouters, D.; Hoeppener, S.; Fustin, C. A.; Gohy, J. F.; Schubert, U. S. “Supramolecular self-assembled Ni(II), Fe(II), and Co(II) ABA triblock copolymers,” Macromolecules, 2008, 41, 2771–2777. 162. Chiper, M.; Winter, A.; Hoogenboom, R.; Egbe, D. A. M.; Wouters, D.; Hoeppener, S.; Fustin, C. A.; Gohy, J. F.; Schubert, U. S. “Synthesis and micellization of coil-rod-coil ruthenium(II) terpyridine assemblies,” Macromolecules, 2008, 41, 8823–8831. 163. Landsmann, S.; Winter, A.; Chiper, M.; Fustin, C. A.; Hoeppener, S.; Wouters, D.; Gohy, J. F.; Schubert, U. S. “Poly(dimethylsiloxane)-substituted 2,2 : 6,2 -terpyridines: Synthesis and characterization of new amphiphilic supramolecular diblock copolymers,” Macromolecular Chemistry and Physics, 2008, 209, 1666–1672. 164. Galler, K. M.; Cavender, A.; Yuwono, V.; Dong, H.; Shi, S. T.; Schmalz, G.; Hartgerink, J. D.; D’Souza, R. N. “Self-assembling peptide amphiphile nanofibers as a scaffold for dental stem cells,” Tissue Engineering Part A, 2008, 14, 2051–2058. 165. Dong, H.; Paramonov, S. E.; Hartgerink, J. D. “Self-assembly of alpha-helical coiled coil nanofibers,” Journal of the American Chemical Society, 2008, 130, 13691–13695. 166. Dong, H.; Paramonov, S. E.; Aulisa, L.; Bakota, E. L.; Hartgerink, J. D. “Self-assembly of multidomain peptides: Balancing molecular frustration controls conformation and nanostructure,” Journal of the American Chemical Society, 2007, 129, 12468–12472. 167. Yan, H.; Frielinghaus, H.; Nykanen, A.; Ruokolainen, J.; Saiani, A.; Miller, A. F. “Thermoreversible lysozyme hydrogels: Properties and an insight into the gelation pathway,” Soft Matter, 2008, 4, 1313–1325. 168. Cui, H.; Muraoka, T.; Cheetham, A. G.; Stupp, S. I. “Self-assembly of giant peptide nanobelts,” Nano Letters, 2009, 9, 945–951. 169. Hamley, I. W.; Krysmann, M. J.; Kelarakis, A.; Castelletto, V.; Noirez, L.; Hule, R. A.; Pochan, D. J. “Nematic and columnar ordering of a PEG-peptide conjugate in aqueous solution,” Chemistry—A European Journal, 2008, 14, 11369–11375.

Cryogenic Transmission Electron Microscopy

317

170. B¨orner, H. G.; Smarsly, B. M.; Hentschel, J.; Rank, A.; Schubert, R.; Geng, Y.; Discher, D. E.; Hellweg, T.; Brandt, A. “Organization of self-assembled peptide-polymer nanofibers in solution,” Macromolecules, 2008, 41, 1430–1437. 171. Matmour, R.; De Cat, I.; George, S. J.; Adriaens, W.; Lecl`ere, P.; Bomans, P. H. H.; Sommerdijk, N.; Gielen, J. C.; Christianen, P. C. M.; Heldens, J. T.; van Hest, J. C. M.; L¨owik, D.; De Feyter, S.; Meijer, E. W.; Schenning, A. “Oligo(p-phenylenevinylene)-peptide conjugates: Synthesis and self-assembly in solution and at the solid-liquid interface,” Journal of the American Chemical Society, 2008, 130, 14576–14583. 172. Marsden, H. R.; Korobko, A. V.; van Leeuwen, E. N. M.; Pouget, E. M.; Veen, S. J.; Sommerdijk, N.; Kros, A. “Noncovalent triblock copolymers based on a coiled-coil peptide motif,” Journal of the American Chemical Society, 2008, 130, 9386–9393. 173. Yang, J. Z.; Alemany, L. B.; Driver, J.; Hartgerink, J. D.; Barron, A. R. “Fullerene-derivatized amino acids: Synthesis, characterization, antioxidant properties, and solid-phase peptide synthesis,” Chemistry—A European Journal, 2007, 13, 2530–2545. 174. Pomerantz, W. C.; Yuwono, V. M.; Pizzey, C. L.; Hartgerink, J. D.; Abbott, N. L.; Gellman, S. H. “Nanofibers and lyotropic liquid crystals from a class of self-assembling beta-peptides,” Angewandte Chemie International Edition, 2008, 47, 1241–1244. 175. Hanson, J. A.; Chang, C. B.; Graves, S. M.; Li, Z. B.; Mason, T. G.; Deming, T. J. “Nanoscale double emulsions stabilized by single-component block copolypeptides,” Nature, 2008, 455, 85–88. 176. Adams, D. J.; Atkins, D.; Cooper, A. I.; Furzeland, S.; Trewin, A.; Young, I. “Vesicles from peptidic side-chain polymers synthesized by atom transfer radical polymerization,” Biomacromolecules, 2008, 9, 2997–3003. 177. Bucak, S.; Cenker, C.; Nasir, I.; Olsson, U.; Zackrisson, M. “Peptide nanotube nematic phase,” Langmuir, 2009, 25, 4262–4265. 178. Rahme, K.; Oberdisse, J.; Schweins, R.; Gaillard, C.; Marty, J. D.; Mingotaud, C.; Gauffre, F. “Pluronics-stabilized gold nanoparticles: Investigation of the structure of the polymer-particle hybrid,” ChemPhysChem, 2008, 9, 2230–2236. 179. Zhang, D. M.; Neumann, O.; Wang, H.; Yuwono, V. M.; Barhoumi, A.; Perham, M.; Hartgerink, J. D.; Wittung-Stafshede, P.; Halas, N. J. “Gold nanoparticles can induce the formation of protein-based aggregates at physiological ph,” Nano Letters, 2009, 9, 666–671. 180. Yu, W. W.; Chang, E.; Falkner, J. C.; Zhang, J. Y.; Al-Somali, A. M.; Sayes, C. M.; Johns, J.; Drezek, R.; Colvin, V. L. “Forming biocompatible and nonaggregated nanocrystals in water using amphiphilic polymers,” Journal of the American Chemical Society, 2007, 129, 2871– 2879. 181. Chen, Y. J.; Bothun, G. D. “Lipid-assisted formation and dispersion of aqueous and bilayerembedded nano-C-60,” Langmuir, 2009, 25, 4875–4879. 182. Hirsch, A. “Amphiphilic architectures based on fullerene and calixarene platforms: From buckysomes to shape-persistent micelles,” Pure and Applied Chemistry, 2008, 80, 571–587. 183. Wang, X. S.; Metanawin, T.; Zheng, X. Y.; Wang, P. Y.; Ali, M.; Vernon, D. “Structuredefined C-60/polymer colloids supramolecular nanocomposites in water,” Langmuir, 2008, 24, 9230–9232. 184. Zhu, Z. X.; Anacker, J. L.; Ji, S. X.; Hoye, T. R.; Macosko, C. W.; Prud’homme, R. K. “Formation of block copolymer-protected nanoparticles via reactive impingement mixing,” Langmuir, 2007, 23, 10499–10504. 185. Goldberg-Oppenheimer, P.; Cosnier, S.; Marks, R. S.; Regev, O. “Preparation and characterization of a novel pyrrole-benzophenone copolymerized silica nanocomposite as a reagent in a visual immunologic-agglutination test,” Talanta, 2008, 75, 1324–1331. 186. Nilsson, L.; L¨of, D.; Bergenst˚ahl, B. “Phenolic acid nanoparticle formation in iron-containing aqueous solutions,” Journal of Agricultural and Food Chemistry, 2008, 56, 11453–11457. 187. Bednarova, L.; Brandel, J.; d’Hardemare, A. D.; Bednar, J.; Serratrice, G.; Pierre, J. L. “Vesicles to concentrate iron in low-iron media: An attempt to mimic marine siderophores,” Chemistry—A European Journal, 2008, 14, 3680–3686.

318

S. Zhong and D. J. Pochan

188. Ogura, T.; Shibata, H.; Sakai, K.; Sakai, H.; Abe, M. “Direct preparation of highly ordered multilayer-type silica nanocapsules using spontaneously formed vesicles as templates,” Chemistry Letters, 2009, 38, 120–121. 189. Lu, Y.; Mei, Y.; Schrinner, M.; Ballauff, M.; M¨oller, M. W. “In situ formation of ag nanoparticles in spherical polyacrylic acid brushes by UV irradiation,” Journal of Physical Chemistry C, 2007, 111, 7676–7681. 190. Schrinner, M.; Polzer, F.; Mei, Y.; Lu, Y.; Haupt, B.; Ballauff, M.; G¨oldel, A.; Drechsler, M.; Preussner, J.; Glatzel, U. “Mechanism of the formation of amorphous gold nanoparticles within spherical polyelectrolyte brushes,” Macromolecular Chemistry and Physics, 2007, 208, 1542–1547. 191. Schrinner, M.; Ballauff, M.; Talmon, Y.; Kauffmann, Y.; Thun, J.; M¨oller, M.; Breu, J. “Single nanocrystals of platinum prepared by partial dissolution of Au-Pt nanoalloys,” Science, 2009, 323, 617–620. 192. Dong, J. S.; Whitcomb, D. R.; McCormick, A. V.; Davis, H. T. “Crystallization of silver carboxylates from sodium carboxylate mixtures,” Langmuir, 2007, 23, 7963–7971. 193. Krichevski, O.; Levi-Kalisman, Y.; Szwarcman, D.; Lereah, Y.; Markovich, G. “Growth of au/ag nanowires in thin surfactant solution films: An electron microscopy study,” Journal of Colloid and Interface Science, 2007, 314, 304–309. 194. Schumacher, M.; Ruppel, M.; Yuan, J. Y.; Schmalz, H.; Colombani, O.; Drechsler, M.; M¨uller, A. H. E. “Smart organic-inorganic nanohybrids based on amphiphilic block copolymer micelles and functional silsesquioxane nanoparticles,” Langmuir, 2009, 25, 3407–3417. 195. Schumacher, M.; Ruppel, M.; Kohlbrecher, J.; Burkhardt, M.; Plamper, F.; Drechsler, M.; M¨uller, A. H. E. “Smart organic-inorganic nanohybrid stars based on star-shaped poly(acrylic acid) and functional silsesquioxane nanoparticles,” Polymer, 2009, 50, 1908–1917. 196. Yan, Y.; de Keizer, A.; Stuart, M. A. C.; Drechsler, M.; Besseling, N. A. M. “Stability of complex coacervate core micelles containing metal coordination polymer,” Journal of Physical Chemistry B, 2008, 112, 10908–10914. 197. Qi, L.; Chapel, J. P.; Castaing, J. C.; Fresnais, J.; Berret, J. F. “Organic versus hybrid coacervate complexes: Co-assembly and adsorption properties,” Soft Matter, 2008, 4, 577–585. 198. Klokkenburg, M.; Ern´e, B. H.; Wiedenmann, A.; Petukhov, A. V.; Philipse, A. P. “Dipolar structures in magnetite ferrofluids studied with small-angle neutron scattering with and without applied magnetic field,” Physical Review E, 2007, 75, 051408. 199. Hussain, I.; Brust, M.; Barauskas, J.; Cooper, A. I. “Controlled step growth of molecularly linked gold nanoparticles: From metallic monomers to dimers to polymeric nanoparticle chains,” Langmuir, 2009, 25, 1934–1939. 200. Klokkenburg, M.; Houtepen, A. J.; Koole, R.; de Folter, J. W. J.; Ern´e, B. H.; van Faassen, E.; Vanmaekelbergh, D. “Dipolar structures in colloidal dispersions of PbSe and CdSe quantum dots,” Nano Letters, 2007, 7, 2931–2936. 201. O’Connell, M. J.; Bachilo, S. M.; Huffman, C. B.; Moore, V. C.; Strano, M. S.; Haroz, E. H.; Rialon, K. L.; Boul, P. J.; Noon, W. H.; Kittrell, C.; Ma, J. P.; Hauge, R. H.; Weisman, R. B.; Smalley, R. E. “Band gap fluorescence from individual single-walled carbon nanotubes,” Science, 2002, 297, 593–596. 202. Bachilo, S. M.; Strano, M. S.; Kittrell, C.; Hauge, R. H.; Smalley, R. E.; Weisman, R. B. “Structure-assigned optical spectra of single-walled carbon nanotubes,” Science, 2002, 298, 2361–2366. 203. Tsyboulski, D. A.; Bakota, E. L.; Witus, L. S.; Rocha, J. D. R.; Hartgerink, J. D.; Weisman, R. B. “Self-assembling peptide coatings designed for highly luminescent suspension of singlewalled carbon nanotubes,” Journal of the American Chemical Society, 2008, 130, 17134– 17140. 204. Witus, L. S.; Rocha, J. D. R.; Yuwono, V. M.; Paramonov, S. E.; Weisman, R. B.; Hartgerink, J. D. “Peptides that non-covalently functionalize single-walled carbon nanotubes to give controlled solubility characteristics,” Journal of Materials Chemistry, 2007, 17, 1909–1915.

Cryogenic Transmission Electron Microscopy

319

205. Backes, C.; Schmidt, C. D.; Hauke, F.; B¨ottcher, C.; Hirsch, A. “High population of individualized swcnts through the adsorption of water-soluble perylenes,” Journal of the American Chemical Society, 2009, 131, 2172–2184. 206. Grunlan, J. C.; Liu, L.; Regev, O. “Weak polyelectrolyte control of carbon nanotube dispersion in water,” Journal of Colloid and Interface Science, 2008, 317, 346–349. 207. Hong, J. S.; Kim, C. “Extension-induced dispersion of multi-walled carbon nanotube in nonnewtonian fluid,” Journal Rheology, 2007, 51, 833–850. 208. Nativ-Roth, E.; Yerushalmi-Rozen, R.; Regev, O. “Phase behavior and shear alignment in swnt-surfactant dispersions,” Small, 2008, 4, 1459–1467. 209. Nativ-Roth, E.; Regev, O.; Yerushalmi-Rozen, R. “Shear-induced ordering of micellar arrays in the presence of single-walled carbon nanotubes,” Chemical Communications, 2008, 17, 2037– 2039. 210. Choi, J. H.; Nguyen, F. T.; Barone, P. W.; Heller, D. A.; Moll, A. E.; Patel, D.; Boppart, S. A.; Strano, M. S. “Multimodal biomedical imaging with asymmetric single-walled carbon nanotube/iron oxide nanoparticle complexes,” Nano Letters, 2007, 7, 861–867. 211. Yeon, S. H.; Seol, J.; Seo, Y. J.; Park, Y.; Koh, D. Y.; Park, K. P.; Huh, D. G.; Lee, J.; Lee, H. “Effect of interlayer ions on methane hydrate formation in clay sediments,” Journal of Physical Chemistry B, 2009, 113, 1245–1248. 212. Negrete-Herrera, N.; Putaux, J. L.; David, L.; De Haas, F.; Bourgeat-Lami, E. “Polymer/laponite composite latexes: Particle morphology, film microstructure, and properties,” Macromolecular Rapid Communications, 2007, 28, 1567–1573. 213. Li, H. Y.; Carter, J. D.; LaBean, T. H. “Nanofabrication by DNA self-assembly,” Materials Today, 2009, 12, 24–32. 214. LaBean, T. H.; Li, H. Y. “Constructing novel materials with DNA,” Nano Today, 2007, 2, 26–35. 215. Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Subramani, R.; Mamdouh, W.; Golas, M. M.; Sander, B.; Stark, H.; Oliveira, C. L. P.; Pedersen, J. S.; Birkedal, V.; Besenbacher, F.; Gothelf, K. V.; Kjems, J. “Self-assembly of a nanoscale DNA box with a controllable lid,” Nature, 2009, 459, 73–76. 216. Andersen, F. F.; Knudsen, B.; Oliveira, C. L. P.; Frohlich, R. F.; Kr¨uger, D.; Bungert, J.; Agbandje-McKenna, M.; McKenna, R.; Juul, S.; Veigaard, C.; Koch, J.; Rubinstein, J. L.; Guldbrandtsen, B.; Hede, M. S.; Karlsson, G.; Andersen, A. H.; Pedersen, J. S.; Knudsen, B. R. “Assembly and structural analysis of a covalently closed nano-scale DNA cage,” Nucleic Acids Research, 2008, 36, 1113–1119. 217. He, Y.; Ye, T.; Su, M.; Zhang, C.; Ribbe, A. E.; Jiang, W.; Mao, C. D. “Hierarchical self-assembly of DNA into symmetric supramolecular polyhedra,” Nature, 2008, 452, 198–201. 218. Zhang, C.; Su, M.; He, Y.; Zhao, X.; Fang, P. A.; Ribbe, A. E.; Jiang, W.; Mao, C. D. “Conformational flexibility facilitates self-assembly of complex DNA nanostructures,” Proceedings of the National Academy of Sciences U. S. A., 2008, 105, 10665–10669. 219. Braet, F.; Wisse, E.; Bomans, P.; Frederik, P.; Geerts, W.; Koster, A.; Soon, L.; Ringer, S. “Contribution of high-resolution correlative imaging techniques in the study of the liver sieve in three-dimensions,” Microscopy Research and Technique, 2007, 70, 230–242. 220. Joni´c, S.; Sorzano, C. O. S.; Boisset, N. “Comparison of single-particle analysis and electron tomography approaches: An overview,” Journal of Microscopy, 2008, 232, 562–579. 221. de la Presa, P.; Rueda, T.; Morales, M. D.; Chich´on, F. J.; Arranz, R.; Valpuesta, J. M.; Hernando, A. “Gold nanoparticles generated in ethosome bilayers, as revealed by cryo-electrontomography,” Journal of Physical Chemistry B, 2009, 113, 3051–3057. 222. Hoffmann, C.; Leis, A.; Niederweis, M.; Plitzko, J. M.; Engelhardt, H. “Disclosure of the mycobacterial outer membrane: Cryo-electron tomography and vitreous sections reveal the lipid bilayer structure,” Proceedings of the National Academy of Sciences U. S. A., 2008, 105, 3963–3967. 223. Luˇci´c, V.; Leis, A.; Baumeister, W. “Cryo-electron tomography of cells: Connecting structure and function,” Histochemistry and Cell Biology, 2008, 130, 185–196.

320

S. Zhong and D. J. Pochan

224. Subramaniam, S.; Bartesaghi, A.; Liu, J.; Bennett, A. E.; Sougrat, R. “Electron tomography of viruses,” Current Opinion in Structural Biology, 2007, 17, 596–602. 225. Sorzano, C. O. S.; Jonic, S.; Cottevieille, M.; Larquet, E.; Boisset, N.; Marco, S. “3D electron microscopy of biological nanomachines: Principles and applications,” European Biophysics Journal, 2007, 36, 995–1013. 226. Yuan, P.; Liu, N.; Zhao, L. Z.; Zhou, X. F.; Zhou, L.; Auchterlonie, G. J.; Yao, X. D.; Drennan, J.; Lu, G. Q.; Zou, J.; Yu, C. Z. “Solving complex concentric circular mesostructures by using electron tomography,” Angewandte Chemie International Edition, 2008, 47, 6670–6673. 227. Norl´en, L.; Masich, S.; Goldie, K. N.; Hoenger, A. “Structural analysis of vimentin and keratin intermediate filaments by cryo-electron tomography,” Experimental Cell Research, 2007, 313, 2217–2227. 228. Taveau, J. C.; Nguyen, D.; Perro, A.; Ravaine, S.; Dugue, E.; Lambert, O. “New insights into the nucleation and growth of PS nodules on silica nanoparticles by 3D cryo-electron tomography,” Soft Matter, 2008, 4, 311–315. 229. Parry, A. L.; Bomans, P. H. H.; Holder, S. J.; Sommerdijk, N.; Biagini, S. C. G. “Cryo electron tomography reveals confined complex morphologies of tripeptide-containing amphiphilic doublecomb diblock copolymers,” Angewandte Chemie International Edition, 2008, 47, 8859–8862.

Polymer Reviews, 50:321–339, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.493256

Advances in the Transmission Electron Microscopy of Polymers MATTHEW R. LIBERA1 AND RAY F. EGERTON2 1

Department of Chemical Engineering and Materials Science, Stevens Institute of Technology, Hoboken, New Jersey 2 Department of Physics, University of Alberta, Edmonton, Canada Transmission electron microscopy (TEM) of polymers involves the problem definition and methodologies associated with the microscopy of both inorganic and biological materials but cannot be categorized within either of these fields alone. On the one hand, like other synthetic materials, polymers offer the ability to control properties through synthesis and processing, and TEM is a powerful method with which to provide information within the synthesis–structure–property paradigm of materials science and engineering. The well-established techniques of bright/dark-field imaging, electron diffraction, high-resolution imaging, and analytical microscopies are thus all used to study polymers. On the other hand, the electron–specimen interactions are more like those in biological systems. Synthetic polymers and biological materials consist largely of light elements whose elastic interactions with energetic electrons are relatively weak. Generating image contrast can thus be a challenge in polymer TEM. The inelastic electron/soft material interactions are, however, relatively strong. These provide for powerful spectroscopies but also lead to radiation damage. The constraints that damage puts on imaging are far more stringent in polymers than in inorganic systems. This review highlights ongoing advances in contrast generation exploiting both elastic and inelastic electron–polymer interactions and outlines the salient issues determining the achievable spatial resolution in radiation-sensitive materials. Keywords polymer, TEM, electron microscopy, radiation damage, image contrast, image resolution, holography, phase contrast, stain

1. Introduction The transmission electron microscope (TEM) has been used extensively for decades to study the morphology of synthetic polymers. It continues to be an essential tool, particularly because of the inherent ability of these materials to hierarchically structure themselves over multiple length scales and produce morphologies that lead to advances in functionality and application. Significantly, instrumentation advances over the past decade have made both established and emerging TEM imaging methods more accessible to an ever-growing number of applications-oriented research groups around the world, and increasing numbers of scientists and engineers are exploiting these methods to quantify polymer morphology.

Received February 10, 2010; accepted April 15, 2010. Address correspondence to Matthew Libera, Department of Chemical Engineering and Materials Science, Stevens Institute of Technology, 1 Castle Point Terrace, Hoboken, NJ 07030. E-mail: [email protected]

321

322

M. R. Libera and R. F. Egerton

There have historically been two central challenges associated with studying polymers in the electron microscope. First, because they are predominantly comprised of low-atomicnumber elements, soft materials display an intrinsically low level of electron-optical image contrast. Second, soft materials tend to be very sensitive to the ionizing radiation used in an intermediate-energy electron microscope and can suffer both chemical and structural change—beam damage—during the microstructural measurement. These two issues continue to provide an incentive to the development of new imaging techniques. A number of excellent and modern monographs have been written on transmission electron microscopy and its application to materials1–3 and two important ones that focus primarily on polymers.4,5 This review highlights some of the most advanced work over the past decade, concentrating on new methods to generate contrast, advances in our understanding of radiation damage in the TEM, and the increasingly important topic of dose-limited spatial resolution.

2. Contrast Mechanisms Some of the basic contrast mechanisms afforded by electron–specimen interactions are outlined schematically by Figure 1. A subset of soft materials exhibits spatial variations in crystallinity, crystal orientation, or density due to the presence of heavy elements. Crystallinity gives rise to Bragg diffraction, and heavy elements can induce significant Rutherford scattering. In both cases, some incident electrons are scattered to relatively high angles where they can be blocked by an objective aperture (Figure 1A), thus producing dark contrast in a final image. In many cases, however, soft-material specimens are amorphous and do not exhibit significant spatial variations in density. Thus, scattering is primarily in the forward direction, and an objective aperture provides little or no contrast. Nevertheless, the rich valence electron structure of soft materials can introduce spatial modulations in the phase or energy of an incident electron wave (Figure 1B). These provide two alternate sources of contrast for imaging soft-materials morphology, and much of the recent and ongoing technique-development work concentrates on phase contrast and spectroscopic

Figure 1. (A) Spatially modulated crystallinity or heavy element stain distribution provides for amplitude contrast when electrons scattered to high angles are blocked by the objective aperture. (B) Changes in electron phase or in electron energy are alternative sources of contrast in amorphous soft materials with small variations in density.

Advances in the Transmission Electron Microscopy of Polymers

323

imaging methods. The following few sections outline and review some of the important aspects of these various imaging methods. 2.1. Bragg Diffraction Diffraction-contrast imaging techniques well established in crystalline inorganic systems have been used extensively to study crystalline and semicrystalline polymers. These have recently been discussed by Michler.4 High-resolution imaging methods, based on the interference of two or more Bragg-scattered beams, have been described by Martin et al.6,7 2.2. Heavy Element Stains Heavy element staining coupled with bright-field imaging has been used for decades to study polymer morphology.5,8 This method takes advantage of the fact that certain elements or compounds will preferentially react with or localize themselves in some particular component of a polymeric material. Osmium tetroxide, for example, reacts with unsaturated carbon–carbon bonds, whereas ruthenium tetroxide preferentially reacts with aromatic rings. Figure 2 illustrates the significant contrast induced by OsO4 staining of a microphase-separated nanoparticle blend of homopolymer polystyrene (PS) and polystyrene-polybutadiene-polystyrene (SBS) triblock copolymer. Other common stains include iodine, uranyl acetate, and phosphotungstic acid. All of them to varying degrees tend to preferentially localize within specimen regions of lower density such as the amorphous parts of semicrystalline polymers.9 The nature and use of these and other stains have been reviewed by Sawyer et al.5 and by Smith and Bryg.8 All of these stains induce contrast by preferentially scattering electrons elastically to high angles. For a uranium atom, the median angle for elastic scattering of 100 keV electrons is about 50 mrad, so a small objective aperture will intercept almost all of these electrons, making uranium-stained regions much more opaque than any surrounding unstained specimen regions. Because of the various complex morphologies that can be displayed over a range of length scales by and, indeed, engineered into, polymeric materials, three-dimensional imaging using transmission electron tomography has been of growing importance. Tomographic imaging is particularly useful in studies of block copolymers where microphase separation occurs at nanolength scales to produce interpenetrated three-dimensional morphologies difficult to analyze by traditional two-dimensional techniques. Much of this work can take great advantage of stained specimens, which produce high contrast and are relatively insensitive to electron-beam damage. Tomography produces a three-dimensional image by reconstructing a large number of two-dimensional images collected from the same specimen but from a large range of different angles.10 Some of the first applications to synthetic polymers were made by Spontak et al.11,12 in the mid 1990s. The integration of powerful computer systems and fully motorized specimen stages on the current generation of microscopes has made the technique far more accessible and its use in polymer-morphology studies continues to accelerate.13 Despite the fact that staining methods have been and will no doubt continue to be used by polymer microscopists, there are several shortcomings associated with their use. Among these is the fact that they are inherently qualitative. Often, a qualitative understanding of morphology is adequate, but sometimes it is not. In addition, stains can introduce artifacts such as inorganic nanostructures14 or nonlinear decoration of interfaces,15 and care must often be exercised when interpreting stain-induced contrast. Perhaps most importantly,

324

M. R. Libera and R. F. Egerton

Figure 2. Bright-field TEM (200 keV) images of the same SBS triblock copolymer–PS homopolymer nanoparticle blend on a carbon support film: (A) unstained, near focus; (B) unstained at −35 µm defocus; (C) OsO4 stained at focus.

however, is the simple fact that there are no good stains for many polymer systems. The various components in some systems often have similar structures and chemistries, limiting the extent to which a particular stain can preferentially label one or more of these components in an interpretable manner. Furthermore, there is increasing interest in studying materials under conditions related to their synthesis or use, such as solvation or hydration, and most staining protocols have limited utility under such conditions. Consequently, there is interest in exploring alternative imaging methods that forego the use of stains and instead generate contrast based on the intrinsic interaction of incident electrons with the polymer specimen itself. 2.3. Phase Contrast In the absence of stains, one alternative source of contrast lies in differential modulations of the electron phase. Phase modulation occurs because the positive Coulombic potential of an atomic nucleus accelerates the incident electron wave. The Columbic potential can be

Advances in the Transmission Electron Microscopy of Polymers

325

represented as a Fourier series, and the orientation-independent zeroth-order term in this series is referred to as the mean inner potential, 0 . An electron-optical index of refraction, neo , can be defined as: neo = 1 +

eo Eo + E E 2Eo + E

(1)

where e is the magnitude of the electron charge, E is the accelerating energy, and E0 is the electron rest energy. Relative to a reference wave that would pass through vacuum rather than through a specimen, and in the absence of dynamical-scattering effects, the electron phase is shifted by an amount: ϕ = 2π (neo − 1) (t/λ)

(2)

where t is the specimen thickness and λ is the electron wavelength. Specimen-dependent spatial variations in 0 will thus give rise to spatial variations in φ and thus to phase contrast. A lot of work has been done related to models and measurements of the mean inner potential for various materials, much of which has concentrated on inorganic solids. Gadjardiska-Josifovska and Carim have provided an excellent overview of much of this work.16 0 varies from approximately 5 to 30 V. It has been measured to be 8.5 V in polystyrene,17 9.09 V in amorphous carbon,18 and 5.9 V and 5.4 V in anthracene and naphthalene,19 respectively. Though the magnitude of the mean inner potential is in great measure determined by the nuclear charge, subtle variations in 0 are related to the redistribution of valence electrons and this effect is the source of phase contrast in many organic polymers and other soft materials. In the absence of thickness differences or significant variations in density, organic materials provide a range of possible valence states. At one extreme is carbon in the fully saturated state, such as found in polyethylene; the other extreme is carbon in a highly conjugated state with delocalized electron states, such as found in polythiophene or polyanilene. Because the information recorded on film or on a charge-couple device (CCD) camera corresponds to the modulus squared of the exit-face wave function, modified by the imaging lenses of the microscope, phase information is lost. The three main techniques used to convert phase information into amplitude information are defocusing, holographic methods, and methods that use some form of Zernike phase plate. All three approaches have been explored in the context of imaging the morphology of amorphous multiphase polymers. Defocusing is a classic technique that employs defocus on the order of 10 µm or more to force the objective-lens transfer properties to convert phase information into amplitude information. It is a method well established in biological systems. It was applied to polymers in the mid 1970s by Petermann and Gleiter20 and used extensively by Petermann until 2002.21 In the early 1980s Thomas et al. applied defocus phase-contrast imaging to a variety of different polymers including polystyrene-polyisoprene (PS-PI) block copolymer systems.22–25 The technique has recently been discussed by Simon et al.26 Figure 2B shows that a defocus of about 35 µm is able to generate phase contrast in the SBS-PS nanoparticle blend. The nature of this contrast, however, bears little direct correspondence to the nanoscale morphology manifested by OsO4 staining (Figure 2C). Defocus imaging has not been extensively used in polymer studies for two reasons. First, the achievable resolution degrades as the defocus increases. Second, the transfer properties of the objective lens become very nonlinear at high defocus and can thus introduce contrast inversions that complicate image interpretation.

326

M. R. Libera and R. F. Egerton

A second method to study polymers and soft materials by phase-contrast imaging is off-axis transmission electron holography.27,28 In contrast to the defocus method, holography can be done with the specimen either in focus or very close to focus. It was first conceived of by Dennis Gabor in 1947 to improve the resolution of a TEM. Based on the interference between an object wave that passes through a specimen and a reference wave that passes through the vacuum adjacent to the specimen, holography is a technique particularly well suited to phase-contrast imaging. The advent of coherent light sources—lasers—in the early 1960s enabled great progress in light-based holography. Electron holography did not make much impact until the late 1980s when coherent electron sources—field-emission gun (FEG) sources—became commercially available and thus accessible to many researchers. It has found some good success imaging magnetic and electrostatic fields. First applied to polymers in the mid-1990s,29–31 holography has been used to study a variety of polymeric systems over the past decade.26,32–36 Significantly, the Lichte group has shown that holographic phase-contrast imaging can differentiate phase shifts generated by the subtle differences in mean inner potential between polystyrene and polyisoprene in an SIS block copolymer.26 Though holography can provide for both high-resolution and quantitative imaging, it has seen relatively little application to polymers. This is in part due to the need for somewhat specialized instrumentation, including a field-emission microscope and an electron biprism. In addition, to obtain an unmodulated reference wave that travels through a vacuum means that only regions of specimen within a few micrometers of an edge can be studied. Even more limiting, however, is the fact that the field of view is limited to hundreds of nanometers when using the standard objective lens and to ∼1–2 µm when using a so-called post-specimen Lorentz lens. Consequently, though holographic imaging may prove particularly useful for imaging polymeric nanostructures, it currently is less useful for studying the mesoscale structure that is so important in a great majority of polymer applications. A third method for phase-contrast imaging involves the use of a post-specimen phase plate inserted at the back-focal plane of the objective lens. The phase plate can take the form of a uniform film of conducting material (e.g., carbon) of suitable thickness, usually chosen to change the phase of scattered electrons by π /2, with a small axial hole such that the phase of the forward-scattered beam in unaltered.37 Alternatively, it can be a threeelectrode structure acting as a miniature einzel lens.38,39 In both cases, problems have arisen from hydrocarbon contamination and/or charging of the phase plate, leading to unstable properties. In addition, the central hole is often not small enough, so that information of low spatial frequency is lost. However, research on phase plates continues because of the potential to provide adequate image contrast at lower dose.40 As with holography, the specimen can be sharply focused, and resolution is therefore optimized. Figure 3, for example, shows a phase-contrast image, taken at focus, of an unstained lamellar PS-PI diblock copolymer.41 Note that this image shows adequate contrast to resolve the two phases, but by itself it does not enable one to determine which phase corresponds to the PS and which to the PI. 2.4. Spectroscopic Contrast The inelastic interactions between energetic electrons and materials provide an alternative source of contrast for quantitative mapping of composition, chemistry, and the distribution of phases in materials without the need for heavy-element stains. These interactions can be analyzed by electron energy-loss spectroscopy (EELS).42–44 Significantly, the past two decades have seen substantial improvements in both the hardware used to collect EELS

Advances in the Transmission Electron Microscopy of Polymers

327

Figure 3. Phase contrast image of an unstained lamellar polystyrene–polyisoprene diblock copolymer (PS-b-PI) collected using a 300-kV microscope equipped with a semicircular phase plate (after Tosaka et al.41).

data and the software used to analyze it, so spectroscopic imaging has become far more accessible to applications-oriented engineers and scientists. Electron energy-loss spectroscopy in the TEM is particularly well suited for analyzing energy losses in the sub-keV regime. This range includes K-shell core excitations from carbon (284 eV), oxygen (532 eV), and nitrogen (401 eV), as well as L-shell and M-shell excitations from a wide range of higher atomic number elements.44 Most soft materials have high concentrations of carbon, nitrogen, oxygen, and hydrogen, and the electronic structure associated with macromolecular compounds of these elements leads to distinctive fine structure in the ionization edges. Often, different polymer phases can be distinguished from each other based on their relative compositions. Significantly, the multiple hybrid bonding states characteristic of carbon, in particular, provide a rich valence-electron fine structure in soft materials. Low-loss spectra—sampling energy losses between a few eV and about 50 eV where valence states manifest themselves—provide a second type of characteristic fingerprint that can be used to differentiate between different polymer phases based on their relative chemistries. The magnitudes of inelastic electron scattering cross sections scale inversely with the energy loss E. Consequently, the spectroscopic intensity associated with low-loss spectra is orders of magnitude higher than that associated with core-loss spectra. Under conditions of dose-limited resolution typical of radiation-sensitive soft materials, as discussed below, a substantially higher signal and better spatial resolution can often be achieved using low-loss rather than core-loss spectroscopy. Imaging based on spectroscopic contrast can be performed in either scanning transmission electron microscope (STEM) mode using spectrum imaging45,46 or in TEM mode using energy-filtering (EFTEM) techniques.47,48 EELS spectrum imaging collects a continuous energy-loss spectrum over a range of energies (e.g., 0 to 80 eV loss) at each pixel in a 2D digital raster over a specimen area. Having complete spectra brings advantages to background modeling and fitting. In addition, all of the inelastically scattered electrons in the collection range can contribute to the imaging, so the incident dose is used most efficiently. In contrast, energy filtering uses either an in-column or post-column filter to collect images using electrons from a defined energy window, often 5–10 eV wide. EFTEM

328

M. R. Libera and R. F. Egerton

has repeatedly proven powerful for so-called zero-loss imaging, where most inelastically scattered electrons are filtered out using a slit centered on the zero-loss peak. It is similarly useful for core-loss imaging to map elemental distribution but is less useful for mapping chemical variations, because these require an energy resolution of a few eV or less. Because a large fraction of inelastic electrons is filtered from the final image, the incident dose is used less efficiently than in spectrum imaging mode. Energy-loss spectroscopy and its application to polymers has been reviewed by Du Chesne49 and by Libera and Disko,50 and over the past decade there has been steady interest in the use of this technique. It has, for example, been used extensively, both in core-loss and low-loss modes, to study several aspects of multiphase polymer morphology in various types of blends, composites and block copolymers.51–60 In addition, spatially resolved EELS has been applied to studies of a variety of polymer–polymer and polymer–inorganic interfaces.61–68 By incorporating cryo-TEM techniques that have been well established in the biological microscopy community and by exploiting the fact that solvents can have characteristic spectral fingerprints substantially different than those of many polymers,69,70 spatially resolved EELS has also been applied to a number of frozen-hydrated71–74 and frozensolvated materials.75,76 Figure 4 provides an example of the problem-solving power of this approach. It summarizes results related to the colloidal synthesis of biphasic poly(dimethyl siloxane) [PDMS]-acrylate copolymer nanoparticles.71,74 Figure 4A shows that the water, PDMS, and acrylate low-loss EELS spectra can be distinguished from each other. Water, for example, displays a sharp onset circa 9 eV loss, which has been attributed to excitation across the band gap69,77 and which effectively differentiates water spectroscopically from many soft materials. Component mapping by cryo-spectrum imaging (Figures 4B–D) shows that acrylate monomer dissolves in the PDMS host precursor. The formation of lobed-type emulsions via the copolymerization of acrylate monomers in a PDMS seed emulsion was confirmed using a high-angle annular-dark-field (HAADF) STEM image (Figure 4E), and subsequent component mapping (Figure 4F) shows the formation of the biphasic nanoparticle with PDMS-rich and polyacrylate-rich lobes. Significantly, however, the copolymer-rich lobe in this biphasic colloid consists of almost pure organic copolymer, whereas the PDMS-rich lobe is partly mixed with the copolymer. This result indicates that the interphase boundary is highly diffuse, a finding that provides important insight into the synthesis–structure–property relationships in this particular polymer system. One instrumentation development that will no doubt have a significant impact on the contrast associated with spectroscopic imaging of polymers is the commercial availability of monochromators.78–84 Typically, the energy resolution associated with EELS-based methods is limited by the energy spread of the electron source, represented by the full-width at half-maximum of the zero-loss peak in an energy-loss spectrum. The energy spread of a Schottky thermally assisted field-emission gun (FEG) electron source is typically about 0.6–1.0 eV. Cold field emission sources have a narrower spread (∼0.3–0.4 eV) due to their lower temperature, whereas thermionic sources (W or LaB6 ) have energy spreads of about 1.5 eV or more. Monochromators filter the emitted electrons at the price of a reduced incident current, and energy resolutions on the order of 0.1 eV can be achieved, sufficient to resolve substantially more fine structure in energy-loss spectra. Figure 5 presents spectra collected from polystyrene using monochromated and unfiltered electrons. The enhanced energy resolution enables the low-energy shoulder (∼5.5 eV) on the π −π ∗ peak (∼7 eV) as well as a series of interband transitions on the rising edge of the bulk plasmon peak to be detected. Early work using monochromated electron sources was performed by the Boersch group, with sub-10-meV energy resolution,79,85 The Ritsko et al.86,87 and Fink et al.88,89 studies were not done using an electron microscope and, hence, lacked spatial resolution.

Advances in the Transmission Electron Microscopy of Polymers

329

Figure 4. (A) Low-loss EELS spectra; component maps before polymerization for (B) acrylate monomer, (C) PDMS, and (D) amorphous ice (water); (E) high-angle annular dark-field (HAADF) cryo-STEM image of biphasic particle after polymerization; and (F) component maps for polyacrylate (right) and PDMS (left) after polymerization (after Kim et al.74). (Figure available in color online)

These studies do, however, suggest that monochromatic energy sources will be able to reveal substantial fine structure in both low-loss and core-loss EELS spectra. In addition, the past two decades have seen substantial work using ultraviolet photoelectron spectroscopy (UPS),90–94 which can probe the valence electron structure up to binding energies of about 40 eV, again with energy resolution on the order of 0.05–0.35 eV. These studies also resolve a rich electronic structure in a variety of different polymeric materials. Such spectral information can not only provide new and better fingerprints to give sharper contrast in spectroscopic imaging studies but new insights to electronic structure, particularly at interfaces in a range of emerging soft electronic devices. Electron monochromators are being increasingly used in studies of inorganic materials but have not yet seen much application to polymeric materials. Instrumentation that can provide <30-meV energy resolution with sub-nanometer spatial resolution is planned for commercial production.84

330

M. R. Libera and R. F. Egerton

Figure 5. Electron energy-loss spectra from polystyrene. Top: Collected using broad-beam 80-keV incident electrons with 0.11-eV energy resolution (after Ritsko and Bigelow86). Bottom: Collected using a focused beam of 200-keV incident electrons with 0.8-eV energy resolution (see Yakovlev and Libera113).

3. Radiation Damage Soft materials are well known to undergo structural and chemical change when irradiated with energetic electrons. Collectively, these effects are known as radiation damage. Damage is not unique to polymers and biological tissue but is prominent in these soft materials. Metals and semiconductors, by contrast, are relatively insensitive to intermediate-energy electron irradiation. Ceramics, broadly speaking, lie somewhere in between. The fact that electron microscopes have been used for several decades to characterize soft materials with great success is an important indicator that this tool is one that can answer important morphological questions despite the concerns over radiation damage. Indeed, radiation damage is rarely an issue in studies of stained specimens, because stains tend to stabilize the structure and distribute the deposited energy over larger volumes of material. Furthermore, in studies of unstained specimens at low resolution the incident electron dose can be distributed over relatively large volumes so that the fraction of chemically or structurally modified material remains relatively small. In addition, materials with intrinsic periodicity—e.g., crystalline polymers—lend themselves to signal averaging so the noisy data collected at low incident doses can be combined to create images with high resolution. Variations of this approach have been responsible for many of the successes in macromolecular electron crystallography enjoyed by the biological community.95–100 The most challenging soft-materials imaging problems are those on unstained amorphous or aperiodic materials such that data are collected at the highest possible spatial resolution. This is the most general type of problem and is one of increasing importance technologically. At this limit, careful experimentation requires that the effects of radiation damage

Advances in the Transmission Electron Microscopy of Polymers

331

be monitored to ensure that the morphological effects being studied are intrinsic to the material and not a consequence of the damage imposed by the experiment itself. There are two general types of electron–specimen interactions that contribute to damage. One type involves high-angle elastic scattering by atomic nuclei, which can transfer sufficient energy to the specimen to cause atomic displacement. Known as knock-on or displacement damage, it occurs when the incident energy is above some threshold value that depends on the atomic number and binding energy of the atoms in the specimen. Low-density materials are particularly susceptible to knock-on damage, because of their relatively low atomic mass. The mean displacement energy of carbon in molecular form is approximately 5 eV, and this much energy can be transferred in a head-on (180-degree) collision by incident electrons of energy 30 keV or more.101 In comparison, the threshold energy to displace an atom in crystalline silicon is 145 keV. A second important type of electron–specimen interaction involves inelastic scattering of incident electrons by atomic electrons. Significantly, the cross sections for both elastic and inelastic electron scattering increase with decreasing incident electron energy. Thus, despite the fact that the effects of knock-on damage due to elastic scattering decreases with decreasing electron energy, the damage due to inelastic scattering increases. Some of the transferred energy is dissipated as heat and under certain circumstances, such as high dose rate and broad-beam illumination of a low-conductivity material, can give rise to an appreciable temperature rise. Calculations suggest, however, that in many practical cases, particularly those involving STEM imaging with a focused electron probe, the temperature rise is limited to a few Kelvin. Often the most significant consequence is mechanical instability of the specimen due to the thermal stresses associated with nonuniform temperature. In the case of low-Tg polymers, however, one must also be aware of beam-induced changes in viscosity and possible local flow. A more profound consequence of inelastic scattering is that it leads to chemical change in the specimen. The same interactions that are exploited by electron energy-loss spectroscopy to assess chemistry and composition, for example, lead to ionization and bond breaking. The underlying reason is related to the fact that, because most polymers are electrical insulators, there is no ready source of free electrons to saturate a radiation-induced free radical in a timescale shorter than that required for other chemical processes to occur. Among these alternative processes are polymeric chain scission, i.e. depolymerization, or crosslinking, or atomic/molecular motion leading to mass loss. Often these effects can happen simultaneously in a given specimen, because (1) different chemical moieties respond to energetic radiation differently, as manifested by the G-value characterization of radiation sensitivity,102 (2) the local structural environment can influence whether and how atoms or molecular fragments move, and (3) the local chemical environment can affect the nature of the possible chemical processes that might occur after ionization. Soft materials having high concentrations of aromatic moieties tend to be somewhat more radiation resistant than fully saturated polymers. This property is usually attributed to the delocalization associated with highly conjugated bonds that provide a source of electron charge and the means to distribute the adsorbed energy over a larger volume of material. However, the situation is very different than that, say, for a good metal where there is a vast sea of valence electrons that can fill electron states created by an ionization event and do so on a timescale faster than many of the possible secondary processes that might otherwise occur. Nevertheless, it has been suggested that damage to highly aromatic compounds may require K-ionization of a carbon atom, a relatively dramatic event that leads to Auger emission and a loss of two electrons from the same atom. If so, one might expect damage to cease below some threshold incident energy related to the carbon K-ionization energy

332

M. R. Libera and R. F. Egerton

(285 eV) and there is some evidence for this,103,104 although the interpretation has been questioned.105 Practically speaking, radiation damage in soft materials manifests itself in ways that can usually be measured as a function of the incident dose. Chemical changes can be monitored spectroscopically. Changes in crystallinity can be monitored by the fading of intensity in particular Bragg reflections. Mass loss can be monitored by changes in transmission, following a Beers law–type approach. Which of these mechanisms of damage is most important depends on the nature of the morphological questions being asked. Chemical changes and loss of crystallinity tend to both follow an exponential decay in signal as a function of dose: D (3) I (D) = I exp − Dc where Dc is the so-called critical dose (or fluence) at which the signal intensity has fallen to 1/e of that characteristic of the unirradiated material. Measuring and reporting Dc for a given experiment is an important part of a careful soft-materials imaging experiment, because it provides a reference for determining the extent to which the doses used during spectral or image data collection cause damage. Among the unresolved physics associated with radiation damage in soft materials are the delocalized effects that occur tens of nanometers or more from the point of electron incidence. This phenomenon increases the measured critical dose and might, for a point analysis, improve the achievable spatial resolution. First reported by Varlot et al.106 in 1999 in poly(ethylene terepthalate), it was confirmed by Siangchaew and Libera in polystyrene.107 In both cases, the dose-dependent decay in the π −π ∗ peak associated with aromatic rings was measured by electron energy-loss spectroscopy. When the probe size was decreased from a diameter on the order of 1 µm to a diameter on the order of a few nanometers, the apparent critical dose increased by several orders of magnitude. Fast secondary electrons generated by inelastic electron specimen interactions can travel within the plane of the thin specimen, radially away from the incident beam, carrying energy that causes delocalized damage,107 but their range and number appear to be insufficient to account for the observed increase in Dc . These measurements were based on the 6-7 eV energy-loss peak associated with the π −π ∗ excitation and use of the Rayleigh-type formula, L ∼ 0.6λ/θ , where θ ∼ (0.5E/E0 )3/4 is a median scattering angle,77 gives a delocalization distance L ∼ 4 nm for E0 = 200 keV. An additional contribution to the large Dc values may thus arise from detection of material outside the incident beam that had not yet been damaged by secondary electrons, but this detection involves energy transfer and presumably some damage.

4. Resolution Limits The limit of spatial resolution in the imaging of polymer morphology using energetic radiation sources is one of increasing interest and importance. This resolution limit is very different from that associated with the imaging of less radiation-sensitive materials such as semiconductors, metals, and many ceramics. In the latter case, the achievable resolution is determined by the quality of the microscope optics, most notably by the spherical aberration of the objective lens, and substantial improvements in resolution have recently been achieved by the development of next-generation aberration correctors108–111 and by holography. The so-called point resolution is a well-defined quantity that can be measured and used to reproducibly specify the achievable spatial resolution associated with a particular

Advances in the Transmission Electron Microscopy of Polymers

333

Figure 6. The Rose criterion defines the minimum feature size, d, which contains N = N counts of intensity, that can be distinguished from the surrounding background with an intensity level of N counts.

microscope and its operating conditions. Achieving this resolution, however, requires relatively high doses of incident radiation. Most inorganic materials can withstand these doses, but most polymers cannot. Consequently, the traditional specification of a microscope’s point resolution, or in STEM mode its probe size, is not very useful. Instead one needs to ask what is the minimum dose to which a particular material can be exposed and generate a sufficient signal-to-noise ratio (SNR) so adjacent pixels can be distinguished. This is a well-known problem, first addressed in the 1940s by Albert Rose in the context of television technology112,113 and extensively discussed in the context of electron microscopy.101,103–105 Some of the insights and limitations of this approach have been discussed by Burgess.114 Rose found that a pattern can be distinguished from a noisy background if the SNR is greater than about 5. We will apply this concept to a single pixel whose lateral dimension is d and whose image intensity, as specified by the number of incident electrons, is N. This intensity differs by an amount N from the background level in adjacent pixels. The contrast ratio is then C = N/N (Figure 6). If we assume that the noise involved is electron-beam shot noise, which is equal to N 1/2, then the SNR = N/N 1/2 = C N 1/2. In order to maximize SNR, we need to increase N until radiation damage threatens to destroy the structure of the specimen, giving N ∼ d2Dc /e where Dc is the critical dose, and e is the electron charge. Then the Rose criterion becomes Cd(Dc /e)1/2 = SNR > 5 so that the smallest useful pixel size (the radiation-limited spatial resolution) is d = (5/C)(e/Dc )1/2. For 20% contrast (C = 0.2) and Dc = 0.01 C/cm2 (typical of many polymers), d ∼ (25) ∗ (Dc /e)−1/2 = 1 nm. For a direct-exposure detector that is capable of counting single electrons,115 this estimate may be realistic. However, most current detectors use a scintillator to convert the electrons into photons, introducing an additional shot-noise term, and the photon detector is a CCD array whose output contains readout noise. Leakage current in the diodes is removed by subtracting the output recorded in the absence of electrons, and this darkcurrent subtraction increases signal contrast but increases the noise content. Individual CCD diodes have slightly different gain, so the resulting fixed-pattern noise is minimized through division by a flat-field response, but this process again increases the statistical noise. Sousa et al.72 measured the standard deviation in the intensity recorded by a CCD detector on a Gatan (Pleasanton, CA) Enfina EELS spectrometer and found it equal to 2.6(N)0.48, which would increase the smallest useful pixel size to about 2 nm under the above conditions.

334

M. R. Libera and R. F. Egerton

In general, the noise performance of an electron detector is described in terms of a detective quantum efficiency (DQE). The actual signal/noise ratio is given by SNR = 1 (DQE) 2 (SNR)ideal , where (SNR)ideal is the ideal signal/noise ratio determined by electronbeam shot noise. A more general expression for the damage-limited spatial resolution is therefore: d = (5/C)(e/Dc )1/2 (DQE)−1/2

(4)

This equation presents a useful means with which to estimate the achievable dose-limited resolution. One caveat is that an exposure to a dose as large as Dc involves substantial damage to the specimen, suggesting that the value of d given by Eq. (4) should be regarded as a lower limit. On the other hand, taking SNR = 5 is probably conservative of the signal-to-noise ratio needed to differentiate adjacent pixels.

5. Conclusion and Outlook The study of polymeric materials using the transmission electron microscope combines challenges and rewards, both in terms of the morphological problems that can be solved as well as the nature of how method and instrument development must occur to better study this class of material. In addition to the well-established and important approach of using heavy element stains to induce contrast between different morphological features, substantial advances have been made over the past two decades in new stain-free imaging methods based either on phase contrast or on spectroscopic contrast. This work is ongoing and will have even greater impact as these new methods become more user friendly and attract the attention of applications-oriented researchers. The damaging effects of the incident ionizing radiation are an ever-present concern, particularly in unstained materials. These effects should not be thought of as a barrier to meaningful morphological measurements in the TEM but rather as an important constraint on how to perform an imaging experiment and how to meaningfully interpret the data.

Acknowledgements The authors thank Aaron Kuo and Dr. Alex Chou of Stevens for their help with Figure 2 and Dr. Ginam Kim of Dow Corning for his help with Figure 4. M. Libera gratefully acknowledges support from the United States Army Research Office under grant number W911NF-07-1-0543. Ray Egerton acknowledges funding from the Natural Sciences and Engineering Research Council of Canada.

References 1. Williams, D. B.; Carter, C. B. Transmission electron microscopy: A textbook for materials science; Plenum: New York, 1996. 2. Egerton, R. F. Physical principles of electron microscopy: An introduction to TEM, SEM, and AEM; Springer: New York, 2005. 3. Reimer, L.; Kohl, H. Transmission electron microscopy: Physics of image formation, 5th ed.; Springer: New York, 2008. 4. Michler, G. H. Electron microscopy of polymers; Springer: Berlin, 2008. 5. Saywer, L. C., Grubb, D. T.; Meyers, G. F. Polymer microscopy, 3rd ed.; Springer: New York, 2008.

Advances in the Transmission Electron Microscopy of Polymers

335

6. Martin, D. C.; Thomas, E. L. “Experimental high-resolution electron microscopy of polymers,” Polymer, 1995, 36, 1743–1759. 7. Martin, D. C.; Chen, J.; Yang, J.; Drummy, L. F.; Kubel, C. “High resolution electron microscopy of ordered polymers and organic molecular crystals: Recent developments and future possibilities,” Journal of Polymer Science, Part B: Polymer Physics, 2005, 43, 1749–1778. 8. Smith, R. W.; Bryg, V. “Staining polymers for microscopical examination,” Rubber Chemistry and Technology, 2006, 79, 520–540. 9. Haubruge, H. G.; Jonas, A. M.; Legras, R. “Staining of poly(ethylene terephthalate) by ruthenium tetroxide,” Polymer, 2003, 44, 3229–3234. 10. Midgely, P. A.; Weyland, M. “3D electron microscopy in the physical sciences: The development of Z-contrast and EFTEM tomography,” Ultramicroscopy, 2003, 96, 413–431. 11. Spontak, R. J.; Williams, M. C.; Agard, D. A. “Three-dimensional study of cylindrical morphology in a styrene-butadiene-styrene block copolymer,” Polymer, 1988, 29, 387–395. 12. Spontak, R. J.; Smith, S. D.; Agard, D. A. “Electron tomography of microstructural elements in strongly segregated block copolymers,” Proceedings—Annual Meeting, Microscopy Society of America, 1993, 888–889. 13. Jinnai, H.; Spontak, R. J. “Transmission electron microtomography in polymer research,” Polymer, 2009, 50, 1067–1087. 14. Chou, T. M.; Prayoonthong, P.; Aitouchen, A.; Libera, M. “Nanoscale artifacts in RuO4-stained poly(styrene),” Polymer, 2002, 43, 2085–2088. 15. Wang, Y.; Coombs, N.; Turak, A.; Lu, Z. H.; Manners, I.; Winnik, M. A. “Interfacial staining of a phase-separated block copolymer with ruthenium tetroxide,” Macromolecules, 2007, 40, 1594–1597. 16. Gajdardziska-Josifovska, M.; Carim, A. “Applications of Electron Holography,” in Introduction to electron holography; Volkl, E., Allard, L., and Joy, D., Eds.; Plenum, New York, 1999; pp. 241–268. 17. Wang, Y. C.; Chou, T. M.; Libera, M.; Voelkl, E.; Frost, B. G. “Measurement of polystyrene mean inner potential by transmission electron holography of latex spheres,” Microscopy and Microanalysis, 1998, 4, 146–157. 18. Wanner, M.; Bach, D.; Gerthson, D.; Werner, R.; Tesche, B. “Electron holography of thin amorphous carbon films: Measurement of the mean inner potential and a thickness-independent phase shift,” Ultramicroscopy, 2006, 106, 341–345. 19. Lunt, R. R.; Kena-Cohen, S.; Benziger, J. B.; Forrest, S. R. “Measurement of the mean inner potentials of anthracene and naphthalene,” Physical Review Letters, 2009, 102, 065504–065507. 20. Petermann, J.; Gleiter, H. “Direct observation of amorphous and crystalline regions in polymers by defocus imaging,” Philosophical Magazine, 1975, 31, 929–934. 21. Liu, T.; Tjiu, W. C.; Petermann, J. “Transmission electron microscopy observations on fine structures of shish-kebab crystals of isotactic polystyrene by partial melting,” Journal of Crystal Growth, 2002, 243, 218–223. 22. Christner, G.; Thomas, E. L. “Visualization of polymer interfaces by phase contrast (“defocus”) electron microscopy,” Journal of Applied Physics, 1977, 48, 4063–4067. 23. Roche, E. J.; Thomas, E. L. “Defocus electron microscopy of multiphase polymers: Use and misuse,” Polymer, 1981, 22, 333–341. 24. Handlin, Jr., D. L.; Thomas, E. L. “Phase contrast imaging of styrene-isoprene and styrenebutadiene block copolymers,” Macromolecules, 1983, 16, 1514–1525. 25. Handlin, Jr., D. L.; Thomas, E. L. “Visualization of ordered spherical microdomains of block copolymers by phase contrast electron microscopy,” Journal of Materials Science Letters, 1984, 3, 137–140. 26. Simon, P.; Adhikari, R.; Lichte, H.; Michler, G. H.; Langela, M. “Electron holography and afm studies on styrenic block copolymers and a high impact polystyrene,” Journal of Applied Polymer Science, 2005, 96, 1573–1583. 27. Tonomura, A. Electron holography; Springer: Berlin, 1993. 28. Volkl, E.; Allard, L.; Joy, D. Introduction to electron holography; Plenum: New York, 1999.

336

M. R. Libera and R. F. Egerton

29. Blackson, J. H.; Zhang, X.; Joy, D. C. “Electron Holography of Polymer Blends,” in Proceedings of the 52nd annual meeting of the microscopy society of america; Bailey, G. W., and GarrattReed, A. J., Eds.; San Francisco Press: San Francisco, 1994; 126–127. 30. Libera, M.; Gajdardziska-Josifovska, M.; Disko, M. M. “Off-Axis Electron Holography of a Two-Phase Polymer,” in Proceedings of the 52nd annual meeting of the microscopy society of america; Bailey, G. W., and Garratt-Reed, A. J., Eds.; San Francisco Press: San Francisco, 1994; 444–445. 31. Libera, M.; Ott, J.; Wang, Y. C. “Transmission Electron Holography of Polymer Microstructure,” in Electron holography; Tonomura, A.; Allard, L. F.; Pozzi, G.; Joy, D. C.; Ono, Y. A. Eds.; Elsevier: Amsterdam, 1995; pp. 231–237. 32. Simon, P.; Huhle, R.; Lehmann, M.; Lichte, H.; Monter, D.; Bieber, T.; Reschetilowski, W.; Adhikari, R.; Michler, G. H. “Electron holography on beam sensitive materials: Organic polymers and mesoporous silica,” Chemistry of Materials, 2002, 14, 1505–1514. 33. Chou, T. M.; Libera, M.; Gauthier, M. “Quantitative phase contrast imaging of arborescent graft polystyrene by off-axis transmission electron holography,” Polymer, 2003, 44, 3037– 3043. 34. Simon, P.; Lichte, H.; Drechsel, J.; Formanek, P.; Graff, A.; Wahl, R.; Mertig, M.; Adhikari, R.; Michler, G. H. “Electron holography of organic and biological materials,” Advanced Materials, 2003, 15, 1475–1481. 35. Simon, P.; Maennig, B.; Lichte, H. “Conventional electron microscopy and electron holography of organic solar cells,” Advanced Functional Materials, 2004, 14, 669–676. 36. Simon, P.; Lichte, H.; Formanek, P.; Lehmann, M.; Huhle, R.; Carrillo-Cabrera, W.; Harscher, A.; Ehrlich, H. “Electron holography of biological samples,” Micron, 2008, 39, 229–256. 37. Danev, R.; Glaeser, R. M.; Nagayama, K. “Practical factors affecting the performance of a thinfilm phase plate for transmission electron microscopy,” Ultramicroscopy, 2009, 109, 312–325. 38. Mtasumoto, T.; Tonomura, A. “The phase constancy of electron waves traveling through boerch’s electrostatic wave plate,” Ultramicroscopy, 1997, 63, 5–10. 39. Majorovits, E.; Barton, B.; Schultheriss, K.; Perez-Willard, F.; Gerthsen, B.; Schroder, R. R. “Optimizing phase contrast in transmission electron microscopy with an electrostatic (boersch) phase plate,” Ultramicroscopy, 2007, 107, 213–226. 40. Malac, M.; Belaggia, M.; Egerton, R. F.; Zhu, Y. “Imaging of radiation-sensitive samples in transmission electron microscopes equipped with zernike phase plates,” Ultramicroscopy, 2008, 108, 126–140. 41. Tosaka, M.; Danev, R.; Nagayama, K. “Application of phase contrast transmission microscopic methods to polymer materials,” Macromolecules, 2005, 38, 7884–7886. 42. Egerton, R. F. Electron energy loss spectroscopy in the electron microscope, 2nd ed.; Plenum: New York, 1996. 43. Egerton, R. F. “Electron energy-loss spectroscopy in the TEM,” Reports on Progress in Physics, 2009, 72, 016502. 44. Ahn, C. C., Ed. Transmission electron energy loss spectrometry in materials science and the EELS atlas, 2nd ed.; Wiley-VCH: Weinheim, 2004. 45. Jeanguillaume, C.; Colliex, C. “Spectrum-imaging: the next step in EELS digital acquisition and processing,” Ultramicroscopy, 1988, 28, 252–257. 46. Hunt, J. A.; Williams, D. B. “Electron energy-loss spectrum-imaging,” Ultramicroscopy, 1991, 38, 47–73. 47. Hofer, F.; Warbichler, P. “Elemental Mapping Using Energy Filtered Imaging,” in Transmission electron energy loss spectrometry in materials science and the EELS atlas, Ahn, C., Ed.; Wiley-VCH: Wienheim, 2004; pp. 159–222. 48. Linares, E. M.; Leite, C. A. P.; Valadares, L. F.; Silva, C. A.; Rezende, C. A.; Galembeck, F. “Molecular mapping by low-energy-loss energy-filtered transmission electron microscopy,” Analytical Chemistry, 2009, 81, 2317–2324. 49. Du Chesne, A. “Energy filtering transmission electron microscopy of polymers—benefit and limitations of the method,” Macromolecular Chemistry and Physics, 1999, 200, 1813–1830.

Advances in the Transmission Electron Microscopy of Polymers

337

50. Libera, M.; Disko, M. M. “Electron Energy Loss Spectroscopy of Polymers,” in Transmission Electron energy loss spectrometry in materials science; Ahn, C., Ed.; Wiley-VCH: New York, 2004; 419–454. 51. Ribbe, A. E.; Hayashi, M.; Weber, M.; Hashimoto, T. “Morphology determination of novel polysulfone-polyamide block copolymers using element spectroscopic imaging in the transmission electron microscopy,” Macromolecules, 2000, 33, 2786–2789. 52. Varlot, K.; Martin, J. M.; Quet, C. “Phase differentiation of a tri-phase polymer composite using enery-filtering transmission electron microscopy,” Polymer, 2000, 41, 4599–4605. 53. Rippel, M. M.; Leite, C. A. P.; Galembeck, F. “Elemental mapping in natural rubber latex films by electron energy loss spectroscopy associated with transmission electron microscopy,” Analytical Chemistry, 2002, 74, 2541–2546. 54. Yamauchi, K.; Takahashi, K.; Hasegawa, H.; Iatrou, H.; Hadjichristidis, N.; Kaneko, T.; Nishikawa, Y.; Jinnai, H.; Matsui, T.; Nishioka, H.; Shimizu, M.; Furukawa, H. “Microdomain morphology in an ABC 3-miktoarm star terpolymer: A study by energy-filtering TEM and 3D electron tomography,” Macromolecules, 2003, 36, 6962–6966. 55. Correa, C. A.; Bonse, B. C.; Chinaglia, C. R.; Hage, Jr., E.; Pessan, L. A. “Characterization of unstained multiphase polymer systems by analytical electron microscopy,” Polymer Testing, 2004, 23, 775–778. 56. Rippel, M. M.; Leite, C. A. P.; Lee, L. T.; Galembeck, F. “Direct imaging and elemental mapping of microgels in natural rubber particles,” Colloid and Polymer Science, 2005, 283, 570–574. 57. Hayakawa, T.; Goseki, R.; Kakimoto, M. A.; Tokita, M.; Watanabe, J.; Liao, Y.; Horiuchi, S. “Self-assembled lamellar nanostructures of wholly aromatic rod-rod-type block molecules,” Organic Letters, 2006, 8, 5453–5456. 58. Sousa, A.; Sengonul, M.; Latour, R.; Kohn, J.; Libera, M. “Selective protein adsorption on a phase-separated solvent-cast polymer blend,” Langmuir, 2006, 22, 6286–6292. 59. Gomez, E. D.; Panday, A.; Feng, E. H.; Chen, V.; Stone, G. M.; Minor, A. M.; Kisielowski, C.; Downing, K. H.; Borodin, O.; Smith, G. D.; Balsara, N. P. “Effect of ion distribution on conductivity of block copolymer electrolytes,” Nano Letters, 2009, 9, 1212–1216. 60. Horiuchi, S.; Hanada, T.; Ebisawa, M.; Matsuda, Y.; Kobayashi, M.; Takahara, A. “Contamination-free transmission electron microscopy for high-resolution carbon elemental mapping of polymers,” ACS Nano, 2009, 3, 1297–1304. 61. Arayasantiparb, D.; McKnight, S.; Libera, M. “Compositional variation within the epoxy/adherend interphase,” Journal of Adhesion Science and Technology, 2001, 15, 1463–1484. 62. Horiuchi, S.; Hamanaka, T.; Aoki, T.; Miyakawa, T.; Narita, R.; Wakabayashi, H. “Characterization of polymer-adhesive interfaces on a nanometre scale by elemental mapping and image EELS using EFTEM,” Journal of Electron Microscopy, 2003, 52, 255–266. 63. Horiuchi, S.; Yin, D.; Ougizawa, T. “Nanoscale analysis of polymer interfaces by energyfiltering transmission electron microscopy,” Macromolecular Chemistry and Physics, 2005, 206, 725–731. 64. Horiuchi, S.; Dohi, H. “Nanoimaging and spectroscopic analysis of rubber/zno interfaces by energy-filtering transmission electron microscopy,” Langmuir, 2006, 22, 4607–4613. 65. Horiuchi, S.; Yin, D.; Liao, Y.; Ougizawa, T. “Study of adhesion and fracture of polymer laminates by imaging of interfaces,” Macromolecular Rapid Communications, 2007, 28, 915– 921. 66. Liao, Y.; Nakagawa, A.; Horiuchi, S.; Ougizawa, T. “Interdiffusion at homopolymer/random copolymer interfaces investigated by energy-filtering transmission electron microscopy,” Macromolecules, 2007, 40, 7966–7972. 67. Gomez, E. D.; Ruegg, M. L.; Minor, A. M.; Kisielowski, C.; Downing, K. H.; Glaeser, R. M.; Balsara, N. P. “Interfacial concentration profiles of rubbery polyolefin lamellae determined by quantitative electron microscopy,” Macromolecules, 2008, 41, 156–162. 68. Bertho, J.; Stolojan, V.; Abel, M. L.; Watts, J. F. “The effect of silane incorporation on a metal adhesive interface: A study by electron energy loss spectroscopy,” Micron, 2010, 41, 130–134.

338

M. R. Libera and R. F. Egerton

69. Sun, S.; Shi, S.; Leapman, R. D. “Water distributions of hydrated biological specimens by valence electron energy loss spectroscopy,” Ultramicroscopy, 1993, 50, 127–139. 70. Sun, S. Q.; Shi, S.-L.; Hunt, J. A.; Leapman, R. D. “Quantitative water mapping of cryosectioned cells by electron energy-loss spectroscopy,” Journal of Microscopy, 1995, 177, 18–30. 71. Kim, G.; Sousa, A.; Meyers, D.; Shope, M.; Libera, M. “Diffuse polymer interfaces in lobed nanoemulsions preserved in aqueous media,” Journal of the American Chemical Society, 2006, 128, 6570–6571. 72. Sousa, A.; Aitouchen, A.; Libera, M. “Water mapping in hydrated soft materials,” Ultramicroscopy, 2006, 106, 130–145. 73. Sousa, A.; Schut, J.; Kohn, J.; Libera, M. “Nanoscale morphological changes during hydrolytic degradation and erosion of a bioresorbable polymer,” Macromolecules, 2006, 39, 7306–7312. 74. Kim, G.; Sousa, A.; Meyers, D.; Libera, M. “Nanoscale composition of biphasic polymer nanocolloids in aqueous suspension,” Microscopy and Microanalysis, 2008, 14, 459–468. 75. Yakovlev, S.; Libera, M. “Cryo-stem eels of nafion saturated with an organic solvent,” Microscopy and Microanalysis, 2006, 12, 996–997. 76. Yakovlev, S.; Misra, M.; Shi, S.; Libera, M. “Specimen thickness dependence of hydrogen evolution during cryo-transmission electron microscopy of hydrated soft materials,” Journal of Microscopy, 2009, 236, 174–179. 77. Parravicini, G. P.; Resca, L. “Electronic states and optical properties in cubic ice,” Physical Review B, 1973, 8, 3009–3023. 78. Egerton, R. F. “New techniques in electron energy-loss spectroscopy and energy-filtered imaging,” Micron, 2003, 34, 127–139. 79. Kothleitner, G.; Hofer, F. “EELS performance measurements on a new high energy resolution imaging filter,” Micron, 2003, 34, 211–218. 80. Kimoto, K.; Kothleitner, G.; Grogger, W.; Matsui, Y.; Hofer, F. “Advantages of a monochromator for bandgap measurements using electron energy-loss spectroscopy,” Micron, 2005, 36, 185–189. 81. Lazar, S.; Botton, G. A.; Zandbergen, H. W. “Enhancement of resolution in core-loss and lowloss spectroscopy in a monochromated microscope,” Ultramicroscopy, 2006, 106, 1091–1103. 82. Egerton, R. F. “Limits to the spatial, energy and momentum resolution of electron energy-loss spectroscopy,” Ultramicroscopy, 2007, 107, 575–586. 83. Dahmen, U.; Erni, R.; Radmilovic, V.; Kisielowski, C.; Rossell, M. D.; Denes, P. “Background, status and future of the transmission electron aberration-corrected microscope project,” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2009, 367, 3795–3808. 84. Krivanek, O. L.; Ursin, J. P.; Bacon, N. J.; Corbin, G. J.; Dellby, N.; Hrncirik, P.; Murfitt, M. F.; Own, C. S.; Szilagyi, Z. S. “High-energy-resolution monochromator for aberration-corrected scanning transmission electron microscopy/electron energy-loss spectroscopy,” Philosophical Transactions of the Royal Society A, 2009, 367, 3683–3697. 85. Katterwe, H. “Object Analysis by Electron Energy Loss Spectroscopy in the Infra-Red Region,” in Electron microscopy—1972; Cosslet, V. E., Ed.; The Institute of Physics: London, 1972; pp. 154–155. 86. Ritsko, J. J.; Bigelow, R. W. “Core excitons and the dielectric response of polystyrene and poly(2-vinylpyridine) from 1 to 400 ev,” Journal of Chemical Physics, 1978, 69, 4162–4170. 87. Ritsko, J. J. “Electron energy loss spectroscopy of pristine and radiation damaged polyethylene,” Journal of Chemical Physics, 1979, 70, 5343–5349. 88. Fink, J.; Scheerer, B.; Stamm, M.; Tieke, B.; Kanellakopulos, B.; Dornberger, E. “Electronic structure of li-doped polyparaphenylene,” Physical Review B, 1984, 30, 4867–4869. 89. Fink, J.; Scheerer, B.; Wernet, W.; Monkenbusch, M.; Wegner, G.; Freund, H.-J.; Gonska, H. “Electronic structure of pyrrole-based conducting polymers: An electron-energy-lossspectroscopy study,” Physical Review B, 1986, 34, 1101–1115. 90. Duke, C. B.; Salaneck, W. R.; T.J., F.; Ritsko, J. J.; Thomas, H. R.; Paton, A. “Electronic structure of pendant-group polymers: molecular-ion states and dielectric properties of poly(2vinyl pyridine),” Physical Review B, 1978, 18, 5717–5739.

Advances in the Transmission Electron Microscopy of Polymers

339

91. Beamson, G. “Conformation effects in the xps valence band spectra of aliphatic polyesters,” Journal of Electron Spectroscopy and Related Phenomena, 2007, 154, 83–89. 92. French, R. H.; Winey, K. I.; Yang, M. K.; Qiu, W. “Optical properties and van der waalslondon dispersion interactions of polystyrene determined by vacuum ultraviolet spectroscopy and spectroscopic ellipsometry,” Australian Journal of Chemistry, 2007, 60, 251–263. 93. Ueno, N.; Kera, S. “Electron spectroscopy of functional organic thin films: Deep insights into valence electronic structure in relation to charge transport property,” Progress in Surface Science, 2008, 83, 490–557. 94. Salaneck, W. R. “Classical ultraviolet photoelectron spectroscopy of polymers,” Journal of Electron Spectroscopy and Related Phenomena, 2009, 174, 3–9. 95. De Rosier, D. J.; Klug, A. “Reconstruction of three dimensional structures from electron micrographs,” Nature, 1968, 217, 130–134. 96. Frank, J. “Averaging of low exposure electron micrographs of non periodic objects,” Ultramicroscopy, 1975, 1, 159–162. 97. van Heel, M.; Frank, J. “Use of multivariates statistics in analysing the images of biological macromolecules,” Ultramicroscopy, 1981, 6, 187–194. 98. van Heel, M. “Classification of very large electron microscopical image data sets,” Optik, 1989, 82, 114–126. 99. Frank, J. “Single-particle imaging of macromolecules by cryo-electron microscopy,” Annual Review of Biophysics and Biomolecular Structure, 2002, 31, 303–319. 100. Subramaniam, S.; Milne, J. L. S. “Three-dimensional electron microscopy at molecular resolution,” Annual Review of Biophysics and Biomolecular Structure, 2004, 33, 141–155. 101. Reimer, L.; Kohl, H. Transmission electron microscopy: Physics of image formation; Springer: Berlin, 2008. 102. Chapiro, A. Radiation chemistry of polymeric systems; Plenum: New York, 1962. 103. Howie, A. “Radiation damage problems in electron microscopy,” Revue de Physique Applique, 1980, 15, 291–295. 104. Glaeser, R. “Retrospective: radiation damage and its associated ‘information limitations’,” Journal of Structural Biology, 2008, 163, 271–276. 105. Yakovlev, S.; Libera, M. “Dose-limited spectroscopic imaging of soft materials by low-loss EELS in the scanning transmission electron microscope,” Micron, 2008, 39, 734–740. 106. Varlot, K.; Martin, J. M.; Gonbeau, D.; Quet, C. “Chemical bonding analysis of electronsensitive polymers by EELS,” Polymer, 1999, 20, 5691–5697. 107. Siangchaew, K.; Libera, M. “Influence of fast secondary electrons on the aromatic structure of polystyrene,” Philosophical Magazine, 2000, 80, 1001–1016. 108. Haider, M.; Rose, H.; Uhlemann, S.; Schwan, E.; Kabius, B.; Urban, K. “A spherical-aberrationcorrected 200 kv transmission electron microscope,” Ultramicroscopy, 1998, 75, 53–60. 109. Evans, J. E.; Hetherington, C.; Kirkland, A.; Chang, L. Y.; Stahlberg, H.; Browning, N. “Lowdose aberration corrected cryo-electron microscopy of organic specimens,” Ultramicroscopy, 2008, 108, 1636–1644. 110. Smith, D. J. “Development of aberration-corrected electron microscopy,” Microscopy and Microanalysis, 2008, 14, 2–15. 111. Urban, K. W. “Studying atomic structures by aberration-corrected transmission electron microscopy,” Science, 2008, 321, 506–510. 112. Rose, A. “Television Pickup Tubes and the Problem of Vision,” in Advances in electronics; Marton, L., Ed.; Academic Press: New York, 1948; 131–166. 113. Rose, A. Vison: Human and electronic; Plenum: New York, 1973. 114. Burgess, A. E. “The rose model, revisited,” Journal of the Optical Society of America A, 1999, 16, 633–646. 115. McMullan, G.; Chen, S.; Henderson, R.; Faruqi, A. R. “Detective quantum efficiency of electron area detectors in electron microscopy,” Ultramicroscopy, 2009, 109, 1126–1143.

Polymer Reviews, 50:340–384, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.495440

The Morphology of Poly(3,4-Ethylenedioxythiophene) DAVID C. MARTIN,1,2,3 JINGHANG WU,1,2 CHARLES M. SHAW,1,2 ZACHARY KING,1,3 SARAH A. SPANNINGA,1 SARAH RICHARDSON-BURNS,1,3 JEFFREY HENDRICKS,1,3 AND JUNYAN YANG1,4 1

The University of Michigan, Ann Arbor, Michigan The University of Delaware, Newark, Delaware 3 Biotectix LLC, Ann Arbor, Michigan 4 The Dow Chemical Company, Freeport, Texas 2

Poly(3,4-ethylene dioxythiophene) (PEDOT) is a chemically stable, conjugated polymer that is of considerable interest for a variety of applications including coatings for interfacing electronic biomedical devices with living tissue. Here, we describe recent work from our laboratory and elsewhere to investigate the morphology of PEDOT in the solid state. We discuss the importance of oxidative chemical and electrochemical polymerization, as well as the critical role of the counterion used during synthesis and film deposition. We have obtained information about the morphology of PEDOT from a number of different complimentary techniques including X-ray diffraction, optical microscopy, scanning electron microscopy, transmission high-resolution electron microscopy, and low-voltage electron microscopy. We also discuss results from ultravioletvisible light spectroscopy (UV-Vis), Fourier transform infrared spectroscopy (FTIR), and X-ray photoelectron spectroscopy (XPS). PEDOT is a relatively rigid polymer that packs in the solid state at a characteristic face-to-face distance (010) of ∼0.34 nm, similar to graphite. These sheets of oriented PEDOT molecules are separated from one another by ∼1.4 nm laterally, with the (100) distance between layers quite sensitive to the choice of counterion used during sample preparation. The order in the films is typically modest, although this also depends on the counterion used and the method of film deposition. The films can be organized into useful structures with a variety of nanoscale dissolvable templates (including fibers, particles, and lyotropic mesophases). When PEDOT is electrochemically deposited in the presence of bromine counterions, highly ordered crystalline phases are observed. It is also possible to deposit PEDOT around living cells, both in vitro and in vivo. Keywords conjugated polymers, polymer morphology, poly(thiophene), polymer microscopy

1. Introduction There has been considerable interest in the development of conjugated polymers because of their potential use in flexible organic electronic devices such as photovoltaics, light-emitting Received February 16, 2010; accepted May 19, 2010. Address correspondence to David Martin, The University of Delaware, Materials Science and Engineering, 201C DuPont Hall, Newark, DE 19716. E-mail: [email protected]

340

Morphology of PEDOT

341

diodes, sensors, actuators, and thin-film transistors.1 The conjugated polymer backbone, consisting of alternating carbon–carbon double bonds, provides for π -orbital overlap along the molecule. This leads to a number of useful properties including relatively easy charge transport and the ability to tailor the absorption and emission spectrum.2 Conjugated polymers that have been investigated in some detail include poly(acetylene), poly(phenylene vinylene), poly(pyrrole), as well as poly(thiophenes) and their various derivatives. Conjugated polymers are typically synthesized by oxidative polymerization using chemical, electrochemical, or chemical vapor deposition methods.3,4 During the reaction covalent bonds are established between the constituent monomers, removing hydrogens from the aromatic phenyl, pyrrole, or thiophene rings depending on the structure of interest. Irregularities in the chemical structure can arise when the monomers have more than one reactive hydrogen present on the conjugated ring.5 One conjugated polymer that has been of particular recent attention is the diethoxysubstituted thiophene poly(3,4-ethylene dioxythiophene), or PEDOT.3,6,7 As shown in Figure 1, the EDOT monomer consists of a five-membered thiophene ring with the hydrogens at the 3 and 4 positions replaced with a dioxy-ethyl substituent group. Note that because the EDOT monomer has only two reactive hydrogens (at the 2 and 5 positions), it is not in principle possible to have chemical defects in the PEDOT chain (other than from variations in the monomer purity). This regularity in the molecular structure has been associated with its outstanding chemical stability.8 The polymer is typically used in its oxidized state where the molecular backbone is loaded with mobile carriers (holes) and thus is electrically conductive. PEDOT can be doped with many different types of anions including smaller molecules (ClO4 -, heparin), as well as macromolecular polyanions such as poly(styrene sulfonate) (PSS).9 At equilibrium it is usual to find a charge density of about one extra positive charge per every three EDOT monomer units.3 It is therefore necessary to have one charge-balancing anion per every three EDOT units (Figure 1). Because of resonance structures that arise from the conjugated backbone, it is expected that the PEDOT chains will be fairly stiff and extended and ought to remain reasonably linear locally. Figure 2 shows a molecular simulation (Compass force field, Materials Studio 4, Accelrys, San Diego, California) of a single chain of PEDOT (with an effective molecular weight of 50,000 g/mol), next to a single chain of PSS (also 50,000 g/mol). As anticipated, the PEDOT molecule is locally rigid and extended, with the planar thiophene rings alternating back and forth along the nearly linear molecular trajectory. Although PEDOT itself is not soluble in water, it is possible to form an aqueous suspension by complexing it with PSS. Given the substantial difference in rigidity between PEDOT and the more globular PSS, it is expected that several PSS coils will be associated with a given PEDOT molecule.6,7 A number of other functionalized thiophene monomers have been synthesized that are chemically similar to EDOT, including 3,4-propylenedioxythiophene (ProDOT) and 3,4-butylenedioxythiopene (BuDOT).3 It is also possible to modify the EDOT monomer with side groups to tailor the surface properties, including designing specific interactions with solid substrates or with living tissue. Examples include hydroxy-methylated EDOT10 and alkoxy-functionalized EDOT.11,12 The synthesis of an EDOT monomer functionalized with a carboxylic acid (EDOT-acid) has recently been reported,13 as well as EDOT-thiol.14 Bifunctional molecules linking together two monomers of EDOT and EDOT derivatives have been created.15 Other interesting examples include PEDOT variants with pendant crown ethers, attached either onto the side of the thiophene ring directly or via a more flexible side chain.16,17

342

D. C. Martin et al.

Figure 1. Top: Chemical structures of 3,4-ethylene dioxythiophene (EDOT) and 3,4-dihydroxy-lphenylalanine (l-DOPA). Middle: Chemical structure of poly(3,4-ethylene dioxythiophene) (PEDOT) and eumelanin. Bottom: In the electrically active, oxidized state, there is one positive charge on the PEDOT molecule per every three EDOT units. This charge on the backbone is balanced with an anion that may be either a small molecule or a macromolecular anion such as poly(styrene sulfonate) (PSS). (Figure available in color online)

The backbone chemistry of PEDOT is quite similar to that of eumelanin, the dark brown–black pigment common in nature. Eumelanin is derived from 3,4-dihydroxy-lphenylalanine, or l-DOPA. As shown in Figure 1, the primary molecular structure of eumelanin shows some close chemical similarities to that of PEDOT. Both molecules have a conjugated backbone (shown in red in the color version of Figure 1, available online), as well as oxygens immediately pendant to the conjugated backbone (shown in green in the color version of Figure 1, available online). Melanins are, of course, abundant as natural pigments in skin and hair, where they serve to absorb light, presumably providing

Morphology of PEDOT

343

Figure 2. Molecular simulation of a single chain of PEDOT next to a single chain PSS, both with effective molecular weights of 50,000 g/mol. The PEDOT is extended and straight, whereas the PSS has a more flexible, globular conformation. Simulated using Materials Studio by Accelrys and the compass force field. (Figure available in color online)

a measure of ultraviolet (UV) protection. However, it is interesting to consider that they are also found in electronically and ionically active living tissues including the retina, the inner ear, and the brain. Indeed, there is a component of the brain stem called the substantia nigra (Latin for black substance), which is dark because of a high local concentration of neuromelanin. The loss of pigmentation in the substantia nigra is known to be correlated with Parkinson’s disease (PD). Autopsies of patients with PD show a much lighter color of the substantia nigra than healthy individuals. Melanin is also highly expressed in the cochlea (inner ear), particularly in the region between the scala media (a chamber filled with a potassium-enriched fluid called endolymph) and the scala tympani (filled with a sodium-enriched fluid called perilymph). This local variation in ionic composition leads to a relatively large voltage difference (the endocochlear potential) of the order 80–100 mV.18 The purpose of this article is to describe recent efforts from our laboratory and elsewhere to characterize the morphology of PEDOT in the solid state. We have been particularly interested in the development of PEDOT as a soft, functional material for interfacing hard, electronically conducting metallic and semiconducting biomedical devices with wet, ionically active living tissue. When considering the design of materials for interfacing electronic biomedical devices with living tissue, it is necessary to consider and optimize many different performance requirements. Electronic biomedical devices are typically engineered from solid inorganic metals (platinum, gold, iridium) or semiconductors (silicon). They are also inert and conduct current electronically. On the other hand, living tissue is organic, wet, and conducts charge ionically. Furthermore, engineered components typically have a relatively flat, uniform surface, whereas surfaces in biological systems are complicated and may facilitate active transport from the interior to exterior of a cell. In order to create materials that are better suited for biomedical device–tissue interface applications, it is important to match both the electrical and mechanical properties of the materials used with those of the living systems. Figure 3 dramatically demonstrates the scale of this design problem, plotting the electrical properties (as resistivity in units of ohm/m) of known engineering materials as a function of their mechanical properties (as modulus in units of Pa), using the CES Selector software version 4.6.1 by Granta Design, Cambridge, UK. The resistivities vary

344

D. C. Martin et al.

Figure 3. Plot of the electrical properties of materials (resistivity in ohm/m) as a function of their mechanical properties (Young’s modulus in Pa) for known engineering materials. Plot generated using CES Selector software version 4.6.1 by Granta Design. The upper right shows the stiff insulating ceramics, and the lower right shows the stiff, conducting inorganic metals and semiconductors. Polymers are typically insulating and are softer than ceramics but not nearly as soft as gels and foams. Tissue is quite soft and is also conducting because it is an electrolyte. There is a need for soft, conducting materials that can interface between the living tissue and the hard, electrically conducting materials typically used in biomedical devices. (Figure available in color online)

over 30 orders of magnitude, from highly conducting (10−9 ohm/m) to highly insulating (1018 ohm/m). Likewise, there are over 12 orders of magnitude of differences in mechanical response, from extremely soft (with moduli of a few Pa) to extremely stiff (1012 Pa). Classes of chemically similar materials (metals, ceramics, polymers) tend to cluster together when plotted in diagrams of this sort.19 In this particular diagram the upper right corresponds to the highly insulating, stiff ceramics, and the lower right is the highly conductive inorganic metals and semiconductors. Polymers are much softer than metals or ceramics and are usually insulating. However, solid, fully dense polymers are still much softer than lowdensity foams or liquid-swollen gels. PEDOT is an example of a conductive polymer with a resistivity approaching that of common metals. Tissue itself is a relatively soft, gel-like material that is swollen with an electrolyte, making it an ionic conductor. The use of soft, organic, conducting polymers like PEDOT seems to be a reasonable way of interfacing tissue with hard, inorganic conductors, because they can help to bridge these large gaps in performance. There is likely to be significant potential in the direct integration of PEDOT into hydrogels with mechanical properties similar to tissue, as we will discuss in more detail in this review.

Morphology of PEDOT

345

We have used a variety of different microscopic characterization methods to interrogate the structure of PEDOT in these studies. Here, we will summarize some of these results and in the process demonstrate how these different techniques provide complimentary information, leading to a detailed understanding of the morphology of this particularly important and interesting polymeric material. Our laboratory has been actively developing a variety of methods for the detailed microstructural characterization of ordered polymers and organic molecular crystals. In particular, we recently reported on the refinement and application of low-dose high-resolution electron microscopy (HREM).20 We have also demonstrated the utility of low-voltage electron microscopy (LVEM).21 This review article focuses on the use of these different microscopy techniques, along with X-ray diffraction, X-ray photoelectron spectroscopy (XPS), and Fourier transform infrared spectroscopy (FTIR) to elucidate the morphology of PEDOT.

2. Mechanisms of EDOT Polymerization During polymerization of EDOT, the monomer is oxidized to a radical cation in a zone adjacent to the electrode (for electrochemical polymerization) or the reactive oxidant species (for chemical vapor deposition [CVD] or chemical polymerization).22 It has been hypothesized that this radical cation then reacts with another radical cation to form a dimeric dication.23 The dimeric dication subsequently loses two protons, leading to the formation of dimers and then eventually higher oligomers and polymers. In the electrochemical method, deposition of PEDOT onto the metal anode is promoted by the relative insolubility of the polymer in the reaction media. We have found that significant local changes in the pH near the electrode can be detected by using indicator dyes during the reaction. Although the molecular weight of electropolymerized films of PEDOT is difficult to measure because of their insolubility, it has been demonstrated that neutral PEDOT can be solubilized in common organic solvents by careful control of the stoichiometry of the iron (III) chloride oxidant during the reaction.6 The PEDOT prepared in this manner can evidently by solubilized in chloroform, dichloromethane, or tetrahydrofuran (THF) and has a number average molecular weight of 1,000 to 1,500 g/mol, corresponding to 7–10 EDOT units (the monomer molecular weight is 144 g/mol).6 In order to create films of PEDOT of interest for practical devices, at least four different methods have been employed. The first is to use an oxidant such as iron chloride to polymerize the monomer and a counterion such as PSS to keep the PEDOT molecules in suspension. These PEDOT/PSS suspensions can be readily spun-cast into thin films with reasonable electrical properties and are the basis for the Baytron/Clevios materials that have been commericalized by H. C. Starck, Leverkusen, Germany. The conductivities of these films are modest, presumably because the PSS forms a shell around the PEDOT that prevents good contact between chains after the films are formed. The second method is to use electrochemical deposition, causing the polymer to be formed on the anode. Electrochemical deposition can be readily done on patterned metallic surfaces and provides films with better electrical properties than chemical polymerization. Because charge transport is required for the films to form, this method ensures that the PEDOT is electrically connected to the conductive substrate. The last two methods involve oxidative polymerization on solid surfaces. Vapor-phase polymerization (VPP) involves the polymerization of EDOT from a surface that has been previously coated with a layer of oxidant. The last technique is oxidative CVD. This method requires fairly specialized high-vacuum equipment.

346

D. C. Martin et al.

To date the most ordered and crystalline thin films of PEDOT have been seen using the VPP and CVD techniques. Chemical polymerizations in solutions or suspensions and electrochemical polymerizations onto metal surfaces have usually led to film with much less crystallinity. However, it is possible to create nanometer-scale structures in PEDOT using templates such as polymer latex spheres or liquid-crystalline mesophases. In addition, extremely highly ordered films, with faceted, optically birefringent crystals many hundreds of micrometers in size, have been observed when electrochemically depositing PEDOT with bromine counterions. 2.1. Vapor-Phase Polymerized PEDOT In the vapor-phase polymerization of PEDOT, the surface of a solid sample is first coated with an Fe(III) salt and then exposed to an atmosphere of EDOT monomer.24,25 The technique results in uniform, relatively thin (submicron) films of PEDOT.25,26 Grazing incidence X-ray diffraction of these films have shown high degrees of order, with the PEDOT chains packing into sheets.27 The chains were oriented parallel to the substrate surface, and the chain–chain edge-to-edge packing distance d100 was around 1.4 nm. The face-to-face packing dimension d010 was about 0.35 nm, as expected for closepacked, nearly planar conjugated molecules. Evidence for an order–disorder transition was seen near 130 C by DSC and XRD; however, this did not lead to dramatic changes in the conductivity of the film, suggesting that the efficiency of electrical transport was not directly related to the degree of crystallinity alone.27 The unit cell for PEDOT consistent with these results is shown in Figure 4. This structure was proposed from earlier studies on more weakly ordered films by Aasmundtveit et al.28,29 The essential picture of PEDOT in the solid state is one in which the oriented,

Figure 4. Crystal structure of PEDOT consisting of PEDOT molecules packed face to face into sheets. Originally published by Aasmundtveit et al.28 Used with permission. (Figure available in color online)

Morphology of PEDOT

347

Figure 5. Schematic diagrams showing the geometry and assembly of single PEDOT chains, twochain PEDOT molecular complexes, and PEDOT sheets. Top: Side view ([010] projection), middle: oblique view, bottom: top view ([001] projection). (Figure available in color online)

rigid, and linear PEDOT chains are packed into close-packed layers. The PEDOT molecules are organized with the chain parallel to the c direction (as is usual for polymers). The repeat distance along c is 0.78 nm or 0.39 nm per monomer unit. The a-axis spacing (∼1.4 nm) corresponds to the distance between chains side to side and, as discussed, this fluctuates from system to system depending on the dopant. The b-axis (0.68 nm) corresponds to twice the distance from molecule to molecule face to face (0.34 nm). Because of the alternating arrangement of the EDOT monomers along the PEDOT backbone, the minimum energy configuration is for two neighboring chains to pack so that these side groups alternate positions, with the monomer units on a given chain lining up over gaps between the monomers in the chain next to it. Figures 5 and 6 show schematically the arrangement of PEDOT chains in the condensed state. In Figure 4 a single chain is shown with the alternating EDOT monomers along the

348

D. C. Martin et al.

Figure 6. Schematic of the variation in microstructure of crystalline PEDOT with different dopants causing the (100) sheets of PEDOT chains to be closer together (green) or farther apart (blue) depending on the chemistry of the counterions used. (Figure available in color online)

backbone. A second chain placed on top of this chain has its minimum energy position so that the monomer side group is placed between the monomers on the lower chain. By stacking many chains like this, a sheet of PEDOT molecules is created (Figures 5 and 6). These sheets then stack at different distances from one another depending on the dopant (Figure 6).

Morphology of PEDOT

349

2.2. Chemically Vapor-Deposited PEDOT Gleason et al. have described the preparation of thin films of PEDOT using CVD.30–32 In this case the films are deposited in a vacuum chamber, and the iron (III) chloride oxidant is introduced by sublimation. CVD can presumably produce films that are substantially thicker than VPP; however, relatively specialized vacuum equipment is required, and the end result is a film that conformally coats the target entirely. It is thus somewhat difficult to locally pattern the films onto specific locations (such as the electrodes themselves) for a given device geometry. 2.3. Chemically Polymerized PEDOT The most common method of preparing PEDOT is to use oxidative chemical polymerization with a catalyst such as iron chloride and then stabilize the resulting suspension with a hydrophilic counterion such as poly(styrene sulfonate). The resulting aqueous suspension can then be cast into films of interest for a wide variety of applications. Studies of the morphology of these films by XPS and ultraviolet photoelectron spectroscopy (UPS) revealed local phase segregation of PEDOT and PSS and were interpreted to correspond to “grains” of about 3–4 nm with a PSS-rich shell and a PEDOT-rich core.33 This general picture was confirmed by scanning tunneling microscopy studies, although the size of the grains can be somewhat larger (30–50 nm).34 The size of these grains has been associated with the particles formed in the processing of suspension and the filtering processes during preparation. The current understanding is that the particles are deformed into pancake-shaped plates during the drying processes involved in film formation.35 Recent efforts have generally been consistent with this picture, suggesting that the individual grains of PEDOT are formed from “tangles” involving the association of several PEDOT molecules with larger, more flexible PSS strands, and that the grains themselves are composed of several tangles into larger clusters.35 2.4. Electrochemically Deposited PEDOT Like other conducting polymers such as polypyrrole, PEDOT can be readily deposited onto metal surfaces using oxidative electrochemical deposition. Both water and acetonitrile can be used as solvents. Certain organic ionic liquids have also proven to be useful as solvent for PEDOT electrochemical deposition.36,37 Electrochemical deposition can be done under constant current (galvanostatic) or constant voltage (potentiostatic) modes. This can also be done by cycling the potential through the oxidation point. The PEDOT films are deposited directly onto a conductive electrode and can have precisely tailored structures by using dissolvable templates such as particles38,39 or nanofibers.40,41 Wide-angle X-ray scattering from electrochemically deposited PEDOT films typically shows only a limited degree of structural order, although this is a function of the dopant used during deposition and the charge density used in the process. Figure 7 shows a series of two-dimensional (2D) XRD patterns with different dopants (including PSS, NaCl, LiF, and CaCl2 ). The lower panel shows azimuthally averaged scans showing the peak near 6–7 degrees 2θ (corresponding to a d-spacing of ∼1.4 nm), showing that the NaCl sample has the highest intensity and sharpest peak, whereas the other dopants have a significantly weaker and somewhat broader peak (Figure 6). Figure 8 shows a powder XRD pattern from PEDOT using NaCl as the dopant. The peak near 6–7 degrees 2θ is associated with the (100) spacing, corresponding to the average

350

D. C. Martin et al.

Figure 7. Top: 2D wide-angle x-ray scattering (WAXS) of scattering from PEDOT with different counterions: PSS, NaCl, LiF, and CaCl2 . Bottom: Azimuthally average WAXS scans of the scattering below 20 degrees showing the development of the ∼1.4 nm (100) intersheet peak with different counterions. (Figure available in color online)

distance between the ladder-like PEDOT molecules. The face-to-face packings between chains are in the 0.3–0.4 nm range, but the limited degree of order in typical PEDOT samples has not made it possible to be any more definitive about the nature of the packing in the solid state. When different dopants are used to prepare PEDOT, it is found that there are systematic variations in the position of the low-angle XRD peak, corresponding to variations in the average d100 distance between the stacks of PEDOT chains. Table 1 shows a listing of the observed d100 spacings, ranging from 1.52 nm (with poly(acrylic acid) [PAA] as the dopant) to 1.15 nm (for F- as dopant). This variation in morphology is shown schematically in Figure 6. Electrochemical deposition makes it possible to deposit PEDOT directly onto a microfabricated metal electrode. Figure 9 shows a film of PEDOT deposited onto a 40-µmdiameter electrode used to make a neural prosthetic device.42 We have found that the deposition of PEDOT films onto metal electrodes leads to a dramatic decrease in the impedance for the biologically important frequencies near 1,000 Hz corresponding to the characteristic

Morphology of PEDOT

351

Figure 8. Powder WAXS from PEDOT films showing the (100), (200), (400), and (020) peaks for crystalline PEDOT. There is also a peak from an underlying Au(111) film used for electrochemical deposition and diffraction pattern calibration. (Figure available in color online)

width of a neural pulse (1 ms).8 This drop in net impedance has been associated with the development of increased surface area that facilitates charge exchange between the electrons in the metallic and semiconducting engineered substrate, the holes along the conducting polymer backbone, and the ions in the dopant and surrounding electrolyte. We have been quite interested in developing methods for increasing the effective area of interaction between the conducting polymer and the surrounding electrolyte to reduce the impedance of these electrodes as much as possible. Rather than create flat films, we focused on methods to make the surface as fuzzy and furry as possible. In this example (Figure 10), poly(acrylic acid) was added as a counterion. We have found that with a molecular weight of PAA near 500K, concentration near 0.5 wt%, and current density near 0.5 mA/cm2, it is possible to reproducibly create nanofibrillar morphologies.43 It has been shown that these nanofibrils are mechanically compliant, and the soft mechanics of the films correlates with their high surface area and hence good electrical properties.42 Figure 11 shows scanning Table 1 (100) d-Spacings of PEDOT as a function of counterion chemistry Counterion

d100 (nm)

PAA− PSS− PTS Cl− (NaCl) C14 SO3 Cl− (CaCl2 ) EBS CSA F−

1.52 1.46 1.39 1.39 1.37 1.33 1.32 1.29 1.15

352

D. C. Martin et al.

Figure 9. Top: SEM images of the electrochemical deposition of PEDOT onto a microfabricated neural probe with a 40-µm-diameter electrode. Bottom: Thickness of the PEDOT film as a function of the total charge passed during the deposition process. Originally published by Yang, 2006.42 Used with permission. (Figure available in color online)

electron microscope (SEM) images of PEDOT nanofibrils produced over large areas on a gold-palladium-coated epoxy substrate. Cross sections of these fuzzy, nanofibrillar films can be obtained by embedding them in epoxy and microtoming. Figure 12 shows a side-by-side set of such thin sections of PEDOT shown in both the transmission electron microscope (TEM) and in the optical microscope.

Morphology of PEDOT

353

Figure 10. SEM image of PEDOT deposited into a fibrillar morphology onto the electrode of a microfabricated neural probe using poly(acrylic acid) as a counterion. Originally published by Yang et al., 2007.43 Used with permission. (Figure available in color online)

Direct correlations of images such as this make it possible to compare the optical properties of the PEDOT films with their electron scattering behavior (predominantly mass thickness contrast). Brightfield (BF) TEM images of the PEDOT films are shown in Figure 13. The dark ∼20-nm layer at the bottom is the polycrystalline gold-palladium thin film used as the conductive substrate for depositing the PEDOT. The PEDOT film itself has a nominal thickness of 500 nm for this sample. The rough outer surface of the film seen in the SEM is readily evident, as well as the local thickness and uniform density of the underlying PEDOT film. Elemental mapping of the cross-sectioned sample can be done with energy dispersive spectroscopy (EDS) in the STEM. The Z-contrast scanning transmission electron microscopy (STEM) image shows that most of the scattering comes from the gold-palladium thin support film, and sulfur maps out the position of the PEDOT film itself (Figure 14).

2.5. X-ray Photoelectron Spectroscopy of PEDOT To compliment our morphological studies of PEDOT by diffraction and microscopy, we have also performed a number of studies using XPS. XPS of PEDOT commonly uses the core spectra of carbon (C1s) near 285 eV, oxygen (O1s) near 532 eV, and sulfur (S 2p) near 165 eV to characterize counterion bonding, surface treatments, and degradation pathways (Figure 15). PEDOT XPS spectra contain peaks from four distinct types of carbon, C C/C H (285 eV), C S (285.3 eV) in the α position, C C O (286.3 eV) in the β position, and C O C (287 eV) bonding in the ethylene bridge,44 in agreement with values previously reported by J¨onsson et al.45,46 and Gelius et al.47 The carbon spectra, in general, also have an asymmetrical tail that resulted from a combination of π →π ∗ shake-up transition45,48,49 and possibly positively polarized or charged carbon.50–52 The prominence of the C 1s asymmetric tail was dependent upon the counterion used. Counterions such as phosphates will result in larger tails than PSS.44 The PEDOT characteristic C O C (533.4 eV) bonding

354

D. C. Martin et al.

Figure 11. SEM images of PEDOT nanofibrils deposited over an extended flat electrode. The nanofibrillar structure is obtained when poly(acrylic acid) is used as a counterion. (Figure available in color online)

was present in the O 1s region53 as well as a C O (532.1 eV) contribution. A second peak at the lower binding energy, here at ∼532 eV, was representative of the counterion such as the SO3 from PSS or ClO4 from LiClO4 . The PEDOT spin-split sulfur coupling, S 2p3/2 (164 eV) and S 2p1/2 (165.1 eV) with a corresponding 1.18 eV separation,54 also had a higher energy broad tail originating from positively charged sulfur within the thiophene ring (delocalization of π electrons).33,50,55,56 Higher binding energy contributions are representative of counterions such as the SO3 for PSS or SO2 from degradation.57,58

Morphology of PEDOT

355

Figure 12. Left: Low-magnification TEM image of the cross section of a PEDOT film embedded in epoxy. Right: Optical microscope image of the same PEDOT film taken in reflected light. (Figure available in color online)

Initial XPS studies on PEDOT focused on the effect of different dopants on the PEDOT binding energy in an effort to deduce how the counterion bound to PEDOT. Chemically polymerized PEDOT, via iron (III) tris-p-toluene sulfonate, was first characterized, followed by PEDOT doped with the large polymeric anion PSS or the small anion tosylate (p-methyl benzyl sulfonate) (TsO-).59 Based upon peak placement, they also found that there could be twice the quantity of PSS- present in comparison to TsO-. Greczynski et al. examined

Figure 13. Brightfield TEM images of the cross section of a ∼400-nm PEDOT thin film grown on a thin Au-Pd sputter-deposited polycrystalline metal film. The PEDOT is relatively uniform in density with an extremely rough surface texture. (Figure available in color online)

356

D. C. Martin et al.

Figure 14. Top: STEM Z-contrast of a PEDOT film on an Au-Pd substrate. Bottom: EDS elemental mapping of the same region of the sample. In this image purple is gold, blue is sulfur, and yellow is chlorine. (Figure available in color online)

the effect of counterions on PEDOT by studying the dopants poly(4-styrenesulfonic acid) (PSSH) and PSS-Na+.33,56 Zotti et al. conducted a three-part study into the doping relationship of PEDOT with counterions in comparison to their respective electrical conductivity. The first portion of the Zotti et al. study focused upon whether the polymeric structure of PSS affected the film conductivity.55 The second and third portions of the Zotti et al. study focused on how different counterions dope PEDOT and how this affects the conductivity. XPS has also been utilized to deduce the chemical binding energies of PEDOT after surface treatments, such as an acid and/or heat treatment of the film.33,56 The effects of different solvents on PEDOT60,61 have been studied, as well as PEDOT degradation. The degradation studies focused on the atmospheric,57 UV-light,57,58 UV-ozone,62 and electron bombardment degradation mechanisms.57,63 In addition to determining information on various counterion bonding, surface treatments, and degradation pathways, XPS has been used to verify the presence of additives within PEDOT or PEDOT derivatives, such as PEDOT/PSS with poly(ethylene glycol),64

Morphology of PEDOT

357

Figure 15. XPS spectra of the carbon (∼288 eV), oxygen (∼533 eV), and sulfur (∼164 eV) regions for PEDOT films. (Figure available in color online)

adenosine triphosphate (ATP),65 gold nanoparticles,66 and PEDOT-coated latex spheres.50 Other studies have focused on PEDOT binding with substrate materials, such as aluminum67 and indium tin oxide.68 2.6. FTIR of PEDOT In many studies FTIR was used in conjunction with XPS to determine the chemical composition of the respective PEDOT. FTIR was also used to help interpret electrochemical characterization. Infrared spectroscopy has been used to investigate the chemical composition of PEDOT with respect to dopant/copolymer,69–71 oxidation state.72–74 and polymerization method.31,32,75 Tehrani et al. used FTIR to determine the effects of overoxidation

358

D. C. Martin et al.

on PEDOT.76 The motivating force behind a significant amount of infrared (IR) work was to prove that the polymerization of PEDOT occurred or to directly compare two separate polymers like polypyrrole and PEDOT.74 Initial IR work focused on in situ PEDOT polymerization and the effects of oxidation and reduction upon PEDOT.72 The in situ PEDOT polymerization was first studied by Kvarnstr¨om et al.72 The results found that the PEDOT spectrum contained peaks at 1,512, 1454, 1394, 1370, and 1168 cm−1, which were attributed to the C C and C C stretch of the thiophene ring. Carbon–sulfur bonding within the thiophene ring was found to contribute peaks at 930, 830, 727, and 697 cm−1. Peaks at 1183, 1144–1128, 1093–1076, and 1052–1047 cm−1 were attributed to the alkylenedioxy group.75 Other studies focused on oxidation.72 This work went on to attribute the 1319, 1200, 1100, and 978 cm−1 bands to the doping bands during oxidation. The Kvarnstr¨om et al. work further expanded by exploring the IR active vibration (IRAV) spectrum differences between p- and n-doped PEDOT.73 This work found p-doping bands at 1513, 1319, 1195, 1090, 1060, 980, and 849 cm−1 and n-doping bands at 1285, 1245, 1195, 1090, and 1060 cm−1 with a maximum of free charge carrier absorption above 4,000 cm−1. 2.7. PEDOT-LiClO4 Flowers Local instabilities during film growth can lead to dramatic morphologies of PEDOT films, such as the flower-shaped structures shown in Figure 16. This film was deposited at 135 µA for 10 min on a 6-mm Au/Pd sputter coated on a polystyrene coverslip electrode. The

Figure 16. PEDOT films with a locally flower-shaped morphology formed on a silicon substrate, presumably due to bubble formation during the deposition process. (Figure available in color online)

Morphology of PEDOT

359

PEDOT-LiClO4 structures were likely caused by the PEDOT growing around an air bubble on the working electrodes surface, which then burst, resulting in the flower-like structure.77,78 2.8. Low-Voltage Electron Microscopy of PEDOT Figure 17 shows a low-voltage transmission electron micrograph of a dispersion of PEDOT/PSS (Baytron P) on a thin (∼2.5 nm) amorphous carbon substrate. The PEDOT molecules and molecular aggregates are the thin, filamentous structures, and the PSS is the

Figure 17. Plan-view LVTEM images of Baytron P dispersed on a 2.5-nm amorphous carbon substrate. (Figure available in color online)

360

D. C. Martin et al.

Figure 18. LVTEM of PEDOT crazes. The crazes show local fibrillation and eventually final complete rupture to form a true crack. (Figure available in color online)

more uniform, gray areas connecting the PEDOT strands. Analysis of images similar to Figure 17 have made it possible to estimate the local persistence length of PEDOT.79 This analysis gives an estimate of 27.0 nm. Because of the limited beam penetration of the low-voltage scanning electron microscopy (LVEM, or low-voltage SEM), cast films of PEDOT become opaque if greater than about 50 nm in thickness. However, PEDOT films have been found to develop crazes during deformation. Examples of crazes in deformed PEDOT films are shown in Figure 18. The process of craze opening, fibril extension, and crack formation of the PEDOT could be directly monitored during imaging in the LVEM. The PEDOT crazes show local fibrillation and eventually final rupture to form true cracks. Figure 19 shows an LVSEM image of an electrochemically deposited PEDOT film that showed evidence for local variations in composition or structure due to the details of

Figure 19. LVSEM of phase-separated PEDOT/PSS:GOD films. Left side: “Material” SEM contrast mode (A+B signal); right side: “relief” contrast mode (A-B). (Figure available in color online)

Morphology of PEDOT

361

processing. The uniformity of the secondary electron contrast reveals that the film structure was locally relatively homogeneous within the two different domains. This particular structure was found during a film that may have experienced oxidation during the deposition process. 2.9. HREM of PEDOT Like other ordered, conjugated polymers, PEDOT can be imaged using HREM techniques using low-dose methods. However, the limited crystallinity of most PEDOT preparation has meant that there is relatively little order in most cases. An example of some HREM images from PEDOT crystals using low-dose techniques is shown in Figure 20. This image is shown with local digital fast fourier transforms (FFTs) superposed on the image

Figure 20. Low-dose HREM image of PEDOT film. There is only weak order in the film, although some regions have evidence for crystallites with ∼1.4 nm (100) fringes corresponding to stacks of PEDOT chains arranged into layers, consistent with the XRD results. (Figure available in color online)

362

D. C. Martin et al.

Figure 21. HREM of PEDOT with superimposed FFTs showing ∼1.4-nm (100) fringes in different regions of the image. (Figure available in color online)

in Figure 21, revealing that there is a single predominant spacing corresponding to the ∼1.4-nm side-to-side packing of the rigid PEDOT molecules. The HREM images show that the PEDOT molecules are evidently locally quite rigid in solid state, with little or no evidence yet for the molecular bending that has been observed in other conjugated polymers like poly(nonylbithiazole) and poly(nopthalene benzobisoxazole).80,81 Figure 22 is a higher-magnification view showing the HREM lattice images of PEDOT more closely. 2.10. Liquid-Crystalline PEDOT It has been shown the PEDOT can be polymerized in surfactant mesophases, leading to templating of the liquid-crystalline structure on the resulting polymer.82–84 Figure 23 is an optical micrograph of a diffusion couple between NP9 surfactant and a PEDOT solution, showing the different characteristic textures that develop, depending on the local symmetry of the most stable mesophase. Figure 24 shows a film of PEDOT templated from the hexagonally ordered phase.84 We have recently reported on the polymerization of PEDOT into bicontinuous cubic phases, resulting in the formation of structures that presumably have both electronic and ionic transport capabilities.85

3. Highly Ordered PEDOT Phases When using bromine as a counterion during electrochemical deposition, extraordinarily large crystals are reproducibly formed. Optical micrographs show highly birefringent, needle-shaped crystals that are hundreds of micrometers in size (Figure 25). This ordered microstructure is quite different than the limited order seen when polymerizing PEDOT with other counterions. The detailed crystallography and composition of these samples are still

Morphology of PEDOT

363

Figure 22. Higher magnification low-dose HREM image of PEDOT, showing locally crystalline domains with ∼1.4 nm (100) fringes. The fringes are quite straight and rigid, with little evidence for local bending seen in HREM images of other conducting polymers. This is consistent with the relatively stiff, rigid nature of the PEDOT chain, as anticipated from its chemical structure. (Figure available in color online)

under active investigation. The crystals are highly textured with a preferred growth plane parallel to the substrate surface. They have well-defined, distinct facets and show sharp melting points in the DSC. FTIR and UV-Vis spectroscopy show characteristic absorptions similar to those seen in PEDOT. XPS has confirmed the presence of peaks characteristic of PEDOT, confirming that at least oligomers have formed. SEM images show an anisotropic fibrillar surface texture oriented perpendicular to the long axes of the needle-shaped crystals. Electron diffraction also indicates sharp, highly oriented reflections. Although the details of these interesting structures are still under active investigation, we hypothesize that the large crystals may have formed due to simultaneous crystallization of the EDOT during electrochemical polymerization in the presence of bromine. This process is somewhat similar to the crystallization-induced transesterification reactions that have been observed in certain types of polyesters.86,87

3.1. PEDOT in Gels There are a number of distinct materials and morphologies possible from joining PEDOT and hydrogels in various ways, each having unique characteristics and potential applications. Some of the early work on a hybrid conducting polymer hydrogel focused on the electropolymerization of polypyrrole in a polyacrylamide gel matrix surrounding the active electrode, which resulted in a fully dense, opaque black electroactive gel.88 Other studies focused on a composite conducting polymer hydrogel material comprised of PEDOTPSS in colloidal suspension (aka Baytron-P) and polyvinylpyridine that when joined with

364

D. C. Martin et al.

Figure 23. Optical micrograph of a diffusion couple formed between water and the nonionic surfactant NP-9. The concentration of nonionic surfactant increases from left to right, corresponding to the formation of a number of different phases with different characteristic textures including M (micellar), H (hexagonal), G (bicontinuous cubic), and L (lamellar). The texture is shown in cross-polarized light (right) and with a full-wave red filter (left). (Figure available in color online)

horseradish peroxidase could be utilized as a bioactive three-dimensional (3D) enzyme sensor.89,90 These studies opened the door for the possibility of hybrid conducting polymer hydrogel coatings that could mediate the mechanical mismatch at the tissue–device interface

Figure 24. Optical micrograph of PEDOT templated in a hexagonal liquid-crystalline mesophase. (Figure available in color online)

Morphology of PEDOT

365

Figure 25. Optical micrographs of electrochemically deposited PEDOT film, using bromine as a counterion. Top image: Brightfield; lower image: in polarized light with a full-wave red filter. The films have highly regular, birefringent, faceted crystals that are tens of micrometers in size. (Figure available in color online)

without compromising device electrical properties that are often already attenuated by foreign body responses and scarring around implanted devices. This mechanical mismatch is of particular concern in soft tissues such as the brain, a tissue that also has the added challenge to interface brain cells with high impedance microelectrodes tens of micrometers in diameter, a size scale similar to the cells themselves. Therefore, thin hydrogel coatings for penetrating neural microelectrodes were developed that contained electropolymerized nanoscale networks of conducting polymer within the hydrogel matrix extending out from individual microelectrode sites on the device (Figure 26).42,91 These hybrid conducting polymer–hydrogel materials were based on calcium cross-linked alginate hydrogels and either electrochemically polymerized PEDOT or polypyrrole. The coatings could be dehydrated for device insertion, rehydrated in vivo, and were found to be functional, recording neural signals for multiple days following implantation.92

366

D. C. Martin et al.

Figure 26. PEDOT grown through a hydrogel coated onto a microfabricated neural electrode. The microelectrode was itself coated with a thin layer of electrospun nanofibers. As seen from the side and top views, the PEDOT grows up and off the metal electrode, creating a black cloud of polymer that reaches toward the hydrogel surface. Originally published by Abidian and Martin, 2009.41 Used with permission. (Figure available in color online)

Unlike previous conducting polymer–hydrogel materials in which the conducting polymer was nonspecifically or densely incorporated into the hydrogel, the morphology of the PEDOT within the gel matrices formed a diffuse yet very low-density conducting polymer network extending tens to hundreds of micrometers out into the gel scaffold while still maintaining an electrical connectivity with the underlying active electrode. BF TEM of the PEDOT in gel revealed that the PEDOT appeared to have used the cross-linked alginate fibrils as a scaffold for polymerization, wrapping itself around the fibrils like bark on a tree (Figure 27). When compared with the unmodified gel, the open pore structure of the hybrid PEDOT-gel was retained, with the PEDOT-wrapped fibrils appearing as somewhat fatter, more electron dense (darker) filaments. The hydrogel matrices or coatings

Morphology of PEDOT

367

Figure 27. Brightfield TEM of an alginate gel (top) showing the extended, interconnected, and cross-linked polysaccharide chains. After growing the PEDOT through the gel, the microstructure remains open and infiltrated with water (bottom). (Figure available in color online)

containing the PEDOT–gel nanofibrillar networks appeared visibly darker yet not opaque or densely black–blue, which might be expected of a PEDOT hydrogel. Figure 28 shows schematically the hypothesized mechanisms of electropolymerization of PEDOT within a hydrogel matrix. The positively charged EDOT monomer has low solubility in water and so it likely associates in close contact with the negatively charged macromolecules. During

368

D. C. Martin et al.

Figure 28. Schematic of PEDOT growth through a gel. Because the EDOT monomer (circles) is not that soluble in water, it presumably precipitates out onto the aginate filaments, creating an EDOTrich zone around the polysaccharide backbone (b). During electrochemical polymerization (c), the PEDOT grows preferentially around the alignate filaments, creating an open network of PEDOT that encapsulates the alginate molecules and makes it possible for water to remain infiltrated into the PEDOT-alginate network (d). (Figure available in color online)

Morphology of PEDOT

369

polymerization, these EDOT-rich zones wrap themselves around the gel scaffold, leaving the intercalating water behind. In addition to retaining their porosity, the hydrogels containing the PEDOT networks also remain swollen with water or biological fluids such as saline or cell media. Accordingly, it was found that live cells could be incorporated into the hydrogel matrices followed by electrochemical polymerization of PEDOT networks within the cell-seeded hydrogels. This resulted in formation of electrically connected PEDOT–alginate hydrogel networks surrounding the living cells. If joined with a specialized electrode device, it is not difficult to imagine that such a material could be further developed into a novel electroactive tissue or cell scaffold for real-time interfacing with live cells in a 3D microenvironment. 3.2. Cells on PEDOT Electrochemically polymerized PEDOT materials are generally considered biocompatible and safe as substrates for living cells, particularly when utilizing counterions or dopants that are nontoxic.93,94 The charged surface and micro- and nanoscale rough, fuzzy features inherent to many electropolymerized PEDOT materials are attractive and pro-adhesive for many cells types including neurons, glia, muscle cells, and fibroblasts. Because PEDOT coatings adhere well to most metal electrodes on biomedical devices, when a device with coated electrode sites is explanted it is common to find numerous cells adhered to the surface of the device and the PEDOT coating itself. Figure 29 is a pseudo-colored SEM image of a microelectromechanical system (MEMS) microelectrode neural probe device with PEDOT-coated electrode sites onto which is adhered many microglia, a type of cell involved in the brain’s immune response to implanted devices. One of the many unique and valuable aspects of electrochemically polymerized PEDOT materials is the diversity of PEDOT topographies that can be achieved by discerning choice of counterion/dopant or use of a templating agent. The technique of templating PEDOT

Figure 29. Colorized SEM image of astrocytes (brown) on a neural probe with a PEDOT-coated electrode (blue). (Figure available in color online)

370

D. C. Martin et al.

Figure 30. Optical micrograph of fibroblasts grown on a film with on spherically templated PEDOT (5-µm-diameter holes). In this fluorescent image the actin cytoskeleton was labeled green and the cell nuclei blue. (Figure available in color online)

films with dissolvable polystyrene microspheres was employed to synthesize a PEDOT substrate with tailored, defined porosity. Figures 30 and 31 are optical and SEM images respectively of fibroblast cells grown on spherically templated PEDOT films (with 5-µmdiameter pores). Note the way that the cells lie within the pores as well as sends filamentous

Figure 31. SEM image of fibroblasts grown a film with spherically templated PEDOT (5-µmdiameter holes). (Figure available in color online)

Morphology of PEDOT

371

cellular processes outside the pores and along the edges to contact neighboring cells (Figure 31). 3.3. PEDOT around Cells Previous studies showed that PEDOT films and matrices can be electrochemically polymerized in the presence of living cells and that the cells remain viable within the PEDOT material for the short term. If the cell material is removed, the resulting PEDOT material bears cell-shaped holes and tunnels similar to the way that polystyrene microspheres can be used to template PEDOT films as described above (Figure 32). A related technique was developed to electropolymerize PEDOT within living tissue such as the skin, peripheral nerve, muscle, and brain, a tissue in which it was found that the PEDOT could be extended out from the active electrode for several hundreds of micrometers95,96 (Figure 33). Similar to findings with electropolymerized PEDOT–hydrogels, it was found that the in situ/in vivo electrochemical polymerization process results in a hybrid PEDOT–tissue material that electrically joins the active electrode to the tissue via a diffuse nanoscale PEDOT network that penetrates the tissue likely using the extracellular matrix material as a scaffold for polymerization. It has also been shown possible to polymerize PEDOT in skin (Figure 34), muscle, and peripheral nerve (Figure 35). This exciting and promising area of research represents a new paradigm for interfacing living tissue with biomedical devices, yet many questions remain, the most obvious being related to the long-term performance and mechanical properties of such a hybrid PEDOT–tissue material and to the extent of tissue injury around the reaction site. In order to begin answering some of the questions related to tissue injury during in situ deposition, we examined by BF TEM of the mouse brain tissue containing the in situ/in vivo electrochemically polymerized PEDOT. Figure 36 shows BF TEM of PEDOT polymerized directly within mouse brain slice using a 75-µm Teflon-insulated Au microwire. The data indicated that PEDOT uses anatomical features as scaffolds for polymerization; the polymer appears to forms dense regions on or around the white matter tracts in the corpus collosum. The PEDOT network polymerized within brain tissue, showing the formation of dark PEDOT coatings on extracellular matrix and cellular processes/filaments. Furthermore, the TEM images of the in situ/in vivo electrochemically polymerized PEDOT in the central nervous systems (CNS) reveals less tissue damage when polymerizing PEDOT than when simply running the current alone through the tissue (Figure 36). Presumably this is because the polymerization reaction is helping to protect the integrity of the tissue by using the electrical current to form PEDOT from EDOT rather than simply cause irreversible redox reactions and harmful chemical reaction by-products in the tissue. Figure 37 shows a mouse brain tissue control sample exposed to current but in the absence of EDOT monomer. The open area at the top of the micrograph is the hole left by the microwire deposition electrode, which was removed prior to sectioning. Note that in the control tissue, the edge near the electrode hole has a jagged, torn, and disorganized appearance as opposed to the edge of the tissue, which was injected with EDOT monomer when electrical current was applied and thus appears smoother, darker, and appears more interconnected, possibly due to the presence of the PEDOT network wrapped around the extracellular matrix components. The higher magnification view of tissue exposed to current only shows that the tissue has a vacuolized and disorganized appearance as well as a noted absence of intact mitochondria. Also, the edge of the PEDOT-infused tissue nearest the electrode has a darker appearance, indicating a more electron-dense region that is likely due to the presence of the PEDOT network around the extracellular components. In general the

372

Figure 32. SEM images (top) and optical micrographs (bottom) of PEDOT grown around cells in vitro. The actin cytoskeleton (labeled green) shows disruption after being embedded in PEDOT. The extent of this cytoskeletal disruption can be reduced by incorporating biologically active agents (with integrin binding sites) in the PEDOT. Originally published by Richardson-Burns et al., 2007.94 Used with permission. (Figure available in color online)

Morphology of PEDOT

373

Figure 33. Brain-DOT: PEDOT polymerized in living brain tissue slices. Originally published by Richardson-Burns et al., 2007.95 Used with permission. (Figure available in color online)

tissue containing the in situ electropolymerized PEDOT has a normal appearance with an abundance of healthy mitochondria and an organized, interconnected extra- and intracellular matrix. In contrast to the control tissue that received electrical current only (no monomer) in which the exposure to the electrical current appears to have had a deleterious effect on the tissue, exposure of the tissue to the electrical current in the presence of monomer solution resulted in electrochemical polymerization of the PEDOT without damaging the tissue. TEM images of brain tissue collected from a region far from the electrode insertion site in the brain tissue in which PEDOT was polymerized show that the tissue has a normal,

Figure 34. Skin-DOT: PEDOT polymerized in living skin tissue. (Figure available in color online)

374

D. C. Martin et al.

Figure 35. Peripheral nerve-DOT: PEDOT polymerized around living peripheral nerves. (Figure available in color online)

Figure 36. TEM of brain-DOT near an implanted electrode showing effect of current on resulting tissue damage. Current alone causes considerable disruption of the cellular architecture (a). There is significantly less disruption of the tissue when the current is used to polymerize the PEDOT (b). (Figure available in color online)

Morphology of PEDOT

375

Figure 37. TEM of brain-DOT showing morphological details of in situ polymerization. (Figure available in color online)

healthy appearance with numerous healthy mitochondria, organized intra- and extracellular matrix, and cell nuclei with normal morphology and heterochromatin structure indicative of healthy cells.

4. Experimental Methods 4.1. Sample Preparation Samples of a commercial PEDOT/PSS suspension (Baytron P, obtained from H. C. Starck) were prepared by atomizing onto ultrathin amorphous carbon support films. The

376

D. C. Martin et al.

suspensions were diluted with deionized water. For LVEM imaging (∼5 kV), we have found that approximately 2.5-nm thin films of amorphous carbon were still strong enough for handling and provided the minimal contrast needed for imaging with this microscope. For HREM (300 kV), we were able to use somewhat thicker support films (∼20 nm). Cross sections of PEDOT films were obtained by electrochemical deposition from a metal film deposited by evaporation onto an epoxy substrate. An encapsulating layer of epoxy was applied after the reaction was completed, and then the sandwich was sectioned in a Reichert FC4 ultramicrotome (Depew, New York, USA) at room temperature. We obtained better sections using an oscillating diamond knife. Excellent, uniform sections were obtained at 25-nm thickness. 4.2. In Situ Electrochemical Polymerization of PEDOT Directly in Tissue and Hydrogels The target site within the tissue or cross-linked hydrogel material was infiltrated with an aqueous EDOT monomer solution containing 0.01 M EDOT (H. C. Starck), 0.02 M poly(sodium styrene sulfonate) (PSS; Sigma-Aldrich, St. Louis, Missouri, USA) prepared in phosphate-buffered saline using either direct injection of monomer through a syringe or submersion of the sample in an excess of monomer. Samples were infiltrated with monomer for 5–15 min prior to PEDOT polymerization and then were immobilized on/in a gelatin scaffold. The electrochemical polymerization was performed by inserting the working electrode (WE) directly into the target site and a counterelectrode (CE) into a site in the tissue or gel at least 5–50 mm away from the WE. Ideally the WE is a Teflon-insulated Au or Pt microwire (25–100 µm diameter) with exposed wire restricted to the tip (either blunt cut or approx. 200-µm-length exposed tip) and the CE is a Pt wire 5–10× wider in diameter than the WE. To polymerize the PEDOT in the tissue, constant current is applied to the WE (1–10 µA), which results in voltages ranging from 1.5–2.5 V (anything over 2.5 V causes water hydrolysis and this is to be avoided when working with living tissues or cell in hydrogels). The electric current is applied using an AutoaLab Potentiostat/Galvonostat (MetrOhm, EcoChemie, Riverview, Florida, USA) for 5–60 min depending on the desired size and density of the resulting PEDOT “cloud” within the tissue or gel. The PEDOT cloud typically grows outward concentrically from the implanted wire and under some circumstances can grow more than 1 mm out from the wire tip. 4.3. TEM Sample Preparation Tissue was fixed overnight in 2.5% glutaraldehyde and then 1 h in 1% osmium tetroxide in phosphate-buffered saline (PBS) then rinsed two times with deionized water (dH2 0). The tissue was then stained with 3% uranyl acetate in dH2 0 for 1 h and then ethanol-dehydrated in a series of ethanol solutions of increasing concentration (15 min each; 50, 70, 90%; two times 100%) Next, the samples were infiltrated with a 3:1 ratio of 100% ethanol to Epon resin, then a 1:1 ratio of 100% ethanol to resin for 1 h each. The samples were incubated overnight in a 1:3 ratio of 100% ethanol to resin and then into 100% resin the next day for 4 h and transferred to fresh resin to prevent resin polymerization. The samples were then arranged in molds and vacuum dessicated for 6 h. Finally, the resin in the samples was polymerized in the oven at 60◦ C for 24 h. After baking, the resin was hard, so rectangular slabs were cut from the blocks using a hand saw. Sections 70 nm thick were cut from the blocks using an ultramicrotome (Reichert Ultracut-E) and placed on copper TEM grids. These samples were now ready for the transmission electron microscope Philips CM-100

Morphology of PEDOT

377

(FEI Company, Hillsboro, Oregon) at the University of Michigan Microscopy and Image Analysis Laboratory (MIL). A few tissue sample sections of approximately 4 µm thick were also cut from the blocks for staining by hematoxylin and eosin dyes to aide in anatomical localization and histological analysis. These were imaged using a Zeiss Axioplan upright light microscope (Carlzeiss Microimaging, Inc., Thornwood, New York, USA).

4.4. Microsphere Templating on Electrode Substrate Monodisperse solutions of latex microspheres (PolySciences, Warrington, PA, or Duke Scientific, Division of Thermo-Fisher Scientific, Waltham, Massachusetts, USA) were prepared to concentrations of approximately 2% solid by volume by spinning down the stock solution, extracting the supernatant, and then resuspending the spheres in chosen volume with sonication. The sphere solution was then deposited onto the chosen substrate and dried by air evaporation or a vacuum desiccator overnight. The electrode substrates used in these studies were flat conductive indium tin oxide (ITO)-coated glass (5 × 10 mm). The microspheres spontaneously form ordered patterns determined by particle diameter and density. Following electrochemical polymerization of the conducting polymer around the microspheres (typically leaving the upper half of the sphere exposed), the electrode was immersed in toluene for 24 h to dissolve the microspheres, resulting in a pore-templated conducting polymer film on the ITO surface. Multilayered microporous networks were generated by either (a) simultaneously evaporating multiple layers of spheres by constraining lateral flow of the solution or (b) performing layer-by-layer microsphere assembly/evaporation and then electrochemical polymerization. Method (a) is faster but less controllable.

4.5. Electrochemical Polymerization of Microporous PEDOT Films The electrode or electrically conductive substrate was placed in an electrically connected reservoir containing the aqueous monomer solution comprised of 0.01 M EDOT (H. C. Starck) in deionized water with 0.02 M polyanionic poly(sodium styrene sulfonate). To polymerize the PEDOT films, galvanostatic current (1–10 µA/mm2) was applied to the electrode and the monomer solution using an AutoaLab Potentiostat/Galvonostat for 1–60 min depending on the diameter of the microspheres, surface area to be covered, and the desired thickness of the polymer film.

4.6. Cell Culture SH-SY5Y neuroblastoma-derived cells were maintained in Dulbecco’s modified Eagle media (DMEM with glucose, with l-glutamine; Gibco/Invitrogen, Carlsbad, CA) supplemented with Pen-Strep mixed antibiotic solution (dilute 1:100 in cell media; Gibco/Invitrogen) and 10% fetal bovine serum (FBS; Gibco/Invitrogen). For experiments in which the PEDOT substrates were biofunctionalized with cell adhesion molecules, cells were maintained in serum-free media. Primary motor neuron culture is described in detail in Corey.97 Briefly, spinal cords of E15 Sprague-Dawley rats were isolated, chopped, and dissociated in 0.05% trypsin/EDTA for 15 min at 37◦ C. Cells were isolated in 5.4% Optiprep in L-15 media by centrifugation for 15 min at 1000 g. Motor neurons were collected and plated in neurobasal culture medium supplemented with 2% B27 and l-glutamine (2 µM).

378

D. C. Martin et al.

4.7. Cell Staining Cells were fixed in 3.7% formaldehyde/PBS at room temperature (RT) for 30 min to 1 h. The F-actin cytoskeleton was labeled by Phalloidin-Oregon Green (Invitrogen-Molecular Probes; 1:300 in PBS, Carlsbad, California, USA) for 1 h at RT or overnight at 4◦ C. Dead cells were identified by positive propidium iodide staining and all nuclei were counterstained with Hoechst 33342 (both from Invitrogen Live/Dead staining kit). For SEM cells were fixed using 1% gluteraldehyde, dehydrated in ethanol series of increasing concentration (50, 75, 95, 100%, 10 min in each) and then immersed overnight in hexamethyldisilazine (HMDS), placed in the fume hood, and allowed to evaporate completely overnight. Cells and substrates were argon plasma-sputtered with a nanometer thin layer of Au prior to SEM imaging.

5. Microscopy and Spectroscopy Optical microscopy was done on a Nikon Optiphot-POL (Mager Scientific, Dexter, Michigan, USA) in transmitted or reflected light. The microscope was also equipped with polarizers and analyzers and a full-wave red filer. Digital images were acquired with a SPOT-RT camera (Diagnostic Instruments; Mager Scientific, Dexter, Michigan, USA). Wide-angle X-ray diffraction patterns were acquired with a Bruker D8 Discover system (Bruker AXS Inc., Madison, Wisconsin, USA), equipped with a 2D position-sensitive wire array detector. Samples were typically oriented in a near-grazing incidence geometry, with the film plane horizontal and the Cu Kα X-ray source at an incoming angle of ∼2 degrees, and the detector collecting the diffracted X-rays from ∼5 to 40 degrees 2θ . LVEM images were obtained in SEM, STEM, and TEM modes on a table-top-sized LVEM-5 from DeLong Instruments (Brno, Czech Republic). The operating voltage was ∼5 kV, and TEM images were acquired using a charge-coupled detector (CCD) camera from the YAG screen of the LVEM. TEM imaging was performed with a 200 kV JEOL 2010F STEM and a JEOL 3011 HREM at the EMAL at the University of Michigan and in the W. H. Keck Electron Microscope laboratory at the University of Delaware. We have recently described the use of low-dose HREM techniques for beam-sensitive polymers. The estimated critical dose of PEDOT has been determined to be ∼0.1 C/cm2. This correlates with the established relationship between electron beam sensitivity and thermal stability.20,98 TEM images were also obtained a Philips CM-100 at 60 kV at the University of Michigan Morphology and Image Analysis core facility. Digital images were acquired using a CCD. XPS spectra were obtained with a Kratos Axis Ultra DLD X-ray Photoelectron Spectrometer (Kratos Analytical Ltd., Manchester, UK) with a monochromatic aluminum source at a vacuum pressure of 10−8 to 10−9 Torr. Initial survey scans were run with a pass energy of 160 eV, and characteristic region scans for C1s, O1s, and S 2p peaks were obtained with a pass energy of 20 eV and step size of 0.1 eV. All spectra were reference to the C C/C H peak at 285 eV. The data were peak-fitted by using the CasaXPS software (Casa Software, Ltd., Teignmouth, UK) using Gaussian/Lorentzian and asymmetry corrections. Molecular models of the PEDOT structure were built using Cerius2 (SGI, UNIX) and Materials Studio 4 (Windows) software (Accelrys, Inc.) and analyzed using the Dreiding and Compass force fields. Diffraction patterns and HREM images were generated using the modules available in these software packages. Crystal structures were also generated and examined using CrystalMaker and SingleCrystal on Macs.

Morphology of PEDOT

379

Acknowledgements The authors gratefully appreciate funding from the National Science Foundation, the National Institute of Health, the Army Research Office, the University of Michigan College of Engineering GAP funding program, Biotectix LLC, the Michigan University Commercialization Initiative, and the Defense University Research Instrumentation Program. The assistance of Dr. John Mansfield and Dr. Haiping Sun at the Electron Microscopy and Analysis Laboratory (EMAL) is appreciated. The support of Ying Qi in the Materials Science and Engineering J. D. Hanawalt X-ray Microanalysis Laboratory (XMAL) is acknowledged. Andre Pergeron, a summer student from Greenhills High School in Ann Arbor, assisted in the LVEM of PEDOT. Dotty Sorenson from the Medical Imaging Laboratory (MIL) in the Medical School provided technical support in the embedding, microtoming, and TEM imaging of alginate and PEDOT-coated alginate hydrogels. DCM, JLH, and SRB are cofounders of Biotectix, a University of Michigan spin-off company interested in the commercialization of materials for interfacing electronic biomedical devices with living tissue.

References 1. Forrest, S. R. “The path to ubiquitous and low-cost organic electronic appliances on plastic,” Nature, 2004, 428, 911–918. 2. Batchelder, D. N. “Colour and chromism of conjugated polymers,” Contemporary Physics, 1988, 29, 3–31. 3. Groenendall, L. B.; Jonas, F.; Freitag, D.; Pielartzik, H.; Reynolds, J. R. “Poly(3,4ethylenedioxythiophene) and its derivatives: Past, present, and future,” Advanced Materials, 2000, 12, 481–494. 4. Groenendaal, L. B.; Zotti, G.; Aubert, P.-H.; Waybright, S. M.; Reynolds, J. R. “Electrochemistry of poly(3,4-alkylenedioxythiophene) derivatives,” Advanced Materials, 2003, 15, 855–879. 5. Schlenoff, J. B.; Xu, H. “Evolution of physical and electrochemical properties of polypyrrole during extended oxidation,” Journal of the Electrochemical Society, 1992, 139, 2397–2401. 6. Kirchmeyer, S.; Reuter, K. “Scientific importance, properties, and growing applications of poly(3,4-ethylenedioxythiophene),” Journal of Materials Chemistry, 2005, 15, 2077– 2088. 7. Kirchmeyer, S.; Reuter, K.; Simpson, J. C. in Conjugated polymers: theory, synthesis, properties and characterization; Skotheim, T. A. and Reynolds, J. R., Eds.; Taylor and Francis: Boca Raton, FL, 2007. Chapter 10, pp. 10-1–10-22. 8. Cui, X.; Martin, D. C. “Electrochemical deposition and characterization of poly(3,4ethylenedioxythiophene) on neural microelectrode arrays,” Sensors and Actuators B: Chemical, 2003, 89, 92–102. 9. Asplund, M.; von Holst, H.; Ingan¨as, O. “Composite biomolecule/PEDOT materials for neural electrodes,” Biointerphases, 2008, 3, 83–93. 10. Ng, S. C.; Chan, H. S. O.; Yu, W.-L. “Synthesis and characterization of electrically conducting copolymers of ethylenedioxythiophene and 1,3-propylenedioxythiophene with w-functional substituents,” Journal of Materials Science Letters, 1997, 16, 809. 11. Lima, A.; Schottland, P.; Sadki, S.; Chevrot, C. “Electropolymerization of 3, 4ethylenedioxythiophene and 3, 4-ethylenedioxythiophene methanol in the presence of dodecylbenzenesulfonate,” Synthetic Metals, 1998, 93, 33. 12. St´ephan, O.; Schottland, p.; Le Gall, P. Y.; Chevrot, C.; Mariet, C.; Carrier, M. “Electrochemical behaviour of 3,4-ethylenedioxythiophene functionalized by a sulphonate group. Application to the preparation of poly(3,4-ethylenedioxythiophene) having permanent cation-exchange properties,” Journal of Electroanalytical Chemistry, 1998, 443, 217–226. 13. Povlich, L. K.; Cho, J. C.; Spanninga, S.; Martin, D. C.; Kim, J. “Carboxylic acid-modified EDOT for bio-functionalization of neural probe electrodes,” Polymer Preprints, 2007, 48, 7–8.

380

D. C. Martin et al.

14. Balog, M.; Rayah, H.; Le Derf, F.; Salle, M. “A versatile building block for EDOT or PEDOT functionalization,” New Journal of Chemistry, 2008, 32, 1183–1188. 15. Reeves, B. D.; Thompson, B. C.; Abboud, K. A.; Smart, B. E.; Reynolds, J. R. “Polymer based on a spiro bipropylenedioxythiophene [(poly(spiroBiProDOT)],” Advanced Materials, 2002, 14, 717–719. 16. B¨aurle, P.; Scheib, S. “Synthesis and characterization of thiophenes, oligothiophenes and polythiophenes with crown ether units in direct Pi-conjugation,” Acta Polymerica, 1995, 46, 124. 17. Rimmel, G.; B¨aurle, P. “Molecular recognition properties of crown ther-functionalized oligothiophenes,” Synthetic Metals, 1999, 102, 1323. 18. von Bekesy, G. “DC resting potentials inside the cochlear partition,” Journal of the Acoustical Society of America, 1952, 24, 72–76. 19. Ashby, M. F. Materials selection in mechanical design; Elsevier Butterworth-Heinemann: Burlington, MA, 2005. 20. Martin, D. C.; Chen, J.; Yang, J.; Drummy, L. F.; K¨ubel, C. “High resolution electron microscopy of ordered polymers and organic molecular crystals: Recent developments and future possibilities,” Journal of Polymer Science: Physics, 2005, 43, 1749–1778. 21. Drummy, L. F.; Yang, J.; Martin, D. C. “Low voltage electron microscopy of polymer and organic molecular thin films,” Ultramicroscopy, 2004, 99, 247–256. 22. Sadki, S.; Schottland, P.; Brodie, N.; Sabouraud, G. “The mechanisms of pyrrole electropolymerization,” Chemical Society Reviews, 2000, 29, 283–293. 23. Schuhmann, W. in Immobilized biomolecules in analysis: A practical approach; Cass, T. and Ligler, F. S., Eds.; Oxford: New York, 1998; 198. 24. Kim, J.; Kim, E.; Won, Y.; Lee, H.; Suh, K. “The preparation and characteristics of conductive poly(3,4-ethylenedioxythiophene) thin film by vapor-phase polymerization,” Synthetic Metals, 2003, 139, 485–489. 25. Winther-Jensen, B.; Breiby, D. W.; West, K. “Bas inhibited oxidative polymerization of 3,4ethylenedioxythiophene with iron (III) tosylate,” Synthetic Metals, 2005, 152, 1–4. 26. Winther-Jensen, B.; West, K. “Vapor-phase polymerization of 3,4-ethylenedioxythiophene: A route to highly conducting polymer surface layers,” Macromolecules, 2004, 37, 4538–4543. 27. Winther-Jensen, B.; Forsyth, M.; West, K.; Andreasen, J. W.; Bayley, P.; Pas, S.; MacFarlane, D. R. “Order-disorder transitions in poly(3,4-ethylenedioxythiophene),” Polymer, 2008, 49, 481–487. 28. Aasmundtveit, K. E.; Samuelsen, E. J.; Pettersson, L. A. A.; Ingan¨as, O.; Johansson, T.; Feidenhans’l, R. “Structure of thin films of poly(3,4-ethylenedioxythiophene),” Synthetic Metals, 1999, 101, 561–564. 29. Aasmundtveit, K. E.; Samuelsen, E. J.; Ingan¨as, O.; Pettersson, L. A. A.; Johansson, T.; Ferrer, S. “Structural aspects of electrochemical doping and dedoping of poly(3,4ethylenedioxythiophene),” Synthetic Metals, 2000, 113, 93–97. 30. Lock, J. P.; Im, S. G.; Gleason, K. K. “Oxidative chemical vapor deposition of electrically conducting poly(3,4-ethylenedioxythiophene) films,” Macromolecules, 2006, 39, 5326– 5329. 31. Im, S. G.; Gleason, K. K. “Systematic control of the electrical conductivity of poly(3,4ethylenedioxythiophene) via oxidative chemical vapor deposition,” Macromolecules, 2007, 40, 6552–6556. 32. Im, S. G.; Olivetti, E. A.; Gleason, K. K. “Systematic control of the electrical conductivity of poly(3,4-ethylenedioxythiophene) via oxidative chemical vapor deposition (oCVD),” Surface and Coatings Technology, 2007, 201, 9406–9412. 33. Greczynski, G.; Kugler, T.; Salaneck, W. R. “Characterization of the PEDOT-PSS system by means of X-ray and ultraviolet photoelectron spectroscopy,” Thin Solid Films, 1999, 354, 129–135. 34. Timpanaro, S.; Kemerink, M.; Touwslager, F. J.; de Kok, M. M.; Schrader, S. “Morphology and conductivity of PEDOT/PSS films studied by scanning-tunneling microscopy,” Chemical Physics Letters, 2004, 394, 339.

Morphology of PEDOT

381

35. Lang, U.; M¨uller, E.; Naujoks, N.; Dual, J. “Microscopical investigations of PEDOT:PSS thin films,” Advanced Functional Materials, 2009, 19, 1215–1220. 36. Ahmad, S.; Deepa, M.; Singh, S. “Electrochemical synthesis and surface characterization of poly(3,4-ethylenedioxythiphene) films grown in an ionic liquid,” Langmuir, 2007, 23, 11430–11433. 37. D¨obbelin, M.; Marcilla, R.; Salsamendi, M.; Pozo-Gonzalo, C.; Carrasco, P. M.; Pomposo, J.A.; Mecerreyes, D. “Influence of ionic liquids on the electrical conductivity and morphology of PEDOT:PSS films,” Chemistry of Materials, 2007, 19, 2147–2149. 38. Yang, J.; Martin, D. C. “Microporous conducting polymers on neural prosthetic devices. II. physical characterization,” Sensors and Actuators A: Physical, 2004, 113, 204–211. 39. Yang, J.; Martin, D. C. “Microporous conducting polymers on neural prosthetic devices. I. electrochemical deposition,” Sensors and Actuators B: Chemical, 2004, 101, 133–142. 40. Abidian, M. R.; Kim, D.-H.; Martin, D. C. “Conducting polymer nanotubes for controlled drug release,” Advanced Materials, 2006, 18, 405–409. 41. Abidian, M. R.; Martin, D. C. “Multifunctional nanobiomaterials for neural interfaces,” Advanced Functional Materials, 2009, 19, 573–585. 42. Yang, J.; Martin, D. C. “Impedance spectroscopy and nanoindentation of conducting PEDOT coatings on neural prosthetic devices,” Journal of Materials Research, 2006, 21, 1124– 1132. 43. Yang, J.; Lipkin, K.; Martin, D. C. “Electrochemical fabrication of conducting polymer poly(3,4ethylenedioxythiophene) (PEDOT) nanofibrils on microfabricated neural prosthetic devices,” Journal of Biomaterials Science—Polymer Edition, 2007, 18, 1075–1089. 44. Spanninga, S. A.; Martin, D. C.; Chen, Z. “X-ray photoelectron spectroscopy study of counterion incorporation in poly(3,4-ethylenedioxythiophene),” Journal of Physical Chemistry C, 2009, 113, 5585–5592. 45. J¨onsson, S. K. M.; Birgerson, J.; Crispin, X.; Greczynski, G.; Osikowicz, W.; Denier Van Der Gon, A. W.; Salaneck, W. R.; Fahlman, M. “The effects of solvents on the morphology and sheet resistance of poly(3,4-ethylenedioxythiophene)-polystyrenesulfonic acid (PEDOT-PSS) films,” Synthetic Metals, 2003, 139, 1–10. 46. J¨onsson, S. K. M.; de Jong, M. P.; Groenendaal, L.; Salaneck, W. R.; Fahlman, M. “Phenylcapped EDOT trimer: Its chemical and electronic structure and its interface with aluminum,” Journal of Physical Chemistry B, 2003, 107, 10793–10800. 47. Gelius, U.; Allan, C. J.; Joii.insson, G.; Siegbahn, II.; AUison, D. A.; Siegbahn, K. “The ESCA spectra of benzene and the iso-electric series, thiophene, pyrrole, and furan,” Physica Scripta, 1971, 3, 237–242. 48. Briggs, D. Surface analysis of polymers by XPS and static SIMS; Cambridge University Press: Cambridge, 1998. 49. Khan, M. A.; Armes, S. P. “Surface characterization of poly(3,4-ethylenedioxythiophene)-coated latexes by X-ray photoelectron spectroscopy,” Langmuir, 2000, 16, 4171–4179. 50. Kang, E. T.; Neoh, K. G.; Tan, K. L. “X-ray photoelectron spectroscopic studies of poly(2,2 bithiophene) and its complexes,” Physical Review B, 1991, 44, 10461–10469. 51. Riga, J.; Snauwaert, P.; De Pryck, A.; Lazzaroni, R.; Boutique, J. P.; Verbist, J. J.; Br´edas, J. L.; Andr´e, J. M.; Taliani, C. “Electronic structure of sulfur-containing conducting polymers,” Synthetic Metals, 1987, 21, 223–228. ¨ 52. Wu, C. R.; Nilsson, J. O.; Ingan¨as, O.; Salaneck, W. R.; Osterholm, J. E.; Br´edas, J. L. “Electronic structure of polythiophene,” Synthetic Metals, 1987, 21, 197–202. 53. Briggs, D.; Beamson, G. “XPS studies of the oxygen 1s and 2s levels in a wide range of functional polymers,” Analytical Chemistry, 1993, 65, 1517–1523. 54. Moulder, J. F.; Stickle, W. F.; Sobol, P. E.; Bomben, K. D. Handbook of X-ray Photoelectron Spectroscopy: A Reference Book of Standard Spectra for Identification and Interpretation of XPS Data. Eden Prairie, MN: Perkin-Elmer Corporation.

382

D. C. Martin et al.

55. Zotti, G.; Zecchin, S.; Schiavon, G. “Electrochemical and XPS studies toward the role of monomers and polymeric sulfonate counterions in the synthesis, composition, and properties of poly(3,4-ethylenedioxythiophene),” Macromolecules, 2003, 36, 3337–3344. 56. Greczynski, G.; Kugler, T.; Keil, M.; Osikowicz, W.; Fahlman, M.; Salaneck, W. R. “Photoelectron spectroscopy of thin films of PEDOT-PSS conjugated polymer blend: A mini-review and some new results,” Journal of Electron Spectroscopy and Related Phenomena, 2001, 121, 1–17. 57. Crispin, X.; Marciniak, S.; Osikowicz, W.; Zotti, G.; Denier Van Der Gon, A. W.; Louwet, F.; Fahlman, M.; Groenendaal, L.; De Schryver, F.; Salaneck, W. R. “Conductivity, morphology, interfacial chemistry, and stability of poly(3,4-ethylenedioxythiophene)-poly(styrene sulfonate): A photoelectron spectroscopy study,” Journal of Polymer Science: Part B: Polymer Physics, 2003, 41, 2561–2583. 58. Marciniak, S.; Crispin, X.; Uvdal, K.; Trzcinski, M.; Birgerson, J.; Groenendaal, L.; Louwet, F.; Salaneck, W. R. “Light induced damage in poly(3,4-ethylenedioxythiophene) and its derivatives studied by photoelectron spectroscopy,” Synthetic Metals, 2004, 141, 67–73. 59. Xing, K. Z.; Fahlman, M.; Chen, X. W.; Ingan¨as, O.; Salaneck, W. R. “The electronic structure of poly(3,4-ethylenedioxythiophene): Studied by XPS and UPS,” Synthetic Metals, 1997, 89, 161–165. 60. Kim, T. Y.; Kim, J. E.; Suh, K. S. “Effects of alcoholic solvent on the conductivity of tosylate-doped poly(3,4-ethylenedioxythiophene) (PEDOT-OTs),” Polymer International, 2006, 55, 80–86. 61. Lee, C. S.; Joo, J., Han, S.; Koh, S. K. “An approach to durable pvdf cantilevers with highly conducting PEDOT/PSS (DMSO) electrodes,” Sensors and Actuators A: Physical, 2005, 121, 373–381. 62. Tengstedt, C.; Kanciurzewska, A.; de Jong, M. P.; Braun, S.; Salaneck, w. R.; Fahlman, M. “Ultraviolet light-ozone treatment of poly (3,4-ethylenedioxy-thiophene)-based materials resulting in increased work functions,” Thin Solid Films, 2006, 55, 2085–2090. 63. Denier Van Der Gon, A. W.; Birgerson, J.; Fahlman, M.; Salaneck, W. R. “Modification of PEDOT-PSS by low energy electrons,” Organic Electronics, 2002, 3, 111–118. 64. Wang, T.; Qi, Y.; Xu, J.; Hu, X.; Chen, P. “Effects of poly(ethylene glycol) on electrical conductivity of poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonic acid) film,” Applied Surface Science, 2005, 250, 188–194. 65. Paczosa-Bator, B.; Peltonen, J.; Bobacka, J.; Lewenstam, A. “Influence of morphology and topography on potentiometric response of magnesium and calcium sensitive PEDOT films doped with adenosine triphosphate (ATP),” Analytica Chimica Acta, 2006, 555, 118–127. 66. Li, X.; Li, Y.; Tan, Y.; Yang, C.; Li, Y. “Self-assembly of gold nanoparticles prepared with 3,4ethylenedioxythiophene as reductant,” Journal of Physical Chemistry B, 2004, 108, 5192–5199. 67. J¨onsson, S. K. M.; Salaneck, W. R.; Fahlman, M. “Spectroscopy of ethylenedioxythiophenederived systems: from gas phase to surfaces and interfaces found in organic electronics,” Journal of Electron Spectroscopy and Related Phenomena, 2004, 137–140, 805–809. 68. Nguyen, T. P.; de Vos, S. A. “An investigation into the effect of chemical and thermal treatments on the structural changes of poly(3,4-ethylenedioxythiophene)/polystyrenesulfonate and consequences on its use on indium tin oxide,” Applied Surface Science, 2004, 221, 330–339. 69. Bruno, F. F.; Fossey, S. A.; Nagarajan, S.; Nagarajan, R.; Kumar, J.; Samuelson, L. A. “Biomimetic synthesis of water-soluble conducting copolymers/homopolymers of pyrrole and 3,4-ethylenedioxythiophene,” Biomacromolecules, 2006, 7, 586–589. 70. Damlin, P.; Kvarnstr¨om, C.; Ivaska, A. “Electrochemical synthesis and in situ spectroelectrochemical characterization of poly(3,4-ethylenedioxythiophene) (PEDOT) in room temperature ionic liquids,” Journal of Electroanalytical Chemistry, 2004, 570, 113–122. 71. Sarac, A. S.; S¨onmez, G.; Cebeci, R. C. “Electrochemical synthesis and structure studies of polypyrroles, poly(3,4-ethylenedioxythiophene)s and copolymers of pyrrole and 3,4ethylenedioxythiophene on carbon fibre microelectrodes,” Journal of Applied Electrochemistry, 2003, 33, 295–301.

Morphology of PEDOT

383

72. Kvarnstr¨om, C.; Neugebauer, H.; Blomquist, S.; Ahonen, H. J.; Kankare, J. “In situ FTIR spectroelectrochemical characterization of poly(3,4-ethylenedioxythiophene) films,” Synthetic Metals, 1999, 101, 66. 73. Kvarnstr¨om, C.; Neugebauer, H.; Ivaska, A.; Sariciftci, N. S. “Vibrational signatures of electrochemical p- and n-doping of poly(3,4-ethylenedioxythiophene) films: An in situ attenuated total reflection fourier transform infrared (ATR-FTIR) study,” Journal of Molecular Structure, 2000, 521, 271–277. 74. Mausavi, Z.; Alaviuhkola, T.; Bobacka, J.; Latonen, R.; Pursiainen, J.; Ivaska, A. “Electrochemical characterization of poly(3,4-ethylenedioxythiophene) (pedot) doped with sulfonated thiophenes,” Electrochimica Acta, 2008, 53, 3755–3762. 75. Kvarnstrom, C.; Neugebauer, H.; Blomquist, S.; Ahonen, H. J.; Kankare, J.; Ivaska, A. “In situ electrochemical characterization of poly(3,4-ethylenedioxythiophene),” Electrochimica Acta, 1999, 44, 2739–2750. 76. Tehrani, P.; Kanciurzewska, A.; Crispin, X.; Robinson, N. D.; Fahlman, M.; Berggren, M. “The effect of pH on the electrochemical over-oxidation in PEDOT-PSS films,” Solid State Ionics, 2007, 177, 3521–3527. 77. Qu, L. T.; Shen, G. Q.; Chen, F.; Zhang, J. X. “Electrochemical growth of polypyrrole microcontainers,” Macromolecules, 2003, 36, 1063–1067. 78. Qu, L.; Shi, G.; Yuan, J.; Han, G.; Chen, F. “Preparation of polypyrrole microstructures by direct electrochemical oxidation of pyrrole in aqueous solution of camphorsulfonic acid,” Journal of Electroanalytical Chemistry, 2004, 561, 149–156. 79. M¨ucke, N.; Kreplak, L.; Kirmse, R.; Wedig, T.; Herrmann, H.; Aebi, U.; Langowski, J. “Assessing the flexibility of intermediate filaments by atomic force microscopy,” Journal of Molecular Biology, 2004, 335, 1241–1250. 80. K¨ubel, C.; Gonzalez-Ronda, L.; Drummy, L. F.; Martin, D. C. “Defect-mediated curvature and twisting in polymer crystals,” Journal of Physical Organic Chemistry, 2000, 13, 816–829. 81. Gonzalez-Ronda, L.; Martin, D. C. “Lattice bending in electrooptically-active polynonylbithiazole and polynonylbisoxazole,” Macromolecules, 2004, 37, 2872–2879. 82. Hulvat, J. F.; Stupp, S. I. “Liquid crystal templating of conducting polymers,” Angewandte Chemie International Edition, 2003, 42, 778–781. 83. Hulvat, J. F.; Stupp, S. I. “Anisotropic Properties of Conducting Polymers Prepared by Liquid Crystal Templating”, Advanced materials 2004, 16, 589–592. 84. Yang, J.; Kim, D.; Hendricks, J.; Martin, D. C. “Ordered surfactant-templated poly(3,4ethylenedioxythiophene) (PEDOT) conducting polymer on microfabricated neural probes,” Acta Biomaterialia, 2005, 1, 124–136. 85. Martin, D. C., Shen, Y., Wu, J., Povlich, L., Leach, M., Spanninga, S., and Shaw, C., “Conducting Polymer Cubic Phases,” PMSE Preprints, Proceedings of the American Chemical Society Meeting, Salt Lake City, UT, March 2009. 86. Lenz, R. W.; Go, S. “Crystallization-induced reactions of copolymers III. Ester interchange organization of poly(cis/trans-1, 4-cyclohexylenedimethylene terephthalate),” Journal of Polymer Science Part A: Polymer Chemistry, 1973, 11, 2927. 87. Lenz, R. W.; Go, S. “Crystallization-induced reactions of copolymers. IV. ester-interchange reorganization of poly(ethylene terephthalate-co-2-methylsuccinate),” Journal of Polymer Science Part A: Polymer Chemistry, 1974, 12, 1–10. 88. Kim, B. C.; Spinks, G. M.; Wallace, G. G.; John, R. “Electroformation of conducting polymers in a hydrogel support matrix,” Polymer, 2000, 41, 1783–1790. 89. Ghosh, S.; Inganas, O. “Conducting polymer hydrogels as 3d electrodes: applications for supercapacitors,” Advanced Materials, 1999, 11, 1214–1218. 90. Asberg, P.; Ingan¨as, O. “Hydrogels of a conducting conjugated polymer as 3-D enzyme electrode,” Biosensors and Bioelectronics, 2003, 19, 199–207. 91. Kim, D.; Abidian; M.; Martin, D. C. “Conducting polymers grown in hydrogel scaffolds coated on neural prosthetic devices,” Journal of Biomedical Materials Research, 2004, 71A, 577– 585.

384

D. C. Martin et al.

92. Joseph, M. D.; Ludwig, K. A.; Richardson-Burns, S. M.; Miriani, R.; Kipke, D. R. “Chronic in-vivo neural recording from bioactive hydrogel-coated neural probes,” 38th Neural Interfaces Conference. NIH Neural Interfaces Conference Proceedings, Cleveland, OH, June 2008. 93. Miriani, R.; Abidian, M.; Kipke, D. R. “Cytotoxic analysis of the conducting polymer PEDOT using myocytes,” 30th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2008, 1–8, 1841–1844. 94. Asplund, M.; Thaning, E.; Lundberg, J.; Sandberg-Nordqvist, A. C.; Kostyszyn, B.; Ingan¨as, O.; von Holst, H. “Toxicity evaluation of PEDOT/biomolecular composites intended for neural communication electrodes,” Biomedical Materials, 2009, 4, 045009. 95. Richardson-Burns, S. M.; Hendricks, J. L.; Foster, B.; Povlich, L. K.; Kim, Kim, D.; Martin, D. C. “Polymerization of the conducting polymer poly(3,4-ethylenedioxythiophene) (PEDOT) around living neural cells,” Biomaterials, 2007, 28, 1539–1552. 96. Richardson-Burns, S. M.; Hendricks, J. L.; Martin, D. C. “Electrochemical polymerization of conducting polymers in living neural tissue,” Journal of Neural Engineering, 2007, 4, L6–L13. 97. Corey, J. M.; Gertz, C. G.; Wang, B.-S.; Birrell, L. K.; Johnson, S. L.; Martin, D. C.; Feldman, E. L. “The design of electrospun PLLA nanofiber scaffolds compatible with serum-free growth of primary motor and sensory neurons,” Acta Biomaterialia, 2008, 4, 863–875. 98. Kumar, S.; Adams, W. W. “Electron beam damage in high temperature polymers,” Polymer, 1990, 31, 15–19.

Polymer Reviews, 50:385–409, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.494238

Structural Characterization of Organics Using Manual and Automated Electron Diffraction UTE KOLB, TATIANA E. GORELIK, ENRICO MUGNAIOLI, AND ANDREW STEWART Institut f¨ur Physikalische Chemie, Johannes Gutenberg Universit¨at, Mainz, Germany In the last decade the importance of transmission electron microscopic studies has become increasingly important with respect to the characterization of organic materials, ranging from small organic molecules to polymers and biological macromolecules. This review will focus on the use of transmission electron microscope to perform electron crystallography experiments, detailing the approaches in acquiring electron crystallographic data. The traditional selected area approach and the recently developed method of automated diffraction tomography (ADT) will be discussed with special attention paid to the handling of electron beam sensitive organic materials. Keywords electron diffraction, simulation methods, automated data acquisition, structure determination

1. Introduction Structural elucidation of materials is a key step in the characterization of their physical properties and therefore is of particular interest to both academics and industry. A vast array of materials are nanocrystalline or consist of nanosize domains that either cannot be crystallized as large crystals or exhibit structural features different from those of large crystals. For many of these materials transmission electron microscopy (TEM) is the only technique because it can deliver structural information from areas as small as tens of nanometers. High-resolution transmission electron microscopy (HRTEM), a well-known and powerful method, can directly visualize structural features of the sample. Additionally, electron diffraction (ED) data can be obtained, typically providing much better resolution in reciprocal space1 and therefore intensively used for structure elucidation.2,3 Recently scanning transmission electron microscopy (STEM) gained more significance as an imaging technique, especially for beam sensitive samples.4–7 Despite intensive developments in the field of electron crystallography, it is still a rather “exotic” method of crystal structure analysis. Most structures are solved by single crystal X-ray diffraction method. If a suitable crystal, with a size of about 1 mm3 for lab X-ray sources down to some µm3 for synchrotron data, cannot be grown, usually X-ray powder structure analysis is performed. A powder X-ray profile comprises one-dimensional data and often severely suffers from peak overlap. Indexing of powder X-ray data containing Received January 11, 2010; accepted May 13, 2010. Address correspondence to Ute Kolb, Institut f¨ur Physikalische Chemie, Johannes Gutenberg Universit¨at, Welderweg 11, 55128 Mainz, Germany. E-mail: [email protected]

385

386

U. Kolb et al.

several phases can be very problematic. Strong preferred orientation, often found especially for organic materials, can influence the intensity distribution within a profile and hamper a structure solution. Lastly, structural features like superstructure effects are usually problematic for powder data. All these are the cases when electron crystallography is the best strategy for structure elucidation. Electron crystallography can deliver single-crystal structure information from nanomaterials, and it is particularly efficient for mixtures of different phases, because the crystals can be imaged and at the same time diffraction data can be collected. Electron crystallography of organic materials is especially powerful; for instance, when working with polymorphic systems. There are different ways a molecule can be packed into a crystal. These different crystal packings of a chemically equivalent molecule are called polymorphs. Molecular crystals can form a variety of polymorphs with significantly different physical properties. Even though the molecular structure is the same, the chemical behavior can be different. The polymorphism issue is a very important aspect of pharmaceutical compounds: different crystalline forms can have different solubility and therefore different bioavailability.8 Organic pigments crystallized as different polymorphs show diverse color and fastness properties.9 In general, everywhere an organic material is used in its crystalline form the knowledge of crystal packing is of great importance.10 Different polymorphs present simultaneously in a sample can be distinguished and characterized structurally by electron crystallography. Polymorphism also exists for proteins,11 but because for these materials the main interest is the conformation of the molecule, the arrangement of the molecules inside a crystal is not a primary concern. The most significant problem for organic specimens is the sensitivity to electron beam–induced damage. Beam damage is defined as any structural change caused by the interaction of the electron beam and the specimen and can be as severe as complete degradation of the material. Beam damage is observed in almost all materials (for example, it is a major issue for zeolites12,13 and hydrated minerals14,15) but is most significant for organic samples16 and metal–organic frameworks.17 To overcome problems with beam damage, a special strategy for electron dose distribution during data acquisition must be elaborated. The well-balanced electron dose distribution was the underlying idea for recently developed automated diffraction tomography (ADT) module. The principles of ADT data collection, processing, and use for structure solution will be described in detail in the following sections. In parallel, structure solution from manually collected electron diffraction data will be presented. Throughout this manuscript electron microscopy applications are illustrated with data of 9,9 -bianthracene-10-carbonitrile (CNBA; Figure 1). This material exhibits nonlinear optical (NLO) activity in solution but, due to the centrosymmetric space group it adapts, the NLO activity vanishes in the crystalline state. Initially the structure was solved from manually collected electron diffraction data using maximum entropy methods.18 Subsequently, the structure was confirmed by single crystal X-ray analysis.19

2. TEM Sample Preparation Because the interaction of electrons with matter is much stronger than for X-rays and neutrons,20 the limiting factor for TEM samples is the electron beam transparency. In other words, the samples have to be very thin.21 The electron transparency depends on the elemental composition. For organic materials composed mainly of light atoms, such as carbon, nitrogen, and oxygen, samples thinner than 200 nm are necessary for TEM and electron diffraction analysis; for low-resolution tomographic reconstructions, cell cuts with

Structural Characterization of Organics

387

Figure 1. Molecular structure of CNBA and STEM image of the crystals selected for diffraction data collection. The white square indicates approximately the area of investigation; the tilt axis is oriented horizontally.

a thickness up to 500 nm can still be used.22 In order to perform HRTEM or to collect kinematic diffraction data the thickness should be less than 10 nm.23 The samples are usually deposited on a copper grid coated with a thin amorphous carbon film (2–5 nm). Alternatively Au, Ni, Mo, Ti, Al, steel, graphite, and nylon grids are available. The carbon film is not only holding the material but serves as a platform for heat dissipation and charge transfer. The beam penetrates through the material and the carbon film. The use of a holey carbon support is not recommended for beam-sensitive material because the reduced heat dissipation gives rise to increased beam damage16 (see section 3). All organic molecules can be subdivided into three groups based on their size and function: small organic molecules, polymers, and large biological macromolecular complexes. The classification of molecules into one of these groups is not strict; some materials such as short oligomers of long polymer chains or larger poly-electrolytes easily bridge the different classes. The three classes generally show different structural features, and therefore usually different preparation and measurement approaches are necessary for characterization. For small-molecule crystals generally there are three approaches for sample preparation: crystallization, dispersion, and sublimation. Crystallization from a solution directly onto a grid can produce evenly distributed thin crystals, ideal for TEM investigations. On the other hand, the method has clear limitations: the material has to be soluble in a suitable solvent, which produces nicely defined but small crystals, and neither the solvent nor the material are allowed to react with the metal of the grid. Another factor to be considered is the possibility to produce a polymorph different from the modification of the bulk phase. This is not the case if the sample is dispersed rather than dissolved and then sprayed onto the grid. Poorly soluble materials, which are likely to refuse the crystallization into crystals suitable for X-ray single crystal structure analysis, can be nicely prepared using this methodology. A third alternative sample preparation technique is sublimation directly onto the carbon coated grid or onto a selected crystalline surface: potassium chloride or mica, either precoated with carbon film or not. The crystals grown on a clean surface must

388

U. Kolb et al.

then be covered with a thin carbon film; the film is detached from the crystalline block by floatation on water and transferred on a TEM grid. The sublimation method is more likely to provide larger and flatter crystals; however, depending on the crystalline support different morphology and sometimes different polymorphs can be produced.24 Molecular crystals often have very strong difference in unit cell parameters, causing morphological anisotropy. Often crystals with one long unit cell parameter and two rather short ones appear as platelets, whereas structures with two relatively long axes tend to appear needle shaped.25–27 Crystals with strongly anisotropic morphology tend to lie on a support film always in the same way. As a result, a strong preferred orientation of the material is observed. In such situations, observation of certain directions can be problematic due to the missing cone problem. Preferred orientation can be avoided by adjusting crystallization parameters.28 Sometimes the employment of epitaxial growth can force the material to change morphology. However, influencing the crystal morphology one could easily run into conditions where a new polymorph is produced.10 Alternatively, the sample can be embedded in an appropriate medium (epoxy resin, ice, etc.) and sliced by a microtome. The thin slices are then placed onto a coated TEM grid. The embedding resin can be removed by an appropriately matched solvent. For long-chain polymer sample preparation usually two strategies are followed: cutting thin slices out of bulk or preparing thin films on a suitable support. Recently, an extended review was published summarizing sample preparation procedures for TEM imaging.29 Polymer films can be extra-ordered during the film preparation, either by mechanical sliding over a glass surface, spin-coating, or epitaxial growth on a template. Relatively short oligomers can be crystallized from dilute solution using a similar methodology to the small-molecule crystals.30 Another possibility is to place a drop of a polymer on the surface of a nonsolvent liquid and to collect the resulting thin film with a TEM grid subsequently. Polymers also show polymorphism; for instance, polymorphs were found for poly(1-butene) or polylactides which crystallized epitaxially.31–33 Large protein macromolecular complexes are flexible and therefore generally difficult to crystallize.34 Considerable time and effort are spent on the crystallization of proteins for X-ray single-crystal analysis. In some cases during the crystallization attempts small crystals are produced with a size not sufficient for X-ray analysis but useable for electron diffraction data collection.35 Electron diffraction of three-dimensional (3D) protein nanocrystals is a relatively new and actively emerging area.36 A large number of proteins, due to their nature,37 crystallize as two-dimensional (2D) crystals and are therefore good candidates for electron crystallography.38 Because 2D or 3D protein crystals are usually only stable within their water-based mother liquor, TEM investigations are typically done using vitrified samples. The suspension is quenched into liquid ethane, propane, or a mixture of both. It is actively discussed which freezing medium is optimal,39 but it is generally accepted that the vitrification procedure is more specimen and operator dependent than hardware specific, so there is no unique procedure. Vitrification not only preserves the material in the natural state but reduces radiation damage.

3. Electron Beam Damage Beam damage is the limiting factor of TEM measurements for organic materials. The electron beam irradiation often leads to structural changes or even to a complete decomposition of the material. The mechanism of the damage has been investigated extensively but still is not fully understood.

Structural Characterization of Organics

389

No matter which mechanism is responsible for the material decay, the damage level depends on the amount of energy deposited in a given volume and the rate of dissipation away from this volume.16 The amount of energy received during irradiation is proportional to the number of incident electrons per unit area. The incident charge density is also called electron dose q and is usually measured in [C cm−2] or [¯e Å−2] with a conversion factor given in Eq. (1). Different concepts of electron dose are used to quantify the damaging process29,40: critical or characteristic dose defined as the point at which the electron diffraction intensities are reduced to 1/e (corresponding to 37% reduction of the initial intensity value) of its initial value, and total endpoint dose at which the crystallinity is completely lost or no further change in intensity is observed. Thus, copper phthalocyanine crystals, which are completely destroyed at a total endpoint electron dose of q = 0.4 C cm−2 = 250 e¯ Å−2,16 will degrade after 25 s of irradiation with the electron beam conditions set at 10 e¯ Å−2 s−1 (experimental conditions are usually described by electron dose rate measured in [¯e Å−2 s−1]). q[C · cm−2 ] =

1016 [¯e · A−2 ] = 625[¯e · A−2 ] 1.6022 · 10−19

(1)

The energy dissipation rate can be difficult to estimate but there are some basic rules. Typically organic crystals appear more stable under the electron beam if the crystal is partially illuminated, so the irradiation induced heating can be transported to the adjacent parts of the crystal. Crystal thickness has a similar effect—thicker crystals are more stable. A large crystal area contact with the supporting film is of crucial importance: crystals with minimal surface contact degrade more rapidly under irradiation. Techniques for increasing the crystal stability are the use of an additional carbon layer on top of the specimen41,42—this increases charge transfer and heat dissipation and therefore the stability of the material—or insertion of an objective aperture in the vicinity of the sample.43 The damage of material occurs due to inelastic scattering, which deposits energy into the material. The deposited energy manifests itself in two ways: one, heating of the specimen (molecular vibrations) or two, excitation of individual molecules, which eventually leads to bond cleavage (ionization). Knock-on damage is usually not considered for organic materials because its contribution to the damage is negligible compared to the other mechanisms. The heat transfer effects can be significantly reduced by sample cooling, either to liquid nitrogen or even down to liquid helium temperatures.44 Cooling reduces molecular motion, keeping the relative position of radicals generated by electron beam and increasing the likelihood for the bond reformation. Radicals created after the primary ionization step usually react further and either produce other longer-living radicals or cross-linked aggregates. These processes are in a dynamic equilibrium and the final damage mechanism may be more dependent on the stability of the intermediates or the final construction than on the dissociation energy in the primary bond cleavage.16 Collective damage to single molecules causes deterioration of periodicity in the crystal structure. The intensities of reflections generally fade as the degree of crystallinity is decreasing. The critical dose, defined as a dose at which the peak intensities of diffraction spots decrease to 1/e (37%) of the initial value can be measured from the diffraction intensities fading curve. Figure 2 shows reflection intensity fall off for a CNBA crystal. A series of electron diffraction patterns was collected sequentially from the same spot on the crystal with 1s exposures. The reflections within the pattern are sorted according to the d-value they represent. The solid black circles show the integrated intensity of the reflections associated

390

U. Kolb et al.

Figure 2. Reflection intensity fall-off due to electron irradiation for CNBA sorted by d-spacing. The plot was drawn on the basis of data recorded in nano-electron beam conditions with a beam size of 100 nm, at an electron dose rate of ∼5 e¯ Å−2 s−1. The 37% decay of the periodicity is reached after 65 s of illumination. Therefore, the characteristic dose of CNBA is around 325 e¯ Å−2.

to interplanar distances (d) larger than 4 Å. These reflections represent long axes in the structure and account for the periodicity of the structure; in other words, for the degree of crystallinity. This curve shows a well-defined plateau typically observed for long-d reflections.45 The hollow circles comprise the integrated intensity of the reflections with d less than 2 Å. These far-outlying high-resolution reflections represent the resolution limit of the diffraction data. The curve falls off exponentially. The different behavior of low- and high-resolution reflections is typically observed in organic crystals.46 It demonstrates the general degradation of the data resolution due to irradiation. The middle grey curve shows the reflections within the range between 2 and 4 Å. These reflections show a rather complicated behavior: the curve falls off but there is a significant redistribution of intensities after approximately 1 min of irradiation. This effect reflects rearrangement of the molecules into a more stable intermediate structure before the crystal collapses completely. This happens when the overall periodicity is significantly disturbed (fall-off of the black curve) and the molecules become flexible. This kind of rearrangement typically occurs for flat molecules that are able to pack into π -stacks. It was demonstrated that for some systems the dose rate rather than the total electron dose describes the stability of the material.47,48 Because the irradiation damage is a multistage process, the particular combination of all factors determines the effective deterioration rate. This may change if, for instance, the effective contribution of some healing effects increases. Sometimes the critical electron dose is correlated with the melting or degradation temperature of the material.40 The melting is associated with the flexibility and mobility of a molecule, whereas irradiation damage is generally connected with radical creation with subsequent cross-linking. Significant deviations from the melting temperature/characteristic dose correlation are observed for conjugated systems. Typically extended aromatic moieties have higher stability, likely connected with the stability of the ionization/dissociation

Structural Characterization of Organics

391

products. Close derivatives of the same molecule can show strikingly different irradiation stability as, for instance, a phthalocyanine molecule being much less stable than its halogenated derivates.49 Material degradation is also dependent upon the energy of the incident electron beam determined by the microscope setup. The electron beam energy is one of the main parameters defining the interaction cross section; therefore, it is natural to expect different damage behavior for different electron beam energies.21 Depending on the exact damage mechanism for a selected material, the damage can be reduced, changing the beam energy. Thus, there are two general ways to improve the stability of a specimen: reducing the electron dose in terms of reducing the number of electrons falling onto a certain area and adjusting the microscope parameters in order to influence scattering (accelerating voltage) or heat transfer (specimen cooling) processes.

4. Adaptation of TEM to Organic Materials 4.1. TEM Imaging Techniques In a transmission electron microscope, data are always collected in projection; therefore, the information along the vertical direction is missing. In order to gain information in the third dimension the specimen must be tilted and additional data collected. This procedure is repeated over a series of images and typically spans a high tilt range. The limiting factors of the tilt range are usually the physical geometric constraints of the apparatus, such as the pole piece gap, the specimen holder, or specimen grid bar, although interference from other parts of the specimen can also be a limiting factor. This unobtainable region of data is known as the missing cone. It is a well-known problem, leading to anisotropic artefacts in the real space tomogram50 and poor coverage of reciprocal space in diffraction data sets, which are more likely to fail during direct methods structure solution attempts.51 Therefore, collection of the maximal possible 3D data is essential for further processing. Because electrons are charged particles, their scattering is caused by the interaction with the effective electrostatic field of the specimen, which is a combination of the atomic nuclei and electron orbitals. An electron traversing through material can be scattered, preserving its original energy (elastic scattering), or losing part of its original energy (inelastic scattering), or with no interaction. Elastically scattered electrons contribute to imaging and diffraction formation. Inelastically scattered electrons carry information about the atomic composition of the material and therefore can be used for spectroscopic analysis. The scattering cross section of electrons characterizing the probability for a scattering event is several orders of magnitude higher than that for X-rays. Therefore, much smaller volumes of material are necessary in order to collect electron scattering information with sufficient signal-to-noise ratio, making TEM a useful tool for structural characterization of nanosized objects. Conversely, the strong interaction of electrons with matter has some negative aspects too, such as multiple scattering of the electron beam as it passes through the specimen.52 Multiple scattering severely affects image and diffraction intensity data, making the interpretation difficult because factors such as defocus, sample thickness, and composition must be accounted for during the data analysis. The kinematical approximation that describes the observed data after a single scattering event (as assumed in X-ray crystallography) in principle cannot be applied to electron data.53,54 Two-beam conditions can be used as a first approximation to describe image and diffraction formation during electron data analysis.55–58 For more accurate analysis of the data a multislice approach can be used and

392

U. Kolb et al.

usually allows matching the experimental data well, if the necessary input parameters are available.59 Nevertheless, the use of multislice simulations for diffraction data is rather limited.60 Fortunately, dynamical scattering effects are less important for organic samples than for inorganic. The application of TEM methods to organic materials primarily consists of reducing the electron beam dose during data acquisition. Organic crystals tend to lose their crystallinity under electron bombardment very fast, making direct high-resolution imaging challenging29 and often impossible. Under the strong irradiation needed in HRTEM conditions the molecular packing is usually destroyed before the image has been recorded. Therefore, one should be very careful with the interpretation of HRTEM images of molecular crystals. Nevertheless, there are examples of crystals that are relatively stable and can be imaged in TEM.61–63 Low-dose imaging was developed specially for beam-sensitive materials.64,65 In this method, focusing and all necessary alignments are undertaken on a testing area close to the area of interest, and then the image is taken as a single exposure at an adjacent part of the sample. This method allows collecting data from fresh areas of the sample that were not illuminated prior to the data collection, thus minimizing the total dose on the area of interest. Another approach to image beam sensitive materials is scanning transmission electron microscopy (STEM). In this mode the data are collected without simultaneously illuminating the whole area but by fast scanning of a focused beam across the area. The signal is collected by a sensitive high angle annular dark-field detector (HAADF). Because the efficiency of the HAADF detectors is very high, in principle, electron dose can be significantly reduced, still allowing collection of data with sufficient signal-to-noise ratio. Because each point of the area is illuminated only for a short time during the image recording, the total electron dose and therefore the beam damage to the specimen are minimized. It was shown that decreasing the acceleration voltage and therefore the electron energy can enhance the stability of material. Usually each TEM has a certain acceleration voltage at which it has the optimal working stability, but the usage of lower voltages is in principle possible. Modern machines offer a broad range of available voltage settings. Objective lens aberration–corrected TEM can produce a low-voltage coherent beam and record high-quality high-resolution images at 60–80 kV (SALVE project66). The aberrations of the objective lens have less effect on diffraction patterns; therefore, principally there is no limitation of collecting diffraction data at low voltage without an image corrector. These investigations are active areas of research.67 In principle, the coherency of the nanobeam used for electron diffraction can be increased by applying a probe corrector (STEM corrector), and therefore the geometry of the diffraction event can be optimized.68 In order to resolve fine details of a structure, high-magnification imaging must be applied. Recording of the image with a reasonable signal-to-noise ratio requires quite long exposure times. This combination unavoidably leads to damage of the sample. In contrast, the use of diffraction mode does not require high magnification. Additionally, the resolution of the diffraction data goes significantly further than that of high-resolution images (which is strongly reduced due to the imperfections of magnetic lens). 4.2. Electron Diffraction Techniques Electron diffraction was discovered by Davisson and Germer in 1927.69 The first structural investigations of organic material based on electron diffraction data recorded in transmission

Structural Characterization of Organics

393

were performed on polymethylene chain compounds. The first powder electron diffraction data of these compounds was recorded in 193170; 2 years later single-crystal electron diffraction patterns of polymers were published.71,72 The first structure analysis of paraffin from single-zone electron diffraction data based on a match between calculated and experimental intensities appeared in 1936.73 The second report on a structure solution (retene) appeared a few years later74 where the molecular packing was pre-optimized geometrically as well as by visual inspection of the diffraction intensities and the obtained structural model was refined by comparison of experimental and calculated intensities. Electron diffraction was collected in these cases as powder patterns in order to obtain cell parameters and as single diffraction patterns in order to obtain intensities. Later, difficulties in orienting single diffraction patterns precisely and drawing 3D intensities useable for structure solution were avoided using texture patterns. This approach started a large series of crystallographic work based on electron diffraction data75 and a variety of structures, for example, diketopiperazine,76 urotropine and urea,77 two polymorphs of thiourea,78,79 and copper salts of some aminoacids,80–83 have been successfully solved. From the 1960s structures of various polymeric chains solved from electron diffraction data appeared in the literature.84–86 With the development of appropriate sample holders the collection of 3D data by a sequential tilt of the sample became possible and has been used for structure solution by direct methods of organic compounds like C60 Buckminsterfullerene87 and nonlinear optically active materials.88–90 Electron crystallography was also very effective in the field of biological objects: the structure of purple membrane obtained from electron diffraction data was reported in 1975.91 Later structures of bacteriorhodopsin,92 plant light-harvesting protein,93 and tubulin94 were published. A number of structural studies from electron diffraction data were published recently.95–102 Electron diffraction data traditionally can be collected in two different ways: via insertion of a selected area aperture (selected area electron diffraction, SAED) in the back focal plane of the objective lens or by selection of a small condenser aperture (10 µm) to generate a parallel beam with small diameter on the sample (nanodiffraction mode). In SAED mode a large area of the sample is unavoidably illuminated, and the diffraction information is only collected from the part inside the selected area aperture. Therefore, this mode is not optimal for beam sensitive materials. In nanodiffraction mode the information is collected only from the area directly illuminated by the beam. The size of this area can be adjusted by focusing the condenser lens and is usually in the range of 50–100 nm. With this approach it is possible to collect up to 100 diffraction patterns until the equivalent SAED area has been fully illuminated. To achieve beam sizes lower than 150 nm with a 10-µm aperture the beam needs to be slightly convergent and diffraction spots spread into small disks, which can then be focused by the diffraction lens. This causes in turn a rotation of the diffraction pattern and changes the effective camera length. These parameters, however, can be easily recalled from the value of the diffraction lens excitement and cannot be considered as drawbacks of the data. Electron diffraction in high-convergence mode (convergent-beam electron diffraction, CBED) recently showed its potential for ab initio structure solution of materials.103 However, this technique has almost no application to organic materials because high convergence of the beam introduces a high electron dose at the sample that the materials cannot tolerate. Furthermore, due to the long crystallographic axis often present in organic crystals, overlap of the diffraction disks would limit the useful information in CBED.

394

U. Kolb et al.

4.3. Recording Media Photographic films read out with high-performance scanning units are still actively used where a large field of view is required together with a relatively high magnification. Modern TEMs are usually equipped with charge-coupled device (CCD) cameras providing reasonable linearity of the signal, high dynamic range, and good sensitivity. Recent developments of CCD cameras reach the field of view of a photographic film or imaging plate. The signal from organic materials using a low-intensity beam is usually very weak. Conversely, the increase of the exposure time only damages the material instead of improving the signalto-noise ratio; therefore, the high sensitivity of the recording media is a primary concern when working with organics. Image plates show characteristics superior to those of CCD cameras but are more complicated in handling and not often used. Special high-sensitivity detectors were developed recently.104 The data are usually available in a digital form or are converted into a digital form for further processing. TEM images, especially HRTEM, usually do not have large contrast ranges and can be stored in 8-bit format. For the diffraction data it is extremely important to use the largest available dynamical range in order to avoid truncating the strongest reflections. Therefore, the diffraction data are typically stored in 16-bit or higher format. Different formats are used for the data storage: some are standard image formats like tif; some are specific to the CCD camera control software (for example, dm3 of Digital Micrograph GATAN, Pleasanton, CA, USA). Regardless of the format, it is important to have full, unaltered data and comprehensive information about the acquisition parameters for further processing. For beam-sensitive material it is crucial to select the appropriate exposure time. If the recording time is too long, the material degrades under the beam too quickly or in the extreme case will deliver a diffraction pattern already distorted by diffraction from a damaged area. If the exposure time is too short, the signal-to-noise ratio is too small and the reflections are lost in the background.

5. Solving Crystal Structures The aim of electron crystallography is to obtain atomic coordinates of a crystalline material. A well-proven methodology of structure solution from Bragg diffraction data can be adopted from single-crystal X-ray structure analysis and applied in electron crystallography with minor adjustments: 1. Collection of diffraction data: this is the most problematic step for beam-sensitive materials. Special techniques for optimizing data collection are discussed in section 6. 2. Determination of unit cell parameters and the orientation matrix. 3. Determination of the space group or a range of possible space groups consistent with extinction conditions and symmetry of intensities. 4. Structure solution: at this point a sensible structural model including most of the atoms should be created. 5. Structure refinement and validation of the structure model: missing atoms are found to complete the structure and the atomic positions are refined to satisfy proper bond geometry. For structure solution the methods commonly used in X-ray analysis can be applied, if the data collected are pusedo-kinematic in nature. When the kinematic scattering formalism

Structural Characterization of Organics

395

is applicable, the most significant differences between X-rays and electrons are in the wavelength of the radiation and in the atomic scattering factors.105,106 Some programs, such as SIR,107 SHELX,108,109 and the maximum entropy package MICE110 have been adopted to work with electron diffraction data sets. A recently developed charge-flipping algorithm111 implemented in Superflip112 can deal with electron diffraction data sets and already delivered promising results; however, the use of electron diffraction data is not included in the publically available version yet. A combination of diffraction data with energy potential for structure solution implemented in Endeavour was recently adapted to electrons as well.113 There is ongoing work on linking electron scattering factors into TOPAS,114 FOX,115 and DASH.116

6. Electron Diffraction Data Collection and Processing Omitting an exotic case of one-dimensional molecular ordering,117 there are two types of organic molecular arrangement spoken of in crystallography: 2D and 3D. Two-dimensional crystals do not have any repetition units in the third direction (in contrast to disordered structures, which do possess the third direction but are not packed periodically along this axis). Diffraction space of a 2D crystal consists of relrods going along the direction with no repetition units. The relrods are not homogeneous; they show intensity modulations due to the structure variation along the third direction. Typically even for 2D crystals the third direction is of interest, and therefore data are also collected at higher tilts. Sometimes, even for 3D crystals the structure can be fully solved from 2D data (if the repetition length in the third direction in real space is very short, and the atomic arrangement is obvious due to rigid molecular conformations as it can be found; e.g., for π -stacked molecules). Such a data set would only include one crystallographic zone. Normally, to solve a 3D crystal structure a 3D data set is necessary. 6.1. Proteins: 2D Electron Crystallography Membrane proteins play the most crucial role in the transport system of an organism. Therefore, these are primary targets for the drug delivery study. Membrane protein crystals tend to organize into 2D networks bound by lipids. Electron crystallography allowed structure determination of various membrane proteins.38 Typically the resolution of models received from electron crystallography is lower than that from X-ray crystallography, but the models have a clear advantage: the protein is embedded in its natural environment. Usually for 2D crystals electron diffraction data are combined with phase information extracted from HRTEM imaging. The Fourier transform of a high-resolution image provides information about the structure phase of the material. However, because the information is only obtained after traversing through the imaging system, it is modified by lens imperfections (contrast transfer function, CTF). In this respect, the amplitudes obtained from electron diffraction data are superior, because they reach higher resolution and are not modified by the imaging system. Therefore, the diffraction amplitudes are usually used as the basis for structure solution. However, because proteins commonly crystallize in noncentrosymmetric space groups,118 the structural phase is not restricted to 0 and π , and phase determination is a crucial issue. Because there is no other source of phase information available, the structure phase extracted from high-resolution images is used. The phase is combined with the diffraction amplitude and after phase extension an inverse Fourier procedure is performed, delivering a structure model in the form of electron density distribution. Experimental data are usually collected at 0 degree tilt and at some higher

396

U. Kolb et al.

tilts in order to resolve the electron density function in the vertical direction. Typically, due to the missing data, the structural features along the vertical direction are smeared out. The quality of a model can be calculated in different ways—differential phase residual, Fourier shell correlation, and the Q-factor43—but basically it evaluates the match between the determined model and the experimental data. During extensive attempts to produce 3D protein crystals large enough for singlecrystal X-ray structure solution, some small 3D nanocrystals are produced. The use of these crystals is an active area of research for electron crystallographers. 6.2. Structural Analysis from Zonal Electron Diffraction Patterns 6.2.1. Data collection. In a typical electron diffraction experiment, oriented low-index crystallographic zones are collected for further processing. These are normally diffraction patterns with a relatively high density of reflections and strong symmetry imprint. Usually, a suitable single crystal is selected in imaging mode. Then the microscope is switched into diffraction mode and one of the main crystallographic axes of the crystal is oriented along the tilt axis of the goniometer. The crystal is subsequently tilted and prominent diffraction patterns are recorded. Together with the diffraction patterns, the reciprocal angular relationship is stored. Practically there are two approaches to the manual data collection: use of a double tilt holder or a rotation tilt holder. A double tilt holder, which is useful for material with good recognizable zones, allows free rotation around the goniometer axis (α, primary tilt axis) and perpendicular to it. In this way all zones appearing within the reachable wedge are accessible. In the rotation tilt holder, which is recommended for material where only few diffraction spots can be detected, prior to the tilting the crystal has to be oriented with its low-index crystallographic axis along the goniometer axis. One of these approaches is normally selected based on the hardware available in the lab or on the preference of the operator. A hybrid double tilt rotational holder is also available (GATAN, Pleasanton, CA, USA) providing an additional degree of freedom. 6.2.2. Unit cell parameter determination. A set of 3D diffraction patterns is reconstructed into a 3D lattice using the angular relationships between the collected zones. From this information all unit cell parameters can be deduced. The procedure includes basic 3D geometry. Knowing the relative tilt ranges between the zones a Vainshtein plot75 can be constructed, which is described in detail below. Software packages are also available to image the 3D network (TRICE,119 numerous homemade software packages). The metric of the unit cell usually is the basis for the crystal system determination. The true symmetry and space group are additionally confirmed by evaluation of the intensity distribution in the recorded diffraction patterns. Below the lattice parameter determination for CNBA (Figure 1) is presented. A manual diffraction tilt series was collected through a tilt around the monoclinic axis b∗ (Figure 3). The starting zone has an index of [100] with the monoclinic axis b∗ horizontal and the axis c∗ vertical. From this pattern the reciprocal lattice parameters b∗ , c∗ , and α ∗ can be measured directly. For the determination of other lattice parameters, the crystal was tilted around the b∗ axis until symmetrical patterns were obtained. In this way zones [401], [301], [201], and [101] were recorded. A Vainshtein plot is essentially a combination of (h0l) lines extracted from each zone plotted according to the angular distances between the zones. Thus, the first zone [100] is plotted as a row of spots along the c∗ axis. From the zone pattern it is evident that there are extinctions along the [001] axis. These are marked as hollow spots in the Vainshtein plot. The next zone reached was [401], which added two spots to the

Structural Characterization of Organics

397

Figure 3. Manual tilt series of CNBA around the monoclinic axis b∗ (horizontal axis in all diffraction patterns) and Vainshtein plot resolving the third lattice parameter a∗ and the monoclinic angle β ∗ .

Vainshtein plot. The next zone was [301], which has the first reflection along the (h0l) line extincted (therefore marked as a hollow circle). The same situation is observed for zone [101]: each odd reflection is extincted. When all available reflections are sketched, a 2D net of reflections can be drawn through them. This net is the [010] zone, not directly observed but reconstructed. The missing lattice parameters a∗ and β ∗ can be measured directly from the plot. Therefore, the complete lattice cell metric can be calculated (see Table 1). 6.2.3. Symmetry determination. The symmetry of the structure can also be deduced from the analysis of the diffraction patterns. All diffraction patterns collected in the CNBA tilt series have mm symmetry. This means that the structure in the projection has either 2, m, or 2/m symmetry. This allows an assignment of the tilt axis as unique monoclinic axis b∗ . Along this axis each odd reflection is missing satisfying the reflection conditions 0k0: k = 2n due to a 21 screw axis. Besides, from the Vainshtein plot it is evident that full rows of reflections are extincted, the observed reflection following the rule h0l: l = 2n, due to a c-type glide plane. Therefore, the space group of the structure can be unambiguously determined as P21 /c. Knowing the cell metric and the space group, all available diffraction patterns can be indexed and intensities of reflections can be integrated. The approach to the reflection Table 1 Lattice parameters of CNBA determined by manual and automated electron diffraction together with those derived from single crystal X-ray diffraction18

Manual tilt ADT XRD ∗

a, Å

b, Å

c, Å

α

β

γ

14.70 14.75 14.70

9.47 9.48 9.47

15.42 15.36 15.42

90◦∗ 89.4◦ 90◦∗

111.8◦ 112.6◦ 112◦

90◦∗ 90.5◦ 90◦∗

Not measured—set to 90◦ based on symmetry considerations.

398

U. Kolb et al.

integration is more or less standardized, and CRISP/ELD (Calidris120) is the most common software package used for extracting intensities of diffraction spots from zonal patterns in order to produce a standard crystallographic data set (h k l Int). There is no strict definition for the determination of intensity error sigma; therefore, the value was estimated to be 10% of the determined intensities.121 Merging of the data coming from different zones is a serious problem: often zones are collected from different crystals, which may have had different thickness, and therefore a scaling factor would be necessary to combine the data from different patterns together. Usually there are some common reflections present in different zones, so based on their intensities a scaling factor can be calculated. It is a known effect that the intensities along a row of reflections may appear different for the same row present in different zone patterns. This effect is associated with multiple scattering paths, which are realized differently when another set of reflections is excited.30 Therefore, even this intuitive approach is basically wrong but commonly used. Nevertheless, a comprehensive data set allowing a structure solution can be extracted from a manual tilt series.18 6.2.4. Structure solution. Although dynamical effects modulating the electron intensities are a more serious issue for stronger scatters such as inorganic material, they may also affect intensities for organics. The effect is nevertheless usually not strong and often ignored. The biggest problem of data sets extracted from the manually collected electron diffraction tilt series is the coverage of the data. Typically only a few relatively low-index zones are obtained, which cover a high number of symmetrically dependent reflections, higher index zones being difficult to recognize, align, and index. Besides, the patterns may have a different degree of structure decomposition or even features of another packing produced at early stages of irradiation. Additionally, a slight misorientation of a zonal pattern may change intensities severely. Despite all these problems, a number of structure solutions for organic materials from electron diffraction data have been published. Usually the data sets extracted from a set of low-index diffraction patterns do not have good enough statistics to allow automatic application of direct methods. There are nevertheless different sources of additional information that help to achieve structure solution from zonal electron diffraction data: phases of reflections extracted from HRTEM images, employment of energetical constraints, or combination with other diffraction data such as X-ray powder diffraction. An intensity data set obtained by the described traditional method for CNBA contained 193 reflections. The Rsym (determined by SIR) is of 21.20%. The data resolution is 1.4 Å, and the coverage of the reciprocal space is only 25%. These data alone are definitely not enough for an ab initio structure solution. In this case additional information is necessary. The structure of CNBA was pre-optimized by semi-empiric calculations.18 For CNBA, a rather rigid molecule, the only free torsion angle is situated between the anthracene moieties and was optimized to 90 degrees. This molecule was manually positioned into the determined unit cell and the agreement of experimental and theoretical kinematic diffraction data was optimized by shift, rotation, and torsion angle changes in a manual way. The resulting structure showed a torsion angle of 96 degrees and reached an R value of 35%. A subsequently performed single-crystal X-ray structure solution delivered a torsion angle of 95.4 degrees and C-C bond length of 1.405 Å. 6.3. Off-Zone Data Collection: Automated Diffraction Tomography A completely different approach is realized in an automated electron diffraction experiment. Instead of tilting around a low-index crystallographic axis with relatively large gaps between

Structural Characterization of Organics

399

Figure 4. Selection of electron diffraction patterns in a range of 9 degrees taken from an automatically performed tilt about an arbitrary axis of a CNBA single crystal; all diffraction patterns show arbitrary cuts; at −54 degrees the reciprocal space is accidentally cut close to [02–1] zone. The tilt axis position in the diffraction patterns is marked by a dotted line.

the zones, the crystal is tilted in small, fixed tilt steps around an arbitrary axis. This allows fine sampling of a large reciprocal volume containing the diffraction information. The patterns are arbitrary slices through the reciprocal space and usually do not show any high symmetry; only by chance low-index zones are recorded. During the ADT data acquisition the crystal is imaged in STEM mode. Simple calculations show that for the beam settings used in the ADT acquisition module, due to the scanning of the beam instead of large-area illumination, the electron dose received by a sample during STEM image acquisition is at least two orders of magnitude lower than during TEM imaging or electron diffraction recording.122 A suitable crystal is selected in STEM mode at the starting tilt position. Then diffraction patterns are recorded at different tilt positions sequentially. Optionally crystal tracking during the tilt can be used. The tracking procedure cross-correlates STEM images taken at neighboring tilt steps and takes account for the shift of the crystal. The application of ADT data for ab initio structure solution of CNBA is demonstrated in the following sections. Selected ADT diffraction patterns for CNBA are shown in Figure 4.

400

U. Kolb et al.

6.3.1. Reciprocal space reconstruction, unit cell parameters determination, reflections indexation. Data processing for unit cell parameter determination and reconstruction of the 3D volume, including background correction, correction of diffraction pattern shift, and refinement of the tilt axis position, was performed by Matlab routines described elsewhere in detail.122,123 Figure 5 shows views of the reciprocal space along the main directions of the 3D reciprocal net together with the corresponding kinematically simulated electron diffraction zonal patterns. There is a principal difference between single zones and the visualization of 3D ADT data provided in Figure 5. Zones comprise “cuts” through the reciprocal space only showing reflections on this cut, whereas 3D ADT data are presented here as projections of the reciprocal space along rows orthogonal to the view direction, producing an overlapping of reflections obscuring small intensities or extinctions by larger reflections. In the projection [100] a row of extinctions along the vertical direction can be observed due to a c-type glide plane placed perpendicular to the b-axis. The unit cell parameters, which are determined automatically by cluster analysis of the difference vector space, are provided in Table 1. From the deduced orientation matrix it turned out that the tilt series was collected almost around the [6-1-2]∗ reciprocal vector

Figure 5. Reconstructed 3D ADT data of CNBA (projection of the reciprocal space) and corresponding kinematically simulated electron diffraction patterns (cut through the reciprocal space).

Structural Characterization of Organics

401

lying only 0.8 degrees apart. For reflections sampled through several neighboring slices the maximal intensity value was used, assuming that this was a cut closest to the center of the reflection. The data were collected in a tilt range of 120 degrees (±60 degrees), delivering a coverage of two thirds of the complete reciprocal volume. A resolution of 0.8 Å was recorded with 9415 reflections being measured which reduced to 3519 independent reflections after merging symmetry equivalents, with an Rmerge of 21.77%. A total of 87% of all possible reflections within this resolution range were recorded of 0.8 Å 9,415 reflections were collected in total being reduced to 3,519 independent reflections after merging symmetry equivalent pairs covering 87% of the reflections possible for the achieved resolution. The Rsym of the data set was 21.77%. 6.3.2. Symmetry determination. Integral extinctions caused by centering of the lattice are clearly visible in the volume and can be identified. Zonal extinctions like glide planes are easily seen in the reconstructed diffraction volume when viewed in an appropriate direction. Serial extinctions due to screw axes running along a single axis are difficult to observe in a 3D reconstruction because they are covered by neighboring reflections. Procedures to cut slices from the 3D volume are under development. The projections along the main axes, given in Figure 5, clearly show two 90 degree angles and an angle of 112 degrees. Symmetry inspections in order to find mirror planes in the projected patterns have to be handled with care because the patterns depend on the orientation of the tilt axis. Nevertheless, a monoclinic metric is evident including a c-type glide plane. Using the collected reflections in SIR2008 package107 the space group could be determined automatically to be P21 /c, leading to a density of d = 1.21 g/cm3 with four molecules in the unit cell. 6.3.3. Structure solution and refinement. The structure was solved ab initio using direct methods implemented in SIR2008. The isotropic thermal factor determined from the Wilson plot U was 0.013 Å2. The best solution delivered the R factor of 24.43%. The structure solution as it came out of SIR2008 is shown in Figure 6. No ghost atoms were detected in the structure and the first 30 positions of the solution could be assigned to the 30 atoms of the molecule. In comparison to the previously conducted single-crystal X-ray analysis of CNBA18 the average deviation of the atom positions was 0.186 Å, with the maximal deviation of 0.524 Å. These deviations are within the range found for other structure solutions performed with ADT.124–126 The average bond length was determined as 1.39(2) Å. As shown in Figure 1, the molecule has a rather rigid conformation. The only degree of conformational freedom is the torsion angle between the two anthracene moieties. As shown in Figure 6, the anthracene fragments were found basically in a flat geometry, but additional refinement of the geometry was clearly necessary. Structure refinement was performed in SHELXS.108,109 It turned out to be stable without any restraints but did not deliver a significant improvement in molecular geometry. In order to improve the geometry, soft restraints on C-C bond lengths of 1.4 Å with a standard deviation 0.05 Å and later apply strong constraints to keeping the anthracene moieties flat. The refinement delivered, as expected, a flat geometry with a torsion angle between the anthracene moieties of 93.8 degrees. The R value increased to 39%. As a last step and in order to verify the results, geometry optimization was performed using forcefield calculations with the recently available COMPASS forcefield as implemented in Material Studio package (Accelrys, Inc. San Diego, CA, USA). The torsion angle was found to be 94.0 degrees. In principle, the correct refinement procedure should be performed based on dynamical calculations. Unfortunately, the only existing program package for dynamical refinement

402

U. Kolb et al.

Figure 6. Structure solution of CNBA using automatically collected electron diffraction data: (a) as derived ab initio from SIR2008; (b) as refined with soft restraints in SHELX; (c) as refined with constraints in SHELX; (d) as optimized using COMPASS forcefield.

MSLS60 is restricted to zonal diffraction data because it is designed for the traditional approach. Understanding and description of the reduced but still present dynamical effects in nonzonal diffraction patterns is crucial for refinement and structure validation based on ADT data. The adaption of dynamical refinement procedures to ADT data is under development. Diffraction data used for this investigation were collected from the crystal shown in Figure 1 from the area marked by a white square. The crystal is not nicely faceted and is relatively thick. These are original crystals prepared during the synthesis, not specially recrystallized for electron diffraction data collection, and are expected to deliver slightly dynamically distorted intensities. The structure solution step was not hampered by this quick and easy preparation approach and in only 1.5 days the whole structure was determined. In order to obtain optimum data and, based on that a significantly improved molecular geometry, the data acquisition has to be performed with more care. The direct crystallization of the material onto the grid or an epitaxial growth is in progress as well as the use of electron beam precession (PED)127 in combination with ADT. The latter is expected to deliver only a slight improvement, allowing an integration of the reflections over the used tilt step.

7. Conclusion Electron diffraction is a well-known and extremely powerful method to gain structural information from beam-sensitive materials. Although the instability of the material under

Structural Characterization of Organics

403

irradiation means that extra care has to be taken, high-resolution TEM images as well as electron diffraction data can be collected by optimizing the acquisition conditions. Highresolution images visualizing structural features are, however, strongly modified by the microscope contrast transfer function and therefore the extraction of information directly from these images may be problematic. Although electron diffraction data are not as intuitive as imaging, they do, however, possess superior information in terms of spatial resolution. Traditional methods for 2D and 3D data collection have been used successfully to determine crystal structures of organic compounds. In general, if not enough reflections were collected, additional information like phases retrieved from high-resolution images or energy considerations can be used to solve the structure. Data sets with appropriate number and quality of reflections can be used for ab initio structure solution by traditional direct methods without any additional information. The fact that electron diffraction data are expected to be slightly dynamical even for organics does not hamper the structure solution but may lead to problems in refinement. The main aim for diffraction data collection from organic materials is therefore maximizing the coverage of the reciprocal space, while simultaneously minimizing the dose to the specimen. Manual electron diffraction data collection by design has low space coverage. Additionally, the need to pre-orient the crystal in order to collect a manual diffraction data set is quite a time-consuming task. Beam-sensitive materials can easily be destroyed during this procedure. Additionally, considerable experience is required to perform the experiment. These technical challenges explain why so few structure solutions of organic materials based on electron diffraction data have been published so far. The development of automated routines for data collection and processing is a significant reduction of the barrier to enter the field of electron crystallography for nonexperts in either TEM or crystallography. Automatic routines allow speeding up the complete structure solution procedure. The full ab initio structure analysis on CNBA (including data collection, processing, and structure solution) from automatically collected electron diffraction data took about 2 days, whereas the structure solution based on manually collected data took over one year. From the timescale of the analysis automated electron diffraction experiments are now approaching single-crystal X-ray data collection and analysis. These benefits of automated procedures will increase the number of crystal structures solved by electron diffraction significantly and will have a considerable impact on the field of electron crystallography and structure analysis in general. The future of electron crystallography is an increasing use of the automated procedures and therefore a wide application of the method, especially to nanocrystalline materials. The ADT approach is still in its infancy and further refinements and optimization of the technique are expected. In particular, optimization of electron dose distribution during the data acquisition will be very beneficial for beam-sensitive materials. The collection of diffraction data with smaller tilt steps should allow more precise reconstruction of the reflection shape and recorded diffraction intensity. Further development of the processing of ADT data to more accurately account for the experimental geometry will lead to more accurate diffraction intensities and in turn more accurate structure solutions. With increasing application of electron diffraction crystallography we hope to see wider adoption of X-ray crystallography structure solution methods such as charge flipping and “shake and bake”128 to the solution of electron crystallographic data. Although the automated technique has been demonstrated as successful in producing structures, there is a problem with refinement of the final structure, due to reduced but still existing dynamical scattering effects. Steps to remedy this problem will be of particular interest in the future.

404

U. Kolb et al.

Acknowledgements The authors are very grateful to Galina Matveeva and Iryna Andrusenko for assistance in the preparation of this manuscript.

References 1. Cowley, J. M., Ed. Electron Diffraction Techniques. Vols. 1–2; Oxford University Press: Oxford, UK, 1992. 2. Andr´e, I.; Mazeau, K.; Tvaroska, I.; Putaux, J.-L.; Winter, W. T.; Taravel, F. R.; Chanzy, H. “Molecular and crystal structures of insulin from electron diffraction data,” Macromolecules, 1996, 29, 4626–4635. 3. Dorset, D. L. “Electron crystallography,” Acta Crystallographica, 1996, B52, 753–769. 4. Keyse, R. J.; Goodhew, P.; Garratt-Reed, A. J.; Lorimer, G. W. Introduction to Scanning Transmission Electron Microscopy; Springer: Dordrecht, The Netherlands, 1998. 5. Haruta, M.; Yoshida, K.; Kurata, H.; Isoda, S. “Atomic resolution adf-stem imaging of organic molecular crystal of halogenated copper phthalocyanine,” Ultramicroscopy, 2008, 108, 545–551. 6. Buban, J. P.; Ramasse, Q.; Gipson, B.; Browning, N. D.; Stahlberg, H. “High-resolution lowdose scanning transmission electron microscopy,” Journal of Electron Microscopy, 2010, 59, 103–112. 7. Hodgkinson, J. L.; Horsley, A.; Stabat, D.; Simon, M.; Johnson, S.; da Fonseca, P. C.; Morris, E. P.; Wall, J. S.; Lea, S. M.; Blocker, A. J. “Three-dimensional reconstruction of the shigella t3ss transmembrane regions reveals 12-fold symmetry and novel features throughout,” Nature Structural and Molecular Biology, 2009, 16, 477–485. 8. Snider, D. A.; Addicks, W.; Owens, W. “Polymorphism in generic drug product development,” Advanced Drug Delivery Reviews, 2004, 56, 391–395. 9. Schmidt, M. U. “Crystal engineering and polymorphism of organic pigments,” Advances in Colour Science and Technology, 2003, 6, 59–61. 10. Bernstein, J. Polymorphism in Molecular Crystals; Clarendon Press: Oxford, UK, 2002. 11. Legrand, L.; Ries-Kautt, M.; Robert, M.-C. “Two polymorphs of lysozyme nitrate: temperature dependence of their solubility,” Acta Crystallographica, 2002, D58, 1564–1567. 12. Treacy, M. M. J.; Newsam, J. M. “Electron beam sensitivity of zeolite l,” Ultramicroscopy, 1987, 23, 411–420. 13. Yuan, Z.-Y.; Zhou, W.; Parvulescu, V.; Su, B.-L. “Electron beam irradiation effect on nanostructured molecular sieve catalysts,” Journal of Electron Spectroscopy and Related Phenomena, 2003, 129, 189–194. 14. Rozhdestvenskaya, I. V.; Kogure, T.; Abe, E.; Drits, V. A. “A structural model for charoite,” Mineralogical Magazine, 2009, 73, 883–890. 15. Zega, T. J.; Garvie, L. A. J.; Dodony, I.; Buseck, P. R. “Serpentine nanotubes in the mighei cm chondrite,” Earth and Planetary Science Letters, 2004, 223, 141–146. 16. Reimer, L. Transmission Electron Microscopy; Springer-Verlag: Springer, New York, USA, 1984. 17. Turner, S.; Lebedev, O. I.; Schr¨oder, F.; Esken, D.; Fischer, R. A.; Van Tendeloo, G. “Direct imaging of loaded metal-organic framework materials (metal@mof-5),” Chemistry of Materials, 2008, 20, 5622–5627. 18. Voigt-Martin, I. G.; Yan, D. H.; Wortmann, R.; Elich, K. “The use of simulation methods to obtain the structure and conformation of 10-cyano-9,9 -bianthryl by electron diffraction and high-resolution imaging,” Ultramicroscopy, 1995, 57, 29–43. 19. Voigt-Martin, I. G.; Yan, D. H.; Yakimansky, A.; Schollmeyer, D.; Gilmore, C. J.; Bricogne, G. “Structure determination by electron crystallography using both maximum-entropy and simulation approaches,” Acta Crystallographica, 1995, A51, 849–868.

Structural Characterization of Organics

405

20. Shmueli, U., Ed. International Tables for Crystallography, Volume B, Reciprocal Space; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2001. 21. Williams, D. B.; Carter, C. B. Transmission Electron Microscopy; Plenum Press: New York, USA, 1996. 22. Milne, J. L. S.; Subramaniam, S. “Cryo-electron tomography of bacteria: progress, challenges and future prospects,” Nature Reviews Microbiology, 2009, 7, 666–675. 23. Spence, J. C. H. High-Resolution Electron Microscopy; Oxford University Press: Oxford, UK, 2003. 24. Fryer, J. R.; Smith, D. J. “High resolution electron microscopy of molecular crystals i. quaterrylene, c40h20,” Proceedings of the Royal Society of London, Series A, Mathematical and Physical Sciences, 1982, 381, 225–240. 25. Bravais, A. Etudes Crystallographiques; Paris: Gauthier-Villars, 1913, p. 1866. ´ 26. Friedel, G. Etudes Sur la loide Bravais. Bull. Soc. Fr. Mineral. 1907, 9:32, 326. 27. Donnay, J. D. H.; Harker, D. “A new law of crystal morphology extending the law of bravais,” American Mineralogist, 1937, 22, 446–467. 28. Swarbrick, J. Encyclopedia of Pharmaceutical Technology, Vol. 4; Informa Healthcare: London, UK, 2007. 29. Martin, D. C.; Chen, J.; Yang, J.; Drummy, L. F.; K¨ubel, C. “High resolution electron microscopy of ordered polymers and organic molecular crystals: recent developments and future possibilities,” Journal of Polymer Science Part B: Polymer Physics, 2005, 43, 1749–1778. 30. Dorset, D. L. Structural Electron Crystallography; New York: New York, USA, 1995. 31. Kopp, S.; Wittmann, J. C.; Lotz, B. “Epitaxial crystallization and crystalline polymorphism of poly(1-butene): forms iii and ii,” Polymer, 1994, 53, 908–915. 32. Di Lorenzo, M. L.; Righetti, M. C.; Wunderlich, B. “Influence of crystal polymorphism on the three-phase structure and on the thermal properties of isotactic poly(1-butene),” Macromolecules, 2009, 42, 9312–9320. 33. Cartier, L.; Okihara, T.; Ikada, Y.; Tsuji, H.; Puiggali, J.; Lotz, B. “Epitaxial crystallization and crystalline polymorphism of polylactides,” Polymer, 2000, 41, 8909–8919. 34. Bergfors, T. M. Protein Crystallization: Techniques, Strategies, and Tips. A Laboratory Manual; International University Line: La Jolla, California, USA, 2002. 35. Georgieva, D. Ph.D. “Electron Crystallography of three dimensional protein crystals. thesis, Leiden University: Leiden, The Netherlands, 2008. 36. Jiang, L.; Georgieva, D.; Zandbergen, H. W.; Abrahams, J. P. “Unit-cell determination from randomly oriented electron-diffraction patterns,” Acta Crystallographica, 2009, D65, 625– 632. 37. Gonen, T.; Cheng, Y.; Sliz, P.; Hiroaki, Y.; Fujiyoshi, Y.; Harrison, S. C.; Walz, T. “Lipidprotein interactions in double-layered two-dimensional aqpo crystals,” Nature, 2005, 438, 633– 638. 38. Renault, L.; Chou, H.-T.; Chiu, P.-L.; Hill, R. M.; Zeng, X.; Gipson, B.; Zhang, Z. Y.; Cheng, A.; Unger, V.; Stahlberg, H. “Milestones in electron crystallography,” Journal of Computer-Aided Molecular Design, 2006, 20, 519–527. 39. Tivol, W. F.; Briegel, A.; Jensen, G. J. “An improved cryogen for plunge freezing,” Microscopy and Microanalysis, 2008, 14, 375–379. 40. Kumar, S.; Adams, W. W. “Electron beam damage in high temperature polymers,” Polymer, 1990, 31, 15–19. 41. Fryer, J. R.; Holland, F. “The reduction of radiation damage in the electron microscope,” Ultramicroscopy, 1983, 11, 67–70. 42. Fryer, J. R. “Radiation damage in organic crystalline films,” Ultramicroscopy, 1984, 14, 227–236. 43. Rossmenn, M. G.; Arnold, E. International Tables for Crystallography, Volume F, Crystallography of Biological Macromolecules; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2001.

406

U. Kolb et al.

44. Fryer, J. R.; McConnell, C. H.; Zemlin, F.; Dorset, D. L. “Effect of temperature on radiation damage to aromatic organic molecules,” Ultramicroscopy, 1992, 40, 163–169. 45. Ohno, T.; Sengoku, M.; Arii, T. “Measurements of electron beam damage for organic crystals in a high voltage electron microscope with image plates,” Micron, 2002, 33, 403–406. 46. Misra, M.; Egerton, R. F. “Electron irradiation damage to biomolecules,” Ultramicroscopy, 1984, 15, 337–344. 47. Fryer, J. R. “The effect of dose rate on imaging aromatic organic crystals,” Ultramicroscopy, 1987, 23, 321–328. 48. Chen, J. Z.; Sachse, C.; Xu, C.; Mielke, T.; Spahn, C. M. T.; Grigorieff, N. “A dose-rate effect in single-particle electron microscopy,” Journal of Structural Biology, 2008, 161, 92–100. 49. Koshino, M.; Kurata, H.; Isoda, S. “Stability due to peripheral halogenation in phthalocyanine complexes,” Microscopy and Microanalysis, 2007, 13, 96–107. 50. Frank, J., Ed. Electron Tomography: Methods for Three-Dimensional Visualization of Structures in the Cell; Springer: Berlin, 2006. 51. Giacovazzo, C.; Monaco, H. L.; Artioli, G.; Viterbo, D.; Ferraris, G.; Gilli, G.; Zanotti, G.; Catti, M. Fundamentals of Crystallography; Oxford University Press: Oxford, UK, 2002. 52. Cowley, J. M. Electron Diffraction Techniques, Vols. 1–2; Oxford University Press: Oxford, UK, 1995. 53. Fultz, B.; Howe, J. M. Transmission Electron Microscopy and Diffractometry of Materials; Springer: Berlin, Germany, 2008. 54. Peng, L.-M.; Dudarev, S. L.; Whelan, M. J. High-Energy Electron Diffraction and Microscopy; Oxford University Press: Oxford, UK, 2004. 55. Pinsker, Z. G. Electron Diffraction; Butterworths: London, UK, 1953. 56. Vainshtein, B. K. “Kinematic theory of intensities in electron diffraction patterns. part 2. patterns from textures and polycrystalline aggregates,” Soviet Physics—Crystallography, 1956, 1, 117–122. 57. Dorset, D. L. “Correlations, convolutions and the validity of electron crystallography,” Zeitschrift f¨ur Kristallographie, 2003, 218, 237–246. 58. Weirich, T. E. “Electron diffraction structure analysis: structural research with low-quality diffraction data,” Zeitschrift f¨ur Kristallographie, 2003, 218, 269–278. 59. Kirkland, E. J. Advanced Computing in Electron Microscopy; Plenum Publishing Corporation, New York, New York, USA, 1998. 60. Jansen, J.; Tang, D.; Zandbergen, H. W.; Schenk, H. “Msls, a least-squares procedure for accurate crystal structure refinement from dynamical electron diffraction patterns,” Acta Crystallographica, 1998, A54, 91–101. 61. Uyeda, N.; Kobayashi, T.; Suito, E.; Harada, Y.; Watanabe, M. “Molecular image resolution in electron microscopy,” Journal of Applied Physics, 1972, 43, 5181–5189. 62. Fryer, J. R. “Molecular imaging of thin-film polymorphs and phase transformations in metal-free phthalocyanine,” Acta Crystallographica, 1979, A35, 327–332. 63. Kobayashi, T.; Ogawa, T.; Moriguchi, S.; Suga, T.; Yoshida, K.; Kurata, H.; Isoda, S. “Inhomogeneous substitution of polyhalogenated copper phthalocyanine studied by high-resolution imaging and electron crystallography,” Journal of Electron Microscopy, 2003, 52, 85–90. 64. Williams, R. C.; Fisher, H. W. “Electron microscopy of tobacco mosaic virus under conditions of minimal beam exposure,” Journal of Molecular Biology, 1970, 52, 121–123. 65. Glaeser, R. M.; Kuo, I.; Budinger, T. F. “Method for processing of periodic images at reduced levels of electron irradiation,” Proceedings of the 29th Electron Microscopy Society of America, 1971. 66. Kaiser, U.; Chuvilin, A.; Schr¨oder, R. R.; Haider, M.; Rose, H. “Sub-Ångstrøm Low-Voltage Electron Microscopy—Future Reality for Deciphering the Structure of Beam-Sensitive Nanoobjects?”, EMC 2008 14th European Microscopy Congress; Aachen, Germany, September 1–5, 2008. 67. Stewart, A.; Kolb, U. Preliminary Investigations into the application of Automated diffraction tomography to 3D protein nanocrystals.

Structural Characterization of Organics

407

68. Dwyer, C.; Kirkland, A. I.; Hartel, P.; M¨uller, H.; Haider, M. “Electron nanodiffraction using sharply focused parallel probes,” Applied Physics Letters, 2007, 90, 151104-1-3. 69. Davisson, C.; Germer, L. H. “Diffraction of electrons by a crystal of nickel,” Nature, 1927, 119, 558–560. 70. Lebedeff, A. A. “A focusing method for producing electron diffraction patterns,” Nature, 1931, 128, 491. 71. Thomson, G. P.; Murison, C. A. “Electron diffraction by films of grease,” Nature, 1933, 131, 237. 72. Natta, G.; Rigamonti, R. “Esame con i raggi di elettroni di alcuni eteri grassi,” Rendiconti Accademia Nazionale Lincei, 1935, 6, 342–348. 73. Rigamonti, R. “La struttura della catena paraffinica studiata mediante i raggi di elettroni,” Gazzetta Chimica Italiana, 1936, 66, 174–182. 74. Karpov, V. L. Zhurnal Fizicheskoi Khimii, 1941, 15, 577–591. 75. Vainshtein, B. K. Structure Analysis by Electron Diffraction; Pergamon Press: New York, 1964. [Electrono76. Vainstein, B. K. “ graphic investigation of diketopiperizine],” Zhurnal Fizicheskoi Khimii, 1955, 29, 327–344. 77. Lobachev, A. N.; Vainshtein, B. K. “ [Electronographic investigation of urea],” Soviet Physics—Crystallography, 1961, 6, 313–317. 78. Dvoryankin, V. F.; Vainshtein, B. K. “ [Electronographic investigation of thiourea],” Soviet Physics—Crystallography, 1960, 5, 564–574. 79. Dvoryankin, V. F.; Vainshtein, B. K. “ [Electronographic investigation of low temperature signetoelectric phase of thiourea],” Soviet Physics—Crystallography, 1962, 6, 765–772. 80. Vainshtein, B. K.; D’yakon, I. A.; Ablov, A. V. “Crystal structure of compounds of copper (II) with threonine,” Soviet Physics—Doklady, 1971, 15, 645–647. 81. D’yakon, I. A.; Kairyak, L. N.; Ablov, A. V.; Chapurina, L. F. “ Cu(DL-Alan)(DL-Ser) [Investigation of atomic structure of copper salt of two different aminoasids Cu(DL-Alan)(DLSer)],” Doklady Akademii Nauk SSSR, 1977, 236, 103–105. 82. D’yakov, I. A.; Donu, S. V.; Chapurina, L. F. “

83.

84.

85. 86. 87.

88.

[Crystal strutures mied compounds of copper (II) with racemic and opticallz active serine and buterine],” Kristallografiya, 1998, 43, 656. Diacon, I. A.; Donu, S. V.; Chapurina, L. F. “ [Crystal structure of compounds of copper (II) with racemic threonine],” Kristallografiya, 2000, 45, 659. Tatarinova, L. I.; Vainshtein, B. K. “ [Investigation of poly-γ -methyl-L-glutamate in α form by the method of electron diffraction],” Visokomolek. Soed., 1962, 4, 261–269. [β-form of polyVainshtein, B. K.; Tatarinova, L. I. “ γ -methyl-L-glutamate],” Soviet Physics—Crystallography, 1967, 11, 494–498. Dorset, D. L. “The interpretation of quasi-kinematical single-crystal electron diffraction intensity data from paraffins,” Acta Crystallographica, 1976, A32, 207. Dorset, D. L.; McCourt, M. P. “Disorder and the molecular packing of c60 buckminsterfullerene: a direct electron crystallographic analysis,” Acta Crystallographica, 1991, A50, 344– 351. Yakimanski, A. V.; Kolb, U.; Matveeva, G. N.; Voigt-Martin, I. G.; Tenkovtsev, A. V. “The use of structure analysis methods in combination with semi-empirical quantum chemical calculations for the estimation of quadratic nonlinear optical coefficients of organic crystals,” Acta Crystallographica, 1997, A53, 603–614.

408

U. Kolb et al.

89. Voigt-Martin, I. G.; Zhang, Z. X.; Kolb, U.; Gilmore, C. “The use of maximum entropy statistics combined with simulation methods to determine the structure of 4-dimethylamino-3cyanobiphenyl,” Ultramicroscopy, 1997, 68, 43–59. 90. Yu, R. C.; Yakimansky, A. V.; Kothe, H.; Voigt-Martin, I. G.; Schollmeyer, D.; Jansen, J.; Zandbergen, H.; Tenkovtsev, A. V. “Strategies for structure solution and refinement of small organic molecules from electron diffraction data and limitations of the simulation approach,” Acta Crystallographica, 2000, A56, 436–450. 91. Unwin, P. N. T.; Henderson, R. “Molecular structure determination by electron microscopy of unstained crystalline specimens,” Journal of Molecular Biology, 1976, 94, 425–440. 92. Henderson, R.; Baldwin, J. M.; Ceska, T. A.; Zemlin, F.; Beckmann, E.; Downing, K. H. “Model for the structure of bacteriorhodopsin based on high-resolution electron cryomicroscopy,” Journal of Molecular Biology, 1990, 213, 899–929. 93. K¨uhlbrandt, W.; Wang, D. N.; Fujiyoshi, Y. “Atomic model of plant light-harvesting-complex by electron crystallography,” Nature, 1994, 367, 614–621. 94. Nogales, E.; Wolf, S. G.; Downing, K. H. “Structure of the alpha beta tubulin dimer by electron crystallography“, Nature, 1998, 391, 199–203. 95. Yakimanski, A. V.; Kolb, U.; Matveeva, G. N.; Voigt-Martin, I. G.; Tenkovtsev, A. V. “The use of structure analysis methods in combination with semi-empirical quantum-chemical calculations for the estimation of quadratic nonlinear optical coefficients of organic crystals,” Acta Crystallographica, 1997, A53, 603–614. 96. Voigt-Martin, I. G.; Kolb, U.; Kothe, H.; Yakimanski, A. V.; Yu, R.; Matveeva, G. N.; Tenkovtsev, A. V. “Electron crystallography and organic materials with non-linear optical properties,” Macromolecular Symposia, 1999, 146, 153–162. 97. Voigt-Martin, I. G.; Kolb, U. “Structure Determination by Electron Crystallography Using a Simulation Approach Combined with Maximum Entropy with the Aim of Improving Material Properties”, in Electron Crystallography; Dorset, D. L., Hovm¨oller, S., and Zou, X. D., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1997; 273–284. 98. Schmidt, M. U.; Bruhne, S.; Wolf, A. K.; Rech, A.; Bruning, J.; Alig, E.; Fink, L.; Buchsbaum, C.; Glinnemann, J.; van de Streek, J.; Gozzo, F.; Brunelli, M.; Stowasser, F.; Gorelik, T.; Mugnaioli, E.; Kolb, U. “Electron diffraction, x-ray powder analysis and pair distribution function analyses to determine the crystal structures of pigment yellow 213, c23h21o9h5,” Acta Crystallographica, 2009, B65, 189–199. 99. Dorset, D. L.; McCourt, M. P.; Li, G.; Voigt-Martin, I. G. “Electron crystallography of small organic molecules: criteria for data collection and strategies for structure solution,” Journal of Applied Crystallography, 1998, 31, 544–553. 100. Voigt-Martin, I. G.; Li, G.; Kolb, U.; Kothe, H.; Yakimanski, A. V.; Tenkovtsev, A. V.; Gilmore, C. “Structure determination to calculate nonlinear optical coefficients in a class of organic material,” Physical Review, 1999, B59, 6722–6735. 101. Kolb, U.; Matveeva, G. N. “Electron crystallography on polymorphic organics,” Zeitschrift F¨ur Kristallographie, 2003, 218, 259–268. 102. Kolb, U.; Gorelik, T. “Electron Crystallography on Polymorphs”, in Electron Crystallography, Novel Approaches for Structure Determination on Nanosized Materials; Weirich, T. E., L´ab´ar, J. L., and Zou, X., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2005; 421–433. 103. Nakashima, P. N. H.; Moodie, A. F.; Etheridge, J. “Structural phase and amplitude measurement from distances in convergent-beam electron diffraction patterns,” Acta Crystallographica, 2007, A63, 387–390. 104. Jin, L.; Milazzo, A.-C.; Kleinfelder, S.; Li, S.; Leblanc, P.; Duttweiler, F.; Bouwer, J. C.; Peltier, S. T.; Ellisman, M. H.; Xuong, N.-H. “Application of direct detection device in transmission electron microscopy,” Journal of Structural Biology, 2008, 161, 352–358. 105. Doyle, P. A.; Turner, P. S. “Relativistic hartree-fock x-ray and electron scattering factors,” Acta Crystallographica, 1968, A24, 390–397.

Structural Characterization of Organics

409

106. Cowley, J. M. “Scattering Factors for the Diffraction of Electrons by Crystalline Solids”, in International Tables of Crystallography, Vol. C; Wilson, A. J. C., Ed.; Kluwer Academic Publishers: Dordrecht, 1992; 223–244. 107. Burla, M. C.; Caliandro, R.; Camalli, M.; Carrozzini, B.; Cascarano, G. L.; De Caro, L.; Giacovazzo, C.; Polidori, G.; Diligi, S.; Spagna, R. “Il milione: a suite of computer programs for crystal structure solution of proteins,” Journal of Applied Crystallography, 2007, 40, 609–613. 108. Sheldrick, G. M. SHELXL97. Program for the Refinement of Crystals Structures; University of G¨ottingen: Germany, 1997. 109. Sheldrick, G. M. “A short history of shelx,” Acta Crystallographica, 2008, A64, 112–122. 110. Gilmore, C. J.; Bricogne, G. “Mice computer program,” Methods in Enzymology, 1997, 277, 65–78. 111. Oszl´anyi, G.; S¨ut˜o, A. “Ab ignitio structure solution by charge flipping,” Acta Crystallographica, 2004, A60, 134–141. 112. Palatinus, L.; Chapuis, G. “Superflip—a computer program for the solution of crystal structures by charge flipping in arbitrary dimensions,” Journal of Applied Crystallography, 2007, 40, 786–790. 113. Putz, H.; Schoen, J. C.; Jansen, M. “Combined method for ‘ab initio’ structure solution from powder diffraction data,” Journal of Applied Crystallography, 1999, 32, 864–870. 114. Coelho, A. TOPAS-Academic V4.1 by Coelho Software; Brisbane, Australia, 2007. 115. Favre-Nicolin, V.; Cerny, C. “Fox, ‘free objects for crystallography’: a modular approach to ab initio structure determination from powder diffraction,” Journal of Applied Crystallography, 2002, 35, 734–743. 116. David, W. I. F.; Shankland, K.; Van de Streek, J.; Pidcock, E.; Motherwell, W. D. S.; Cole, J. C. “Dash: a program for crystal structure determination from powder diffraction data,” Journal of Applied Crystallography, 2006, 39, 910–915. 117. Stribeck, N.; N¨ochel, U. “Direct mapping of fiber diffraction patterns into reciprocal space,” Journal of Applied Crystallography, 2009, 42, 295–301. 118. Wukovitz, S. W.; Yeates, T. O. “Why protein crystals favour some space-groups over others,” Nature Structural Biology, 1995, 2, 1062–1067. 119. Zou, X.; Hovm¨oller, A.; Hovm¨oller, S. “Trice—a program for reconstructing 3d reciprocal space and determining unit-cell parameters,” Ultramicroscopy, 2004, 98, 187–193. 120. Zou, X.; Sukharev, Y.; Hovm¨oller, S. “Quantitative electron diffraction—new features in the program system eld,” Ultramicroscopy, 1993, 52, 436–444. 121. Dorset, D.; Gilmore, C. “Prospects for kinematical least-squares refinement in polymer electron crystallography,” Acta Crystallographica, 2000, A56, 62–67. 122. Kolb, U.; Gorelik, T.; K¨ubel, C.; Otten, M. T.; Hubert, D. “Towards automated diffraction tomography: part i—data acquisition,” Ultramicroscopy, 2007, 107, 507–513. 123. Kolb, U.; Gorelik, T.; Otten, M. T. “Towards automated diffraction tomography. part ii—cell parameter determination,” Ultramicroscopy, 2008, 108, 763–772. 124. Rozhdestvenskaya, I.; Mugnaioli, E.; Czank, M.; Depmeier, W.; Kolb, U.; Reinholdt, A.; Weirich, T. “The structure of charoite, (k,sr,ba,mn)15–16 (ca,na)32 [(si70 (o,oh)180 )](oh,f)4.0 * nh2 o, solved by conventional and automated electron diffraction,” Mineralogical Magazine, 2010, 74, 159–177. 125. Schade, Ch. S.; Mugnaioli, E.; Gorelik, T.; Kolb, U.; Panth¨ofer, M.; Tremel, W. “Solution synthesis of a new thermoelectric zn1+x sb nanophase and its structure determination using automated electron diffraction,” Journal of the American Chemical Society, 2010. in press. 126. Kolb, U.; Gorelik, T.; Mugnaioli, E. “Automated diffraction tomography combined with electron precession: a new tool for ab initio nanostructure analysis,” Materials Research Society Symposium Proceedings, 2009, 1184-GG01-05. 127. Own, C. S. System Design and Verification of the Precession Electron Diffraction Technique; Ph.D. thesis, Northwestern University, Evanston, IL, 2005. http://www.numis.northwestern. edu/Research/Current/precessions. 128. Weeks, C. M.; Miller, R. “The design and implementation of snb 2.0,” Journal of Applied Crystallography, 1999, 32, 120–134.

Polymer Reviews, 50:411–419, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.515766

Perspective Perspectives on Organic Photovoltaics JIANGENG XUE Department of Materials Science and Engineering, University of Florida, PO Box 116400, Gainesville, FL 32611, USA Organic photovoltaics have been considered as a future low-cost solar energy conversion solution, and have attracted tremendous academic and industrial interests in recent years due to many technological advantages of organic semiconductors (including both small molecules and conjugated polymers). Here we describe the basic design and operation of organic photovoltaic cells as well as the important metrics for characterizing their performance. An overview of the three review articles included in this special issue on organic photovoltaics is provided, followed by discussions on the remaining significant challenges for the further development and commercialization of the organic photovoltaic technologies. Keywords organic photovoltaics, organic electronic materials, small molecular weight organic compounds, conjugated polymers, solar energy, solar cells

1. Solar Energy and Photovoltaic Cells Energy is essential to every activity of mankind. Fossil fuels, which include oil, coal, and natural gas, have been the major energy source powering modern society for many years. However, the finite supply of fossil fuel sources, the adverse environmental effect caused by burning fossil fuels, and the concern about the nation’s energy security arising from the heavy reliance on imported fossil fuels, coupled with the rising world demand for energy, have made finding sufficient supplies of clean energy urgent and of utmost importance in the next half century.1 One of the most promising, yet vastly under-utilized, alternative energy sources is solar energy, which is clean, renewable, safe, ubiquitous, and abundant—covering 0.1% of the land on Earth with 10% efficient solar conversion systems would be sufficient to power the world. Using photovoltaic (PV) cells, also commonly known as solar cells, to directly convert sunlight to electricity is one of the major methods to capture and convert solar energy; however, such-generated solar electricity only provided a very small portion of the world’s electricity, less than 0.1%, despite its recent growth of 35–40% per annum.2

Received July 29, 2010; accepted August 6, 2010. Address correspondence to Prof. Jiangeng Xue, Department of Materials Science and Engineering, University of Florida, PO Box 116400, Gainesville, FL 32611, USA. E-mail: [email protected]

411

412

J. Xue

Economics has been the limiting factor here, as solar electricity is about five to ten times more expensive than electricity generated from fossil fuels (especially coal). The high costs of PV module manufacturing and installation are two dominant cost parameters for solar electricity. For “first-generation” PV modules based on crystalline silicon wafers, a substantial portion of module manufacturing costs (approximately $300/m2) comes from the costs of materials, most of which are already made in high volumes and have little potential for further cost reduction. The “second-generation” thin film PV technologies based on materials such as amorphous Si, CdTe, and Cu(In,Ga)Se2 offer advantages of lower material costs; however, at least till now, such advantages are somewhat negated by the lower efficiencies of these devices. Several different “third-generation” PV cells with ultrahigh efficiencies and/or ultra-low costs have been proposed,3 which, if successfully demonstrated, would significantly reduce the cost-to-efficiency ratio and make solar electricity competitive against or even cheaper than fossil fuel generated electricity.

2. Organic Photovoltaic Cells One potential third-generation PV technology is based on π -conjugated organic electronic materials, including both small molecular weight organic compounds or small molecules and conjugated polymers. The π -conjugation in these materials leads to an energy difference between the lowest unoccupied and highest occupied molecular orbitals (LUMO and HOMO, respectively) somewhere in the range of 1 to 3 eV, thus exhibiting semiconducting behavior and strong interactions with visible and near infrared photons. Compared with the more conventional, i.e. inorganic, semiconductors, these organic materials generally have significantly lower material costs. They can be easily made into thin films through inexpensive room-temperature processes. They are compatible with flexible substrates that can be light weight and inexpensive, which not only makes them suitable for high throughput, low cost roll-to-roll processing, but will greatly reduce the module installation costs. Moreover, many properties of organic materials can be tailored to suit particular applications through chemical synthesis. Hence, organic semiconductors are particularly suited for large area, low cost, light weight, and flexible device applications, such as photovoltaics. Those technological advantages have attracted many people in recent years to study the physical properties of organic materials and their potential applications in electronic and optoelectronic devices.4 Many organic devices have been demonstrated over the years, including organic light-emitting devices (OLEDs),5,6 organic photovoltaic (OPV) cells,7–9 photodetectors,9–11 thin film transistors,12–15 and memories.16,17 Commercial products of ultrathin, full-color, flat-panel displays based on OLEDs are now available in digital cameras, mp3 players, cellular phones, and television sets.4 Spurred by the need to find a low-cost solar energy conversion technology, the research interests in organic photovoltaics have been growing tremendously in recent years. Figure 1 shows the exponential growth in the number of peer-reviewed articles on organic photovoltaics published in scientific journals since 1985. Nearly 1200 papers on organic photovoltaics were published in 2009, and the annual growth rate in the last decade has been over 30%. With the introduction of new materials and device architectures, the power conversion efficiency of OPV cells have progressed steadily in the last decade, reaching nearly 8% in state-of-the-art laboratory devices.18–26 The operation of an OPV cell is somewhat different from that of an inorganic semiconductor based PV cell, especially on how charge carriers are generated. For an inorganic PV cell, upon absorption of photons by the inorganic semiconductor, free electrons and holes

Perspectives on Organic Photovoltaics

413

Figure 1. The year-by-year number of peer-reviewed journal publications on organic photovoltaics from 1985 to 2000 according to the ISI Web of Science database, obtained using the search criterion of “Topic = ((organic or polymer) and (photovoltaic or solar cell)).” The solid line highlights the over 30% annual growth rate in the last decade. Note that the 2010 data is incomplete (as of July 2010).

are generated in the conduction and valence bands, respectively, whose transport towards respective electrodes leads to a photocurrent or photovoltage. However, unlike the covalent bonding among atoms in inorganic semiconductors, organic semiconductors are held together by the much weaker van der Waals type intermolecular interactions, resulting in highly localized charges.27,28 Absorption of an incident photon thus leads to a bound electron-hole pair, or an exciton, on one single molecule or on neighboring molecules. The binding energy of an exciton ranges from 0.1 eV to 2 eV.29–31 Hence, almost all efficient OPV cells are based on the donor-acceptor (DA) heterojunction (HJ) structure, first demonstrated by Tang in 1986.7 As shown in Fig. 2 (lower right panel), with appropriate energy level alignment at the DA interface, the exciton binding energy can be overcome and efficient dissociation of excitons can occur through a rapid charge transfer process, leading to holes in the HOMO of the donor material (which has a smaller ionization potential) and electrons in the LUMO of the acceptor (which has a larger electron affinity).9 Driven by the built-in electric field or concentration gradients, these holes (electrons) are subsequently transported through the donor (acceptor) molecules towards the anode (cathode) where they are collected, generating a photocurrent or photovoltage. In an organic DA HJ, excitons need to diffuse to the hetero-interface in order to contribute to the photocurrent. However, the distance excitons can diffuse in organic materials before recombination (i.e. electron and hole in the same exciton annihilate each other) is typically very short, 10 nm or less, due to the weak van der Waals interactions among organic molecules.9,27 As a comparison, the thickness one needs for an organic film to absorb most of the incident light is approximately 100 nm. To circumvent this “exciton-diffusion bottleneck,”32 researchers have designed the so-called bulk heterojunction structure33,34 by blending the donor and acceptor materials to form a spatially distributed DA interface throughout the active layer, such that all excitons would easily find a nearby interface and dissociate into free charges. It is important to control the degree of phase separation between

414

J. Xue

Figure 2. Schematic illustration of the photovoltaic process in an organic donor-acceptor heterojunction, which include four consecutive steps: light absorption to generate an exciton, diffusion of the exciton to the donor-acceptor interface, charge transfer at the interface to cause the exciton to dissociate, and finally the transport of the hole in the HOMO of the donor material and electron in the LUMO of the acceptor material to respective electrodes.

the donor and acceptor materials such that the domain sizes are approximately the same as the respective exciton diffusion lengths, i.e. ∼10 nm. The transport properties of the bulk heterojunction are also critical to ensure photo-carriers be transported to the respective electrodes without much loss, which due to the energy level offsets at the DA interface requires the existence of percolated pathways for both the donor and acceptor materials. A main focal point of research in the polymer photovoltaics community, therefore, has been to realize bulk heterojunction structures with the optimal degree of phase separation and percolation by employing appropriate materials and/or appropriate processing conditions.

3. Important Performance Metrics for Organic Photovoltaic Cells Figure 3(a) shows the schematic device structure of a laboratory OPV cell. The device is fabricated on a glass (or plastic) substrate, and a transparent conducting layer of indium tin oxide (ITO) coated on the substrate functioning as the anode. Light is incident on the device through the transparent substrate and the ITO anode. After the deposition of the organic active layers using appropriate methods, the device is completed by depositing a reflecting metal cathode (such as aluminum). One of the most important figures of merit for a PV cell is its power conversion efficiency, ηP , or the ratio of the maximum electrical power output to the power of the incident radiation. As illustrated in Fig. 3(b), from the current-voltage (I-V) characteristics of a PV cell under illumination, the power conversion efficiency can be calculated as: ηP =

Imp Vmp ISC VOC max Pout = = FF, PO A PO A PO A

Perspectives on Organic Photovoltaics

415

Figure 3. Schematic illustrations of (a) device structure and (b) current-voltage (I-V) characteristics of an organic photovoltaic cell. The I-V characteristics under illumination can be used to determine the power conversion efficiency of the device, which is the ratio of the maximum electrical power output Pm, out to the power of the incident light. VOC and ISC are the open-circuit voltage and short-circuit current, respectively.

where PO is the illumination intensity, A is the device area, Imp and Vmp are the current and voltage at the maximum power point, respectively, ISC is the short-circuit current, VOC is the open-circuit voltage, and FF = (Imp Vmp )/(ISC VOC ) is the fill factor representing the “squareness” of the I-V curve under illumination. The three parameters, ISC (or more accurately, ISC /PO A), VOC , and FF, can be individually or collectively varied by material properties and/or device architectures. Other important characteristics of OPV cells include external and internal quantum efficiencies. The external quantum efficiency (EQE, ηEQE ), also known as the incident photon-to-electron conversion efficiency (IPCE), is the ratio of the number of photogenerated electrons collected at the electrode to the number of photons incident onto the device. On the other hand, the internal quantum efficiency (IQE, ηIQE ) is the ratio of the number of electrons collected to the number of photons absorbed in the device, and therefore is related to EQE through ηEQE = ηA × ηIQE , where ηA is the absorption efficiency. Given the photovoltaic process illustrated in Fig. 2, the EQE can be expressed as follows:9 ηEQE = ηA × ηED × ηCT × ηCC = ηA × ηI QE , where ηED is the exciton diffusion efficiency or the probability of excitons diffusing to the DA interface, ηCT is the charge-transfer efficiency at the interface (which could be close to unity with proper energy level alignment at the interface),9 and the charge collection efficiency ηCC is the probability of photogenerated charges collected at the electrodes. The quantum efficiencies are functions of both wavelength λ and applied voltage V. Although often times only the quantum efficiency values under short-circuit conditions (V = 0) are measured and reported, important information on the photocarrier behavior could be extracted from their wavelength and voltage dependencies.35

4. Overview of this Special Issue on Organic Photovoltaics This special issue of Polymer Reviews contains three reviews on the state-of-the-art of organic photovoltaics, covering different types of materials and different types of device architectures.

416

J. Xue

The review36 by Zheng and Xue from the University of Florida is dedicated to the use of small molecules in OPV devices. These types of materials are generally processed into thin films using vacuum thermal evaporation, which provides a clean environment to prepare multilayer structures and is also compatible with the deposition of metals such as aluminum as an electrode. The organic multilayer structures afford the opportunity to control the transport and dynamics of excitons and charge carriers,37 which has been shown to be critical for achieving high efficiency OLEDs suitable for commercial applications.38 For OPV cells, this feature has also been employed to make a double heterojunction structure and different structures of donor-acceptor heterojunctions with different morphological structures and/or different operation characteristics, as discussed in detail in the review by Zheng and Xue. The review39 by Liang and Yu from the University of Chicago is focused on the development of conjugated polymers for use in solution-processed polymer:fullerene OPV cells. The authors have collaborated with Yang Yang from the University of California at Los Angeles and Solarmer Energy Inc., a start-up company in California, and reported the world record efficiency for OPV cells, at nearly 8% in laboratory cells.26 Their review in this issue details the design, synthesis, and optimization of a series of polymers towards achieving better absorption of near infrared photons, providing a higher voltage output with the fullerene acceptor, and forming a more preferred nanoscale morphology in the polymer:fullerene bulk heterojunctions. The conventional OPV device structure shown in Fig. 2(a) involves the vacuum deposition of a metal electrode to complete the device, which provides significant challenges to integrate with the roll-to-roll processing of the polymer active layers. The review40 by Hau et al. from the University of Washington discusses the development of an inverted polymer solar cell structure that offers advantages to the conventional structure. This device structure enables the deposition of various layers onto flexible substrates using solution processing techniques that are compatible with industrial roll-to-roll fabrication. This review also describes the use of nanoparticles of metal oxides (ZnO, TiO2 ) and self-assembled monolayers to modify the interface electronic and chemical structures to influence the active layer morphology and device performance.

5. Future Development and Commercialization Prospect of the OPV Technologies There have been a lot of industrial interests in developing the OPV technologies, especially in recent years. Large companies such as BASF and start-up companies including Konarka Technologies, Inc, Plextronics, Inc., Solarmer Energy Inc., and Heliatek GmbH have been developing various materials and device technologies for OPV applications. In spite of much industrial optimism there are still many significant challenges ahead to eventually deliver the OPV technologies to the commercial market. It is an impressive feat to achieve a power conversion efficiency of nearly 8% in a laboratory cell. However, such efficiencies are significantly below what is needed to trigger large-scale commercial application, particularly considering the potential significant efficiency drop from laboratory devices to commercial modules. Further improvement in the device efficiency is surely needed in the future, which may depend on the development of improved active materials, processing methods, and/or device structures. A deeper understanding of the photovoltaic processes and the combination with theoretical studies could accelerate such development. Early work on OPV cells has mostly been focused on how to improve their efficiencies with new materials and/or new device structures. However, as more and more

Perspectives on Organic Photovoltaics

417

efficient devices have been reported and now the highest efficiency has reached nearly 8%, other practical factors of the devices become increasingly important. Among these are the lifetime/stability considerations of the OPV cells and the manufacturability of these technologies. Conventional photovoltaic modules can easily offer operation lifetimes in excess of 25 years. There have been encouraging results on OPV cells achieving operating lifetimes of a few thousand hours. However, overall the lifetime and stability of organic-based photovoltaic devices are very much understudied, as with the means to improve device stability and enhance lifetime. Another critical issue for the commercialization of OPV technologies is related to the manufacturability of OPV devices. As the main interest in these technologies lies in their premise to deliver low-cost solar energy conversion, the realization of large-scale, high-throughput manufacturing processes will be crucial to the wide-spread adoption of the technologies. Polymers appear to be more advantageous in this regard, due to the availability of many solution-based printing technologies.41 However, small molecules are not without roll-to-roll manufacturing options that can be integrated with vacuum or vapor deposition processes. The ease of depositing small molecule-based multilayer structures offers unique opportunities to scientifically study various physical processes such as charge generation/recombination/transport and exciton generation/diffusion/dissociation in an organic donor-acceptor heterojunction. With a well-defined molecular structure and higher material purity, small molecules may also have advantages over polymers in terms of quality control and long-term stability.

Acknowledgments The author gratefully acknowledges financial support from the National Science Foundation, the US Department of Energy Solar Energy Technologies Program (SETP), and the Florida Energy Systems Consortium (FESC) for research on organic-based photovoltaics.

References 1. Basic Research Needs for Solar Energy Utilization; US Department of Energy Report of the Basic Energy Sciences Workshop on Solar Energy Utilization, 2005. 2. Surek, T. “Crystal growth and materials research in photovoltaics: progress and challenges,” J. Crystal Growth, 2005, 275, 292–304. 3. Green, M. A. “Third generation photovoltaics: concepts for high efficiency at low cost,” Proc. Electrochem. Soc., 2001, 10, 3–19. 4. Forrest, S. R. “The path to ubiquitous and low-cost organic electronic appliances on plastic,” Nature, 2004, 428, 911–918. 5. Tang, C. W.; Vanslyke, S. A. “Organic electroluminescent diodes,” Appl. Phys. Lett., 1987, 51, 913–915. 6. Baldo, M. A.; O’Brien, D. F.; You, Y.; Shoustikov, A.; Sibley, S.; Thompson, M. E.; Forrest, S. R. “Highly efficient phosphorescent emission from organic electroluminescent devices,” Nature, 1998, 395, 151–154. 7. Tang, C. W. “2-layer organic photovoltaic cell,” Appl. Phys. Lett., 1986, 48, 183–185. 8. Brabec, C. J.; Sariciftci, N. S.; Hummelen, J. C. “Plastic solar cells,” Adv. Funct. Mater., 2001, 11, 15–26. 9. Peumans, P.; Yakimov, A.; Forrest, S. R. “Small molecular weight organic thin-film photodetectors and solar cells,” J. Appl. Phys., 2003, 93, 3693–3723. 10. Yu, G.; Pakbaz, K.; Heeger, A. J. “Semiconducting polymer diodes—large-size, low-cost photodetectors with excellent visible-ultraviolet sensitivity,” Appl. Phys. Lett., 1994, 64, 3422–3424.

418

J. Xue

11. Xue, J.; Forrest, S. R. “Carrier transport in multilayer organic photodetectors: II. Effects of anode preparation,” J. Appl. Phys., 2004, 95, 1869–1877. 12. Katz, H. E. “Organic molecular solids as thin film transistor semiconductors,” J. Mater. Chem., 1997, 7, 369–376. 13. Dimitrakopoulos, C. D.; Mascaro, D. J. “Organic thin-film transistors: A review of recent advances,” IBM J. Res. Dev., 2001, 45, 11–27. 14. Horowitz, G. “Organic thin film transistors: From theory to real devices,” J. Mater. Res., 2004, 19, 1946–1962. 15. Xue, J.; Forrest, S. R. “Organic thin-film transistors based on bis(1,2,5-thiadiazolo)-p-quinobis (1,3-dithiole),” Appl. Phys. Lett., 2001, 79, 3714–3716. 16. Moller, S.; Perlov, C.; Jackson, W.; Taussig, C.; Forrest, S. R. “A polymer/semiconductor writeonce read-many-times memory,” Nature, 2003, 426, 166–169. 17. He, J.; Ma, L. P.; Wu, J. H.; Yang, Y. “Three-terminal organic memory devices,” J. Appl. Phys., 2005, 97, 064507. 18. Uchida, S.; Xue, J.; Rand, B. P.; Forrest, S. R. “Organic small molecule solar cells with a homogeneously mixed copper phthalocyanine: C60 active layer,” Appl. Phys. Lett., 2004, 84, 4218–4220. 19. Xue, J.; Uchida, S.; Rand, B. P.; Forrest, S. R. “Asymmetric tandem organic photovoltaic cells with hybrid planar-mixed molecular heterojunctions,” Appl. Phys. Lett., 2004, 85, 5757– 5759. 20. Li, G.; Shrotriya, V.; Huang, J. S.; Yao, Y.; Moriarty, T.; Emery, K.; Yang, Y. “High-efficiency solution processable polymer photovoltaic cells by self-organization of polymer blends,” Nat. Mater., 2005, 4, 864–868. 21. Ma, W. L.; Yang, C. Y.; Gong, X.; Lee, K.; Heeger, A. J. “Thermally stable, efficient polymer solar cells with nanoscale control of the interpenetrating network morphology,” Adv. Funct. Mater., 2005, 15, 1617–1622. 22. Xue, J.; Rand, B. P.; Uchida, S.; Forrest, S. R. “A hybrid planar-mixed molecular heterojunction photovoltaic cell,” Adv. Mater., 2005, 17, 66–71. 23. Kim, J. Y.; Lee, K.; Coates, N. E.; Moses, D.; Nguyen, T. Q.; Dante, M.; Heeger, A. J. “Efficient tandem polymer solar cells fabricated by all-solution processing,” Science, 2007, 317, 222–225. 24. Kroon, R.; Lenes, M.; Hummelen, J. C.; Blom, P. W. M.; de Boer, B. “Small bandgap polymers for organic solar cells (polymer material development in the last 5 years),” Polym. Rev., 2008, 48, 531–582. 25. Liang, Y. Y.; Feng, D. Q.; Wu, Y.; Tsai, S. T.; Li, G.; Ray, C.; Yu, L. P. “Highly efficient solar cell polymers developed via fine-tuning of structural and electronic properties,” J. Am. Chem. Soc., 2009, 131, 7792–7799. 26. Chen, H. Y.; Hou, J. H.; Zhang, S. Q.; Liang, Y. Y.; Yang, G. W.; Yang, Y.; Yu, L. P.; Wu, Y.; Li, G. “Polymer solar cells with enhanced open-circuit voltage and efficiency,” Nat. Photon., 2009, 3, 649–653. 27. Pope, M.; Swenberg, C. E. Electronic Processes in Organic Crystals and Polymers; Oxford University Press: New York, 1999. 28. Silinsh, E. A.; Capek, V. Organic Molecular Crystals: Interaction, Localization, and Transport Phenomena; AIP Press: New York, 1994. 29. Hill, I. G.; Kahn, A.; Soos, Z. G.; Pascal, R. A. “Charge-separation energy in films of piconjugated organic molecules,” Chem. Phys. Lett., 2000, 327, 181–188. 30. Knupfer, M.; Peisert, H.; Schwieger, T. “Band-gap and correlation effects in the organic semiconductor Alq(3),” Phys. Rev. B, 2002, 65, 033204. 31. Barth, S.; Bassler, H. “Intrinsic photoconduction in PPV-type conjugated polymers,” Phys. Rev. Lett., 1997, 79, 4445–4448. 32. Forrest, S. R. “The limits to organic photovoltaic cell efficiency,” MRS Bull., 2005, 30, 28–32. 33. Halls, J. J. M.; Walsh, C. A.; Greenham, N. C.; Marseglia, E. A.; Friend, R. H.; Moratti, S. C.; Holmes, A. B. “Efficient photodiodes from interpenetrating polymer networks,” Nature, 1995, 376, 498–500.

Perspectives on Organic Photovoltaics

419

34. Yu, G.; Gao, J.; Hummelen, J. C.; Wudl, F.; Heeger, A. J. “Polymer photovoltaic cells-enhanced efficiencies via a network of internal donor-acceptor heterojunctions,” Science, 1995, 270, 1789–1791. 35. Myers, J. D.; Tseng, T.-K.; Xue, J. “Photocarrier behavior in organic heterojunction photovoltaic cells,” Org. Electron., 2009, 10, 1182–1186. 36. Zheng, Y.; Xue, J. “Organic photovoltaic cells based on molecular donor-acceptor heterojunctions,” Polym. Rev., 2010, 50, 411–419. 37. Xue, J.; Forrest, S. R. “Carrier transport in multilayer organic photodetectors: I. Effects of layer structure on dark current and photoresponse,” J. Appl. Phys., 2004, 95, 1859–1868. 38. Eom, S. H.; Zheng, Y.; Chopra, N.; Lee, J.; So, F.; Xue, J. “Low voltage and very high efficiency deep-blue phosphorescent organic light-emitting devices,” Appl. Phys. Lett., 2008, 93, 133309. 39. Liang, Y.; Yu, L. “Development of semiconducting polymers for solar energy harvesting,” Polym. Rev., 2010, 50, 454–473. 40. Hau, S. K.; Yip, H.-L.; Jen, A. K.-Y. “Development of the inverted polymer solar cells,” Polym. Rev., 2010, 50, 474–510. 41. Krebs, F. Polymeric Solar Cells: Materials, Design, Manufacture; DEStech Publications, Lancaster, Pennsylvania, 2010.

Polymer Reviews, 50:420–453, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.516051

Reviews Organic Photovoltaic Cells Based on Molecular Donor-Acceptor Heterojunctions YING ZHENG AND JIANGENG XUE Department of Materials Science and Engineering, University of Florida, PO Box 116400, Gainesville, FL 32611, USA Small molecule based organic photovoltaic cells have attracted considerable interest as a potential solution for low-cost solar energy conversion in the near future. In this review, we first discuss the operation, advantages, and limitations of three different molecular donor-acceptor heterojunction structures. We then discuss the phase separation in molecule donor-acceptor mixtures and its impact on the photovoltaic performance. Methods to achieve bulk heterojunctions with controlled nanoscale geometry are also reviewed. Finally, we provide an outlook on major remaining challenges to the further development of this photovoltaic technology. Keywords organic photovoltaic cells, donor-acceptor heterojunctions, small molecules, organic semiconductors, photovoltaics, tandem cells

1. Introduction Organic photovoltaic (OPV) devices based on π -conjugated organic molecules or polymers have attracted great interest in the last two decades due to their potential for providing low-cost solar energy conversion.1–3 Compared with inorganic semiconductors used in commercially available solar modules, organic semiconductors have several technological advantages such as their low material cost, ease of processing, compatibility with flexible substrates and high-throughput manufacturing processes, and tunability of material properties via chemical structure modification.4 Starting from the 1986 seminal paper by Tang in which he demonstrated efficient charge generation by using an organic donor-acceptor (DA) heterojunction (HJ),5 there have been many significant advances in the field. With the introduction of new device architectures and incorporation of new materials,6–21 the power conversion efficiency of OPV devices has steadily increased over the years, and now reaches nearly 8%.22 In this review article, we will discuss several important issues related to the use of small molecular weight organic compounds or small molecules as the photoactive layers in Received July 29, 2010; accepted August 12, 2010. Address correspondence to Professor Jiangeng Xue, Department of Materials Science and Engineering, University of Florida, PO Box 116400, Gainesville, FL 32611, United States. E-mail: jxue@ mse.ufl.edu

420

Organic Photovoltaic Cells

421

Figure 1. Absorption spectra and molecular structures of four small molecular weight organic compounds used in photovoltaic cells. Here CuPc and PbPc are copper and lead phthalocyanines, respectively, and PTCBI = 3,4,9,10-perylene tetracarboxylic bis-benzimidazle.

OPV cells. These materials are generally deposited into thin films using vacuum thermal evaporation,23 and in some cases via vapor phase deposition techniques,24,25 allowing for the opportunity to readily prepare multilayer structures to control the transport of charge carriers and excitons.2,26,27 Figure 1 shows the absorption spectra of some of the small molecules used in OPV cells along with their molecular structures. The peak absorption coefficient of these organic molecules in visible and near infrared (IR) can reach (1–2) × 105 cm−1, suggesting that a film of 50–100 nm thick is sufficient for absorbing most of the incident light. Strong absorption in the visible and near IR spectral regions by these materials along with their good thermal stability thus makes them very attractive for photovoltaic applications. While there have been several reviews on the use of small molecules in OPV cells covering topics such as the physics of the photovoltaic process, various materials used, and efficiency limits,2,28–31 this review specifically focuses on the impact of various donor-acceptor heterojunction architectures on device performance. In Section 2, we first describe three different types of molecular donor-acceptor heterojunctions, and compare their photovoltaic properties. In Section 3, we discuss the understanding and the control of the phase separation process in molecular donor-acceptor mixtures. Works on achieving molecular bulk heterojunctions with a controlled morphology are reviewed in Section 4, which is then followed by conclusions and an outlook on the remaining challenges in Section 5. Discussions on polymer-based OPV devices can be found in the two companion reviews in this special issue.32,33

2. Types of Molecular Donor-Acceptor Heterojunctions For vacuum-deposited molecular donor-acceptor heterojunctions, several different heterojunction structures have been developed relying on the sequential and/or co-evaporation of the donor and acceptor materials. Shown in Fig. 2 are the three commonly referred heterojunction structures sandwiched between the anode and cathode of the OPV cell: (a) planar or bilayer HJ, (b) mixed or bulk HJ, and (c) planar-mixed HJ. The planar HJ structure is formed by the deposition of a pure donor layer followed by a pure acceptor layer. The simultaneous evaporation of the donor and acceptor materials leads to the formation of a mixed donor-acceptor layer, or a mixed or bulk HJ. By sandwiching a mixed layer

422

Y. Zheng and J. Xue

Figure 2. Schematic structure of three different molecular donor-acceptor heterojunction (HJ) structures: (a) planar or bilayer HJ, (b) mixed or bulk HJ, and (c) hybrid planar-mixed HJ.

in between pure donor and acceptor layers, a planar-mixed HJ can be formed, which has been shown to provide the highest performance when properly implemented.13,34,35 In this section, we will discuss the operation and the limitations/advantages for each HJ type. 2.1 Planar Donor-Acceptor Heterojunctions The first demonstration of efficient photovoltaic response in an organic heterojunction structure was based on the bilayer or planar HJ structure.5 The bilayer structure is also readily achievable using sequential deposition of the two organic materials in vacuum, and therefore has often been used as the structure to quickly screen new materials. As shown in Fig. 3, the interface between the donor and acceptor layers is where efficient exciton dissociation occurs, leading to holes (or more precisely, hole polarons) in the

Figure 3. Schematic energy level diagram of a donor-acceptor heterojunction. Here HOMO and LUMO are the highest occupied and lowest unoccupied molecular orbitals, respectively. The photovoltaic process consists of four steps: photon absorption, exciton diffusion, charge transfer (exciton dissociation) at the hetero-interface, and transport of separated charges toward respective electrodes for collection.

Organic Photovoltaic Cells

423

highest occupied molecular orbital (HOMO) level of the donor molecules and electrons (or electron polarons) in the lowest unoccupied molecular orbital (LUMO) level of the acceptor molecules. It is necessary to have a sufficiently large energy level offset at the DA interface to overcome the exciton binding energy, which is generally considered to be in the range of 0.1 to 2 eV.36–38 These separated charges then transport towards the respective electrodes where they are collected to form a photocurrent or photovoltage. Again, due to the energy level misalignment at the DA interface, holes are confined and transported in the donor material whereas electrons are confined and transported in the acceptor material. One of the major limiting factors to the efficiency of these planar HJ devices is the short exciton diffusion length of organic materials, LED , or the average distance excitons can diffuse in these materials before recombination. Due to the weak van der Waals type interactions among organic molecules,36 the exciton diffusion length is short, generally on the order of a few nm.2 In comparison, the light absorption length, LA , for these photoactive organic materials is typically on the order of 50–100 nm (corresponding to an absorption coefficient of (1–2)×105 cm−1, see Fig. 1). Hence, only excitons created very close to the DA interface, e.g. roughly within an exciton diffusion length of the interface, can diffuse to the interface with a high probability and contribute to the generation of separated charge carriers. Excitons generated farther away from the interface are more likely to undergo recombination and have less significant contribution to the photovoltaic process. To maximize the light absorption efficiency ηA , thick organic layers with thicknesses of at least LA are needed; however, to maximize the internal quantum efficiency (ηIQE ), thin layers with thicknesses on the order of LED are preferred so that most of the photogenerated excitons can diffuse to the interface and contribute to the photocurrent. As the external quantum efficiency ηEQE = ηA × ηIQE , a fundamental trade-off exists between optical absorption and exciton diffusion in these planar HJ devices. For the active materials used in the Tang cell,5 the exciton diffusion length is about 10 nm for the donor molecule, copper phthalocyanine (CuPc), and 3 nm for the acceptor molecule, 3,4,9,10-perylene tetracarboxylic bis-benzimidazle (PTCBI).2 This leads to a maximum external quantum efficiency at about 10%, and an overall power conversion efficiency of ηP ≈ 1%.5 The bilayer structure was further modified by Peumans et al. with the insertion of an additional wide-gap transparent exciton-blocking layer (EBL) between the acceptor layer and the cathode to form a double heterojunction (DHJ) structure.39,40 This DHJ structure is advantageous for devices with thin active layers. The transparent layer serves multiple functions including as a buffer layer to absorb damages to the organic layer during the deposition of the metal cathode, and as an optical spacer to enhance the optical field near the DA HJ when a thin acceptor layer is used. Furthermore, with its high energy gap, this layer also reduces the metal cathode induced quenching of excitons in the acceptor layer by blocking the close contact of excitons with the cathode, hence the name exciton-blocking. Using the DHJ structure, Peumans et al. achieved nearly the same ηEQE in CuPc/PTCBI cells with active layers as thin as 10 nm.39 While the overall power conversion efficiency of such a device was still only at approximately 1%, that efficiency was achieved with very low absorption of the incident light. Hence significant increases in ηP can be achieved by using various light trapping techniques to prolong the optical path of the incident light inside the active layer.39,41–43 Alternatively, stacking multiple thin cells vertically to form the tandem cell structure,44 more incident photons can be harvested without compromising ηIQE for each absorbed photon. A more than two-fold increase in ηP was also demonstrated using this technique2,10 which also benefited from the enhancement in absorption caused by plasmons in Ag nanoclusters used as the charge recombination zone connecting the subcells.2,45

424

Y. Zheng and J. Xue

Figure 4. I −V characteristics of CuPc/C60 organic photovoltaic cell under AM1.5G illumination of variable intensity. The maximum power output is indicated for illumination intensities of ≥150 mW/cm2, illustrating the effect of the series resistance on the FF. Reprinted with permission from Ref. 46. Copyright 2001, American Institute of Physics.

An obvious means to alleviate this so-called “exciton diffusion bottleneck”28 is to use active materials with longer exciton diffusion lengths. A material that has now been commonly used as the acceptor is C60 . The strong intersystem crossing in C60 leads to long-lived triplet exciton states, and it was shown that C60 has LED ≈ 40 nm, significantly longer than most other acceptor materials used for OPV devices.2 Using the combination of CuPc/C60 DA system with an EBL, Peumans et al. demonstrated ηP up to 3.6% under simulated AM1.5G spectrum with a device series resistance of 6.2 ·cm2 (see Fig. 4).46 Compared with the CuPc/PTCBI device, the significant increase in ηP in the CuPc/C60 device certainly benefited from the ∼10 times longer LED in C60 than in PTCBI. In addition, in the spectral region where CuPc absorbs strongly (550–750 nm), C60 does not absorb light very strongly but PTCBI does (see Fig. 1). With a longer LED in CuPc than in PTCBI (10 nm vs. 3 nm), this leads to improved harvesting of the incident light in this spectral range, and hence higher ηEQE for the CuPc/C60 cell compared with the CuPc/PTCBI cell. Xue et al. further showed that by reducing the series resistance of the CuPc/C60 cells to as low as 0.1 ·cm2, the current-voltage characteristics can be well described by the modified Shockley diode equation,47 and a maximum ηP up to 4.2% under simulated AM1.5G spectrum could be achieved at an incident intensity of several suns (1 sun = 100 mW/cm2).48 Another example of molecules with long LED is pentacene, a molecule that has been widely studied as the channel material for organic field-effect transistors.49–51 The fieldeffect mobility for holes in pentacene crystals has been measured to be on the order of 1 cm2/V·s, among the highest mobilities for carriers in organic materials.51,52 Experimental results on pentacene/C60 cells revealed that LED in pentacene could be up to 65 nm, much longer than that in CuPc, as shown in Fig. 5.53 The power conversion efficiency of pentacene/C60 cells, however, is not higher than that of CuPc/C60 devices for two main reasons. First, the long-wavelength absorption edge of pentacene is approximately 50 nm shorter than that of CuPc, leading to poorer absorption of near infrared photons. Second,

Organic Photovoltaic Cells

425

Figure 5. (a) External quantum efficiency (EQE) vs the wavelength (λ) of the incident light. (b) Calculated distribution of the squared electric field strength, |E|2, at λ = 670 nm normalized by that of the incident light, |E0 |2. Vertical lines indicate the location of each interface. The distance, z, is measured from the glass/ITO interface. (c) Calculated EQE as a function of pentacene diffusion length (LP ) at λ = 670 nm. LC60 = 40 nm was assumed. Inset: (n,k) values used for the calculation in (b) and (c). Reprinted with permission from Ref. 53. Copyright 2004, American Institute of Physics.

the open-circuit voltage, VOC , of the pentacene/C60 cells is approximately 0.1 V lower than that of the CuPc/C60 cells, which is attributed to the more significant non-geminate recombination of separated charge carriers at the hetero-interface in the pentance-based devices.54 2.2 Mixed or Bulk Donor-Acceptor Heterojunctions For polymer-based PV cells, bulk HJs formed from blends of the donor and acceptor materials using solution processes have been used to create an interpenetrating network with a spatially distributed DA interface.6,7 By controlling the phase separation of the two materials to several nanometers, the same length scale as the exciton diffusion lengths, one can expect most photogenerated excitons to have high probability of diffusing to a nearby DA interface and generate separated charge carriers upon dissociation. Furthermore, percolated pathways need to be formed for both materials in the bulk HJ for efficient collection of these charge carriers. Any morphological islands, bottlenecks, or cul-de-sacs in the bulk HJ will negatively affect the transport of charges and may lead to trapping and eventually the recombination of the charges. The formation of a good bulk HJ structure, however, is not trivial, as the morphology of the bulk HJ strongly depends on the properties of the donor and acceptor materials and the solution processing conditions, as well as

426

Y. Zheng and J. Xue

Figure 6. Comparison of the power conversion efficiency (ηP ), short-circuit current density (JSC ), open-circuit voltage (VOC ), and fill factor (FF) as functions of the illumination light intensity (PO ) for optimized planar and mixed HJ cells based on CuPc/C60 .

any post-processing treatments.8,55–57 A minor modification on the polymer structure or a seemingly trivial change of the solvent used for the materials could lead to a significant change in the bulk HJ morphology, and eventually the performance of the corresponding PV cell. Readers are referred to the two companion reviews in this special issue for more details.32,33 For vacuum-deposited small molecules, similar bulk HJ structures can be realized by co-evaporating the donor and acceptor molecules in vacuum to form a mixed thin film.2,11,34,58–62 Very similar to a polymer-based bulk HJ, the morphology of the molecular mixed HJ strongly impacts its photovoltaic properties. Specifically, the optimal degree of the phase separation between the donor and acceptor molecules should lead to domains with sizes similar to the exciton diffusion lengths, while at the same time achieving a percolated network for each of the molecular species. The degree of phase separation, however, depends on a number of parameters including the properties of the molecules involved, the composition of the mixture, and the processing conditions (deposition rate, substrate temperature), which will be further detailed in Section 3. Figure 6 shows a comparison of the OPV performance parameters of a planar HJ and a mixed HJ CuPc-C60 cell.48,59 The planar HJ consisted of a bilayer of 20 nm thick CuPc and 40 nm thick C60 layer, whereas the main active layer in the mixed HJ consisted of a 33 nm thick 1:1 mixed CuPc:C60 film. At 1 sun illumination intensity (or PO = 100 mW/cm2), the two cells have nearly the same ηP and the open-circuit voltage (VOC ). However, the mixed HJ cell has a higher short-circuit current density (JSC ), whereas the planar HJ cell has a higher fill factor (FF). Furthermore, the dependencies of these parameters on the illumination intensity are quite different for the two cells. At PO <1 sun, VOC of the mixed HJ cell tends to be higher than that of the planar HJ cell. The FF of the planar HJ cell slightly increases with PO and remains at over 0.6 even at >10 suns illumination. However, for the mixed HJ cell, the FF decreases very noticeably with the increase of PO , from 0.54 at PO = 1 mW/cm2 to <0.42 at PO >2 suns. These intensity dependencies lead to a mostly constant ηP for the mixed HJ cell over the intensity range plotted here, whereas ηP of the planar HJ cell increases logarithmically with PO before it saturates under 4–10 suns of illumination intensity. The different PV characteristics of these two cells can be well understood based on the morphological structure and transport properties of the respective active layers. We have found that there is not much significant phase separation in CuPc:C60 mixed films with

Organic Photovoltaic Cells

427

a 1:1 mixing ratio (by weight).34,63 Hence excitons generated in the mixed HJ can easily reach a nearly CuPc/C60 hetero-interface and dissociate to create separated electrons and holes. At short-circuit, when most of these photogenerated charge carriers are collected, the mixed HJ cell thus has a higher JSC than the planar HJ in which many excitons generated away from the interface between the (pure) CuPc and C60 layers cannot reach the interface. Based on the charge carrier mobility measurement, we have also found that the hole transport across CuPc molecules and electron transport through C60 molecules in the mixed film are approximately ten times inferior to that in pure CuPc and C60 films, respectively, due to the larger average spacing among the same species of molecules in the mixture.63 This also suggests that there do not exist long, percolated transport paths for charges in the CuPc:C60 mixture. The lower charge mobilities in the mixed film also suggest that the extraction of the photogenerated charge carriers has a stronger dependence on the electric field inside the layer, and a higher electric field is needed to extract most of the photocarriers. In terms of device performance, this means that the magnitude of the photocurrent has a strong dependence on the applied bias, leading to a reduced FF as compared to the planar HJ cell. The reduction in FF is more significant at higher illumination intensities due to the higher concentration of photogenerated carriers. The differences in VOC for these two cells, however, have more to do with the dark current than with the photoresponse of the different active layers. Shown in Fig. 7 is a comparison of the current density–voltage (J −V) characteristics of the two cells in the dark.48,59 While both devices exhibit strong diode characteristics with rectification ratios >106 at ± 1V, the mixed HJ shows a somewhat lower dark current than the planar HJ cell at V <0.5 V. The difference in the dark current under reverse bias is approximately one order of magnitude. In both cases, the J −V characteristics under forward-bias can be fit to the modified Shockley diode equation:48,64 q(V − JRSA ) −1 , J = Js exp nkT

Figure 7. Comparison of the current density—voltage (J −V) characteristics in the dark for the CuPc/C60 based planar and mixed HJ cells. The solid lines are fits to the modified diode equation.

428

Y. Zheng and J. Xue

where q is the electron charge, k the Boltzmann’s constant, and T the temperature. The fitting yields a specific series resistance of RSA in the range of 0.1–0.2 ·cm2 for both cells, which is mostly related to the sheet resistance of the transparent ITO anode (∼20 /square).48 An ideality factor of n ≈ 2.0 was obtained for the planar HJ cell, which is reduced to 1.5–1.6 for the mixed HJ cell. The fitting also shows that the reverse-bias saturation current density, Js , is approximately one order of magnitude higher for the planar HJ device than the mixed HJ device, 5 × 10−7 A/cm2 (planar) vs. 3 × 10−8 A/cm2 (mixed). These fitting results can be explained by considering the relative contributions of the diffusion-emission current and recombination current in the dark. For the planar HJ device, n ≈ 2.0 suggests that the recombination current dominates,48,64 which is reasonable as the large energy level offsets46 of ∼1 eV for both HOMO and LUMO levels at the CuPc-C60 DA HJ significantly suppress the diffusion-emission current. However, with lower charge carrier mobilities in the mixture, the recombination current is expected to be reduced,65 which leads to a reduced overall dark current and an ideality factor somewhere in between that for the recombination current (n = 2) and for the diffusion-emission current (n = 1). Considering that the total current is the sum of the dark current and the photocurrent Jph , we can now obtain the following expression for VOC :48 VOC

−Jph (VOC ) JSC nkT nkT ln ln ∝ ln(PO ). = +1 ≈ q Js q Js

The logarithmic dependence of VOC on PO is therefore apparent as long as the approximation of JSC ≡ Jph (0) ≈ Jph (VOC ) can be reasonably justified. In addition, the slope in the semi-log plot of VOC vs. ln PO is proportional to the diode ideality factor (n ≈ 1.5 for the mixed HJ cell and ≈ 2.0 for the planar HJ cell). Overall, due to its lower dark current at V <0.5 V, the mixed HJ cell exhibits a higher VOC under illumination intensities up to 1 sun, as compared to the planar HJ cell (see Fig. 6). 2.3 Planar-Mixed Donor-Acceptor Heterojunctions As discussed above, exciton diffusion is a limiting process in a planar HJ cell, whereas charge transport/collection limits the performance of a mixed HJ cell. Hence, while thick layers are desired to maximize the absorption of the incident photons, thin layers are preferred to achieve high exciton diffusion efficiency ηED (for planar HJ cells) or charge collection efficiency ηCC (for mixed HJ cells). A new type of HJ, called hybrid planar-mixed HJ, was therefore developed to combine the advantages of these two types of HJs while mitigating their problems to some extent.34,35 The planar-mixed HJ (PM-HJ) consists of a mixed donor-acceptor layer sandwiched in between a pure donor layer and a pure acceptor layer. With relatively thin layers, the charge collection in the mixed layer and exciton diffusion in the pure layers are not seriously limited. The low absorption by thin layers is compensated by the fact that there are three layers in the PM-HJ all contributing to light absorption. The schematic cross-section and energy level diagram of a CuPc:C60 PM-HJ cell are shown in Figs. 8(a) and (b). Here dD , dm , and dA are the thicknesses of the donor, mixed, and acceptor layers, respectively. In this device, exciton dissociation and photocarrier generation occur throughout the mixed layer as well as at its interfaces with the adjacent pure donor and acceptor layers. Figure 8(c) shows a comparison of the photovoltaic characteristics of an optimized CuPc:C60 PM-HJ cell (dD = 15 nm, dm = 10 nm, and dA = 35 nm) with those of the optimized planar and mixed HJ cells.34,35,48,59 With a thin mixed layer, the PM-HJ

Organic Photovoltaic Cells

429

Figure 8. (a) Schematic cross-section device structure and (b) energy level diagram of a CuPcC60 planar-mixed HJ (PM-HJ) cell; (c) comparison of the photovoltaic performance parameters for optimized planar, mixed, and planar-mixed HJ cells. Reprinted with permission from Ref. 35. Copyright 2005, American Institute of Physics.

cell has an as high FF as the planar HJ cell at 1 sun illumination, significantly higher than that in the mixed HJ cell which has a 33 nm thick mixed layer, much thicker than the mixed layer in the PM-HJ cell. With thinner donor and acceptor layers and the presence of a mixed layer, the PM-HJ cell possesses ∼25% enhanced JSC as compared to the planar HJ cell, and is nearly the same as JSC of the mixed HJ cell. Overall, the PM-HJ structure leads to an approximately 40% improvement in ηP at 1 sun illumination over the planar or mixed HJ structure, clearly demonstrating the effectiveness of this new HJ structure. One should also note that the effectiveness of the PM-HJ structure depends on how the layer thicknesses compare with the relevant charge collection or exciton diffusion lengths. Using a simple phenomenological model to describe charge transport in lowmobility materials and assuming uniform charge generation, we have obtained ηCC for photogenerated charge carriers within a mixed layer as a function of the mixed layer thickness dm as:34,66 dm Lc 1 − exp − , ηCC = dm Lc where Lc is the charge collection length or the characteristic length for the transport of photogenerated charge carriers within the mixed layer without experiencing recombination.34,66 A similar equation could be obtained for ηED by replacing Lc with the exciton diffusion length LED and replacing dm with dD or dA , assuming uniform exciton generation (which may not be very accurate due to interference of the incident light and the light reflected off

430

Y. Zheng and J. Xue

Figure 9. Comparison between model predictions (solid lines) and experimental current density vs voltage (J −V) characteristics of a hybrid PM-HJ cell (open squares) and a mixed HJ cell (open circles) under simulated AM1.5 solar illumination. Reprinted with permission from Ref. 35. Copyright 2005, American Institute of Physics.

the metal cathode).2 As the drift lengths for electrons and holes are both proportional to the electric field, the charge collection length in the mixed layer is field-dependent:34,35,66 Lc ≈ Lc0

Vbi − V . Vbi

Here Vbi is the built-in potential of the device and Lc0 is the charge collection length at short-circuit. Assuming that the only field-dependent term contributing to the photocurrent is from the charge transport in the mixed layer, we can thus model the J −V characteristics of a mixed HJ and PM-HJ cells, as shown in Fig. 9, and obtain Lc0 = (45 ± 5) nm from the curve fitting. For the optimized PM-HJ device described above, we have dm << Lc0 , which results in a near unity ηCC . For devices with a thicker mixed layer, inevitably ηCC will be reduced, leading to a reduced FF for the device. Similarly, thicker donor or acceptor layers than those that have been used in the optimized PM-HJ device will lead to lower ηED . In either case, the effectiveness of the PM-HJ structure will be compromised. Note that similar three-layered structures have been reported previously, where 40 nm of a thick mixed layer of phthalocyanine and perylene derivative molecules were employed.67 The short-circuit charge collection length in such mixtures is, however, very short, Lc0 <10 nm.34,35 Hence, with dm >4 Lc0 , the advantage of the PM-HJ structure was not realized in those devices.

3. Phase Separation in Organic Donor-Acceptor Mixtures It has been discussed earlier that a bulk HJ composed of a mixture of donor and acceptor molecules with percolated structures presents advantages in achieving a high exciton diffusion efficiency.2,6,34,68 In an optimized bulk HJ structure, nanoscale three-dimensional

Organic Photovoltaic Cells

431

percolated networks of the donor and acceptor phases lead to both good exciton diffusion and charge collection. It has been found that the morphology of the DA networks strongly depends on the donor-to-acceptor mixing ratio and various processing conditions during film deposition as well as the molecular properties. The film morphology may even be further optimized by employing thermal or solvent post-annealing processes.11,16,55 One important parameter for determining the film morphology, the degree of phase separation in a DA mixture, strongly depends on both the intrinsic molecular properties and the thermodynamic/kinetic processes involved during the intermixing of the two molecular species. In this section, we review the phase separation process in mixtures of two molecular DA systems and discuss how it affects the final morphology within the mixtures. 3.1 Morphology of CuPc:C60 Mixtures The morphology of CuPc:C60 mixtures is found to strongly depend on the mixing ratio of the two molecular species.34,63 Figure 10 shows the scanning electron microscope (SEM) and atomic force microscope (AFM) images for three representative CuPc:C60 mixed films.63 Due to the tendency of forming orderly stacks of CuPc molecules, the neat CuPc film

Figure 10. [(a), (c), and (e)] Scanning electron microscope (SEM), and [(b), (d), and (f)] tapping mode atomic force microscope (AFM) images of 1000 Å thick [(a) and (b)] CuPc, [(c) and (d)] 19:1 CuPc:C60 , and [(e) and (f)] 3:1 CuPc:C60 films deposited on Si(100). The scale bar in the SEM images is 1 µm, and the scan size in the AFM images is 500 × 500 nm2 with the height given in nanometer. Reprinted with permission from Ref. 63. Copyright 2005, American Institute of Physics.

432

Y. Zheng and J. Xue

Figure 11. X-ray-diffraction patterns for CuPc, C60 , and various mixtures of CuPc:C60 films as well as for the quartz substrate, taken in the θ–2θ geometry using Cu Kα radiation. All organic films are 1000 Å thick. Reprinted with permission from Ref. 63. Copyright 2005, American Institute of Physics.

(Figs. 10(a) and (b)) displays a corrugated surface with small grains approximately 40 nm in diameter, which leads to a root-mean-square (rms) surface roughness of Rrms = 4.8 nm. However, introducing 5 wt% of C60 molecules into the neat CuPc film can partially suppress the stacking of CuPc molecules. As a result, the surface of the 19:1 CuPc:C60 mixed film (Figs. 10(c) and (d), Rrms = 2.9 nm) becomes smoother than that of the neat CuPc. Further increasing the C60 concentration to 25% leads to very smooth and nearly featureless surface morphology (Figs. 10(e) and (f), Rrms = 0.5 nm) with no visible evidence of grain structures. This suggests that the stacking of CuPc molecules is further suppressed with more C60 in the mixture. Another way to monitor the change of CuPc aggregates inside the CuPc:C60 mixed film is to compare the X-ray diffraction (XRD) patterns of films with different composition. Figure 11 shows a series of XRD patterns of pure and mixed CuPc and C60 films.63 For the pure CuPc film, a characteristic diffraction peak corresponding to the crystalline α-CuPc phase can be observed at 2θ = (6.8 ± 0.05)◦ .69 The pure C60 film, in contrast, shows no apparent diffraction peak, indicating that the film is amorphous, and consistent with its smooth surface morphology.63 For the CuPc:C60 mixed films, the intensity of the CuPc (200) peak gradually decreases as the C60 concentration increases. This diffraction peak becomes indiscernible in the 3:1 CuPc:C60 film. The mean crystallite size of the CuPc aggregates in the mixed films can also be estimated via the Scherrer formula by assuming the shape factor to be 0.9.63 The average grain size of L = 34 nm is obtained for the neat CuPc and 19:1 CuPc:C60 films, whereas L = 8 nm for the 6:1 and 3:1 CuPc:C60 films. These data suggest that negligible phase separation occurs in the mixed film with the C60 concentration larger than 25%. This agrees with conclusions inferred from observation of the film surface

Organic Photovoltaic Cells

433

Figure 12. The zero-field mobility, µ0 , for electrons (filled symbols) and holes (open symbols) in various pure and mixed films containing CuPc (diamonds), C60 (squares), and PTCBI (triangles), vs percent of C60 or PTCBI in the mixture. The mixed films are all CuPc:C60 except for the two circled symbols, which correspond to 5:4 CuPc:PTCBI. Reprinted with permission from Ref. 63. Copyright 2005, American Institute of Physics.

morphologies in Fig. 10, where the presence of CuPc aggregates induces corrugated surface morphologies with small polycrystalline grains and increased C60 concentration results in a smooth surface and amorphous structure.63 The electrical properties, especially the charge mobility, could also strongly depend on the level of phase separation inside the mixed films. Higher charge mobility usually results in better charge collection efficiency, which helps to improve the fill factor of OPV cells. The electron and hole mobility of the CuPc:C60 films as function of the mixing ratios is shown in Fig. 12.63 The charge mobility is obtained by modeling the J −V characteristics for electron- and hole-only devices using the space charge limit current (SCLC) theory36 (note that caution needs to be exercised when comparing the SCLC mobility to fieldeffect mobility obtained from field-effect transistor measurements due to the differences in the measurement geometry, surface vs. bulk effects, electric field strength, and carrier density).70 Starting at approximately 10−3 cm2/V·s, the hole mobility decreases by several orders of magnitudes when the CuPc concentration decreases from 100% to around 5%, whereas the electron mobility increases from ∼10−6 to ∼0.05 cm2/V·s as the C60 content increases from 5% to 100%. As the mean distance between neighboring molecules of the same species is larger in the DA mixed film than in a homogeneous layer, the hopping mobilities for charge carriers are expected to decrease upon intermixing of the molecular species. The mobility data of CuPc:C60 films confirm this trend. The hole mobility decreases significantly when the C60 concentration is greater than 25%, which is consistent with the observation from the morphology studies that the CuPc aggregates disappear in the 3:1 CuPc:C60 film.

434

Y. Zheng and J. Xue

3.2 Morphology of Pentacene:C60 Mixtures To achieve stronger phase separations than that in CuPc:C60 to improve charge transport in the mixture, we hypothesized to use molecules with stronger intermolecular interactions. Compared with CuPc, pentacene molecules are expected to show a stronger tendency to form stacks to maximize their π -π interaction. Field-effect transistors based on polycrystalline pentacene thin film show field-effect mobility as high as µeff = 0.3 ∼ 1.5 cm2/V·s, while that of CuPc is only 1.0 × 10−3 cm2/V·s,51,71 suggesting stronger π -π interaction among pentacene molecules in the crystalline phase. Since the OPV cell based on pentacene/C60 planar HJ show reasonable good PV performance (ηP as high as 2.7%),53,72 it is expected that higher cell efficiency may be achieved by simply forming pentacene:C60 mixed HJs based on the knowledge from the CuPc:C60 system. However, the devices actually suffer from significant performance loss after forming mixed HJs, which mainly due to the strong phase separation in the mixtures.73 Figure 13 compares the AFM images of 3 nm thick neat pentacene or C60 and their mixed films grown on Si substrates.73 For neat pentacene (Fig. 13(a)), the layer-plusisland growth mode can be observed. The first monolayer comprises of µm-sized domains, on top of which islands of second and subsequent monolayers are grown. The neat C60 film (Fig. 13(e)), on the other hand, shows a smooth and nearly featureless surface under AFM. However, the pentacene:C60 mixed films (Figs. 13(b)–(d)) show remarkably different morphologies from the two neat films. The surface roughness increases from Rrms = 2.0 nm in the 4:1 pentacene:C60 film to a maximum value of Rrms = 2.9 nm in the 1:2 pentacene:C60 film, both of which are higher than that of either neat film (1.4 nm for pentacene and 1.5 nm for C60 ). “Islands” with a typical size of 200 nm and height of a few nanometers as well as relatively flat domains in between the islands can be identified in the 4:1 pentacene:C60 film. As the C60 concentration increases, the island features become more prominent with

Figure 13. Tapping mode atomic force microscope (AFM) images of 3 nm thick films on Si (100) substrates: (a) neat pentacene, (e) neat C60 , pentacene:C60 = (b) 4:1, (c) 1:1 and (d) 1:2 mixed films. The scanning areas is 5 × 5 µm2, and the height scale is 25 nm for all the images.

Organic Photovoltaic Cells

435

Figure 14. X-ray diffraction patterns for 30 nm thick pentacene and pentacene:C60 mixed films with various mixing ratios, taken in the θ-2θ geometry. Reprinted with permission from Ref. 73. Copyright 2009, American Institute of Physics.

increased height, whereas the flat domains gradually diminish and become indiscernible as the C60 concentration becomes larger than 50%. Structural information of pentacene:C60 films can be obtained from the XRD study of films with different mixing ratios. Figure 14 shows the XRD patterns of 30 nm thick films of pentacene and pentacene:C60 films deposited on Si substrates.73 The neat pentacene film displays diffraction peaks from the thin-film (00n ) and bulk (00n) phases of pentacene.49,74 The presence of higher order diffraction peaks indicate the good crystalline order in the pentacene film. In 3:1 pentacene:C60 film however, the less prominent pentacene (001) peak disappears, and the intensity for the (001 ) peak is reduced significantly. Additionally, all higher-order diffraction peaks can hardly be discerned, suggesting the loss of long-range ordering among the pentacene stacks. As the C60 concentration continues to increase, the (001 ) peak intensity further decreases and eventually disappears in the 1:2 pentacene:C60 film. These data suggest that crystalline domains of pentacene aggregates exist in the pentacene-rich films, although the disorder increases with the C60 concentration. Combined with the information from the AFM study, the flat domains that appeared in the pentacene-rich films correspond to the crystalline pentacene aggregates which contribute to the characteristic pentacene diffraction peaks, whereas the islands could be pentacene aggregates with an amorphous structure. It is noted that the general trends observed in CuPc:C60 and pentacene:C60 are similar where crystalline aggregates are formed in the donor-rich films. However, the pentacene molecules exhibit a stronger tendency to aggregate compared to CuPc. For instance, the crystalline domain size in pentacene:C60 (∼200 nm) is larger than that in CuPc:C60 (∼40 nm). Moreover, the aggregation of CuPc can be strongly suppressed by introducing a small amount of C60 (15% by weight) into the mixture, whereas the crystalline pentacene phase can still be observed with C60 content as high as 50%. This could be attributed to the stronger intermolecular interactions among pentacene molecules. In addition, the energy barrier for pentacene molecules to rearrange and form stacks in the mixtures is much lower than that of CuPc due to their smaller size.73

436

Y. Zheng and J. Xue

3.3 Controlling the Degree of Phase Separation The different degrees of phase separation, hence different morphologies, in CuPc:C60 and pentacene:C60 could be mainly attributed to the different intermolecular interactions among the donor molecules. In addition, various ways have been used to manipulate the degree of phase separation in organic DA mixtures based on the understanding of thermodynamic/kinetic processes. Generally, the formation of bulk HJs starts with the highly intermixing of the two molecular species upon deposition, followed by a kinetic process of phase separation. Therefore, it may be possible to enhance or suppress the phase separation by simply controlling the amount of kinetic energy provided to the system. Peumans et al. showed that by annealing in a confined geometry, the phase separation in the CuPc:PTCBI mixtures can lead to the formation of the desired bulk HJs for OPV cell.2 Normally, the growth of mixed layers at elevated substrate temperatures leads to stronger phase separation and larger crystalline domains, which is advantageous for improving the charge collection efficiency. However, the increased domain sizes will induce lower exciton diffusion efficiency and increased surface roughness. With this consideration, the annealing of the mixed layer was conducted in a confined geometry with a metal cathode on top. The metal cathode stresses the organic film during annealing, preventing the concomitant formation of rough surface morphology while permitting moderate phase separation to occur in the mixed film. Figure 15(a)–(d) shows the cross-section SEM images of the layer structure of ITO/500 nm CuPc:PTCBI (4:1)/100 nm Ag before and after annealing at T = 450 K, 500 K, and 550 K for 15 min.11 The as-grown film does not exhibit any morphological features, whereas the domain features become more apparent as the annealing temperature

Figure 15. SEM images of cross-sections of CuPc:PTCBI (4:1) film on ITO. a, Not annealed. b–d, Annealed for b, 15 min at 450 K, c, 500 K and d, 550 K. e–h, The simulated effects of annealing on the interface morphology of a mixed-layer photovoltaic cell. Reprinted with permission from Ref. 11. Copyright 2003, Nature.

Organic Photovoltaic Cells

437

Figure 16. X-ray diffraction patterns of annealed CuPc:PTCBI films on ITO using the Cu-Kα line. Reprinted with permission from Ref. 11. Copyright 2003, Nature.

increases. Domain sizes of ∼20 nm are observed after annealing at 550 K. The XRD data in Fig. 16 also indicate the formation of crystalline domains in the mixed layer after annealing.11 Upon annealing, the intensity of diffraction peaks corresponding to the αCuPc phase increases with the annealing temperature.11 Simulation of the phase separation process upon annealing was also conducted for the 1:1 CuPc:PTCBI mixtures. As shown in Fig. 15(e)–(h), it can be seen that increasing the annealing temperature will dramatically alter the morphology of the mixed layer, and the results qualitatively match with the observed cross-section SEM images. After optimization of the mixing ratio, the annealing temperature and device structure, a maximum power conversion efficiency of ηP = 1.5% was achieved in the bulk HJ device, whereas only ηP = 1.3 × 10−2% was obtained in the as-grown bulk HJ device.2 Moreover, the annealed bulk HJ device shows twice the enhancement in ηP over the planar HJ device.5,39 These data suggest that enhanced phase separation is needed in CuPc:PTCBI to create the desired bulk HJ morphology to provide the optimized device performance. However, for pentacene:C60 bulk HJs, phase separation needs to be suppressed to achieve the desired DA morphology. As discussed earlier, the pentacene:C60 mixed layer has a very strong tendency to have phase separation, leading to a corrugated surface and large domains (with size of ∼200 nm) for poor exciton collection.73 For pentacene:C60 , the strong phase separation can be suppressed by increasing the C60 concentration or the deposition rate.73 Figure 17 shows the SEM and AFM images of 50 nm thick 1:1, 1:2, and 1:4 pentacene:C60 films deposited at a total rate 0.6 Å/s. The strong phase separation in the mixture results in numerous ridge-like structures on the surface of these films. The height of the ridges reaches nearly 200 nm in 1:1 pentacene:C60 , substantially higher than the nominal film thickness of 50 nm. As the C60 concentration increases from 50% to 80%, the density and height of the

438

Figure 17. [(a), (c), and (e)] Tapping mode atomic force microscope (AFM), and [(b), (d), and (f)] scanning electron microscope (SEM) images of 50 nm thick pentacene:C60 = 1:1 [(a) and (b)], 1:2 [(c) and (d)] and 1:4 [(e) and (f)] deposited at 0.6 Å/s (total rate) on Si substrate. The scale bar in the SEM images is 1 µm, and the scanning area in the AFM images is 5 × 5 µm2 while the height scale is 300 nm.

Organic Photovoltaic Cells

439

Table 1 Comparison of the short-circuit current density (JSC ), open-circuit voltage (VOC ), fill factor (FF), and power conversion efficiency (ηP ) of two bulk heterojunction devices based on pentacene:C60 = 1:1, 1:2 and 1:5.5 (by weight) under 120 mW/cm2 simulated AM 1.5G solar illumination

Pentacene:C60 (1:1) Pentacene:C60 (1:2) Pentacene:C60 (1:5.5)

JSC (mA/cm2)

VOC (V)

FF

ηP (%)

9.7 × 10−3 1.7 × 10−2 1.48

0.45 0.56 0.60

0.14 0.12 0.29

5.1 × 10−4 9.5 × 10−4 0.21

ridges decrease significantly, which leads to a reduction of roughness from Rrms = 50 nm in 1:1 pentacene:C60 to Rrms = 20 nm in 1:4 pentacene:C60 . Further suppressing the phase separation involves depositing the mixtures at a much higher rate. The high deposition rate hinders the diffusion or rearrangement of molecules on the surface, and promotes the intermixing of the two molecular species. Figure 18 shows the morphology of 1:1, 1:4, and 1:5.5 pentacene:C60 films grown at 6 Å/s. The long ridges previously observed in films deposited at a low rate are now broken down into much shorter ones with lower heights. With the same 1:1 mixing ratio, the surface roughness is reduced from Rrms = 50 nm for the film grown at 0.6 Å/s to 26 nm for the film with 6 Å/s. Full suppression of the phase separation in pentacene:C60 may require a combination of a high deposition rate and high C60 concentration, which is implied by the morphology observed in 1:5.5 pentacene film in Fig. 18. That film shows almost featureless morphology under AFM and SEM with a low roughness of Rrms = 6 nm. The photovoltaic performance based on pentacene:C60 bulk HJ confirms that reduced phase separation is necessary in order to achieve better efficiency. Table 1 compares the photovoltaic performance for the device with a structure of ITO/pentacene:C60 (50 nm)/bathocuproine (BCP) (8 nm)/Al.73 The active layers with different mixing ratio are deposited at ∼6 Å/s to avoid the extremely rough surfaces, which could cause poor contact between the active layer and the metal cathode. The photovoltaic performance shows great dependence on the mixing ratio of the pentacene and C60 molecules. The strong aggregation in 1:1 pentacene:C60 results in low JSC = 9.7 × 10−3 mA/cm2 and ηP = 5.1 × 10−4%. However, reasonably good photovoltaic performance is obtained in the 1:5.5 pentacene:C60 device with JSC = 1.48 mA/cm2 and ηP = 0.21%, which is a significant improvement over 1:1 pentacene:C60 . This is mainly achieved by reducing the phase separation so that a desired morphology is formed in the mixed layer for both efficient exciton diffusion and charge collection. Ideally, a certain level of phase separation in a DA bulk HJs is considered to be beneficial for enhancing PV performance; both a too strong or too weak phase separation could be disadvantageous. Strong phase separation may induce domains with sizes significantly larger than the exciton diffusion length, whereas weak phase separation may lead to poor charge transport across the mixed layer due to the absence of percolated transport pathways. Depending on the intermolecular interaction between the molecular species in different DA mixtures, the as-grown mixed film may suffer from insufficient, CuPc:PTCBI for instance,2 or too strong phase separation, such as sexithiophene:C60 and pentacene:C60 .73,75 In these cases, different processing methods should be applied to manipulate the degree of phase separation. In general, slow growth and post-annealing serve as ways to promote the phase separation,2,55 although solvent annealing and chemical additives have also been reported

440

Figure 18. [(a), (c), and (e)] Tapping mode atomic force microscope (AFM), and [(b), (d), and (f)] Scanning electron microscope (SEM) images of 50 nm thick pentacene:C60 = 1:1 [(a) and (b)], 1:4 [(c) and (d)], and 1:5.5 [(e) and (f)] deposited at 6 Å/s (total rate) on a Si substrate. The scale bare in the SEM images is 1µm, and the scanning area in the AFM images is 5 × 5 µm2 while the height scale is 300 nm.

Organic Photovoltaic Cells

441

for solution processable polymer OPVs to enhance phase separation.16,57 On the other hand, fast growth on cool substrates could suppress the phase separation in the mixtures. Further enhancing the photovoltaic performance based on bulk HJs will require a combination of molecular design and optimized processing conditions.

4. Controlled Bulk Heterojunctions Although controlling the phase separation in DA bulk HJs serves as an effective way to achieve better OPV performance, there are still some limitations for the bulk HJs achieved in such fashion, one of which is that the random interpenetrating DA network obtained from phase separation could lead to poor charge transport due to charge trapping at the bottlenecks or cul-de-sacs along the conducting paths. Another limitation is the lack of control over morphology in the phase separated DA bulk HJs. As has been shown in the previous section, the degree of phase separation strongly relies on the strength of the interaction between the molecules as well as the processing conditions. One way to overcome the above limitations in DA bulk HJs is to produce a bulk HJ composed of nanoscale interdigitated DA phase as indicated in Fig. 19. Such structure can secure both efficient exciton diffusion and charge collection if the lateral dimension is about the same as the exciton diffusion lengths. Since the charge transport within a pure phase is superior to that in a mixture, thicker films can be used to increase the overall light absorption without compromising the device fill factor. Here, two different methods to fabricate such interdigitated bulk HJ are reviewed and their effects on improving OPV cell performance are discussed. 4.1 Controlled Bulk Heterojunction Grown by Organic Vapor Phase Deposition Different from conventional vacuum thermal evaporation (VTE) systems widely used for the growth of organic thin films,4,23 organic vapor phase deposition (OVPD) is carried out under low vacuum and with a carrier gas.24,25 In an OVPD system, the organic materials placed in crucibles are sublimated by heating under the reduced pressure (0.1∼1 Torr). The organic vapors are picked up by an inert carrier gas, such as dry nitrogen, and are transported to a cold substrate by directing the flow patterns. Deposition is intentionally controlled only on the substrate by keeping the chamber wall at above the gas-to-solid condensation temperature. OPVD has been employed to fabricate organic light emitting diodes,24 OPV cells,14,76 and organic thin-film transistors.49,77

Figure 19. Schematic of ideal interdigitated bulk heterojunction, where the lateral dimension of the donor and acceptor phase is close to the exciton diffusion length. The presence of straight conducting path for both electrons and holes secure efficient charge collection.

442

Y. Zheng and J. Xue

Figure 20. SEM and AFM images of CuPc films. a, Surface of a CuPc film with a continuous wetting layer plus short needle-like crystals grown on a silicon surface by organic vapor phase deposition (OVPD). b, Surface of a CuPc film grown on ITO-coated glass substrate under the same grown conditions as the film in a. c, Surface of CuPc film grown on top of ITO-coated glass by OVPD. This structure containing a very high density of the protrusions is suitable as the bottom layer of a bulk heterojunction. Inset: AFM image of the same sample. d, Surface of a 500-Å-thick CuPc film deposited on ITO glass by VTE. Reprinted with permission from Ref. 14. Copyright 2005, Nature.

Yang et al. found that by controlling the growth conditions in OVPD, organic thin films with a variety of morphologies can be achieved.14,76 Among them, nanostructured thin films comprise of elongated organic wires which can be used to create interdigitated bulk HJ and show a very promising result in improving OPV cell performance. Figure 20 compares the SEM images of various nanostructured CuPc films grown with OVPD with that grown with VTE.14 Morphologies significantly different from the VTE-grown CuPc film (Fig. 20(d)) were achieved with OVPD, and the film grown on ITO-coated glass (Figs. 20(b) and (c)) and the one grown on Si (Fig. 20(a)) show significantly different morphologies. With the same deposition time, needle-like CuPc crystals with a length of a few micrometers were grown on ITO (Fig. 20(b)), whereas shorter and denser pillar shape CuPc crystals were observed on Si (Fig. 20(a)). However, the size of the above features is too large for efficient interdigitated bulk HJs. As shown in Fig. 20(c), further tuning the processing conditions in the OVPD allows the creation of a highly folded surface with short “protrusions” suitable for bulk HJs. Referred from the AFM images shown in the insets, the average peak-to-valley height is 35 nm in the sample shown in Fig. 20(c), which is much higher than that grown with VTE (average peak-to-valley height of 3 nm).14 The XRD measurements have shown that the OVPD grown CuPc layer and the VTE grown planar CuPc film have very similar crystallinity.14 With the obtained nanostructured CuPc nanorod films, filling acceptor molecules into the spacing between the donor protrusions could be challenging in the conventional VTE. The ballistic trajectory of the incident molecules in VTE induces a strong shadowing effect, leading to voids and an unfilled region in the shadows created by the protrusions. In the

Organic Photovoltaic Cells

443

Figure 21. Typical current-voltage characteristics of the bulk heterojunction (BHJ) device in comparison with a planar heterojunction (PHJ) produced by OVPD, PHJ by VTE, and BHJ by annealing at incident power level of 1 sun using AM 1.5G simulated solar irradiation. Reprinted with permission from Ref. 14. Copyright 2005, Nature.

OPVD case, however, the molecules have much shorter mean free paths due to the higher chamber pressure, and the molecules arrive at the surface from random directions upon scattering with carrier gas molecules, which not only fill up the gaps but also smooth the top surface. Device performance based on the CuPc/PTCBI interdigitated bulk HJ is compared with that of the planar HJ device as well as the device with annealed CuPc:PTCBI bulk HJ. Figure 21 shows the J—V characteristics of the devices under 1 sun simulated AM 1.5 G illumination.14 With similar VOC , the controlled bulk HJ device shows much improved JSC = 11 mA/cm2 and ηP as high as 2.7%, which is nearly three times higher than the planar HJ device. Table 2 summarizes the photovoltaic performance of the four devices under study.14 The two planar HJ devices grown using the VTE and OVPD method have very similar performance. The annealing of the bulk HJ show improved JSC and ηP = 1.4% was achieved, which is mainly attributed to the improved exciton diffusion. However, the high series resistance of the annealed bulk HJ device suggests the poor charge transport within the mixed layer, leading to the lowest FF among the four devices. The interdigitated bulk HJ device shows the highest JSC and ηP . More importantly, the series resistance is significantly lower than the planar HJ and annealed bulk HJ device, which indicates that the interdigitated DA structure provides good charge transport across the film. Thus, FF

Table 2 Comparison of performance of several ITO/CuPc/PTCBI/BCP/Ag photovoltaic cells

Planar VTE HJ Annealed Bulk HJ Planar OVPD HJ Controlled Bulk OVPD HJ

J sc (mA/cm2)

Voc (V)

FF

ηp (%)

RSA ( cm2)

6 9 5 11

0.49 0.50 0.48 0.49

0.49 0.40 0.47 0.58

1.0 ± 0.1 1.4 ± 0.1 1.1 ± 0.1 2.7 ± 0.1

30 ± 10 60 ± 10 18.2 ± 0.5 2.2 ± 0.1

444

Y. Zheng and J. Xue

Figure 22. Schematics of oblique angle deposition: (a) Nuclei distributed across the surface lead to ballistic shadowing of the surrounding regions during oblique angle deposition. The incident angle is α. (b) Restricted growth on tops of the nuclei induces formation of one-dimensional nanostructure. The nanorods will grow oriented toward the source, forming an angle β with the substrate normal with β < α.

as high as 0.58 can be achieved. These data suggest that a controlled formation of bulk HJs with interdigitated DA phase is effective in securing both good exciton diffusion and charge collection for an efficient OPV cell.

4.2 Oblique Angle Deposition Grown Nanostructure Besides OVPD, oblique angle deposition (OAD) has also been demonstrated as a versatile way to fabricate one-dimensional nanostructures, which could then be utilized to produce interdigitated bulk HJs with controlled morphology.78–80 OAD, which is also known as glancing angle deposition, has been widely explored to grow inorganic nanostructures for memory, photovoltaic, and sensor application.80–85 Figure 22 schematically illustrates the OAD process. Unlike conventional VTE processes in which the ballistic molecular beams arrive at the substrate surface from near-normal directions, the molecular beams in the OAD process arrive at the surface with a large angle, α, from the substrate normal. Due to the self-shadowing effect by the deposited molecules and the limited diffusion of adsorbed molecules on the surface, nanorod arrays with various morphologies can be obtained by controlling the deposition conditions, such as the incident angle, molecule surface diffusivity, etc.80–85 Figure 23 shows a series of SEM and AFM images of OAD-grown CuPc films with nominal thicknesses of 100 nm with α ≈ 45◦ , 55◦ , and 80◦ . A corrugated surface with CuPc “bumps” ranging from 20 nm to 50 nm in size can be observed in the film grown at α = 45◦ . The roughness of Rrms = 9.5 nm for this film is not much greater than that of planar CuPc films (Rrms = 4.8 nm).63 However, as the incident angle increases, the features on the surface evolve from short “bumps” to elongated nanorods. This is accompanied by the increase in roughness from Rrms = 9.5 nm for α = 45◦ film to Rrms = 73 nm for α = 80◦ . CuPc nanorod arrays with nearly 200 nm length and 20 to 50 nm width are observed with α = 80◦ . The CuPc nanorod morphology can be further controlled by introducing substrate rotation during the OAD. Instead of creating shadows along one specific direction with a stationary substrate, the shadowing effect can be created in all orientations with a rotational substrate. Figures 24(a) and (b) show the cross-section SEM images of CuPc nanorods grown with substrate either stationary (Fig. 24(a)) or rotated at a speed of 5 rpm (Fig. 24(b)). With the same incident angle of α = 75◦ , the stationary substrate mode generates slanted nanorods with 20–50 nm in diameter, tilting towards the direction of the incoming molecular flux. The CuPc nanorods grown with rotational substrate, however, have diameters between 40 to 70 nm and are mostly upright. XRD patterns of the OAD

Organic Photovoltaic Cells

445

Figure 23. SEM and AFM images of 100 nm thick CuPc nanorod film fabricated with incident angle of α ≈ 45◦ [(a) and (d)], 55◦ [(b) and (e)], and 80◦ [(d) and (f)]. The scale bar is 100 nm in the SEM images, while the scanning area is 2 × 2 µm2 for the AFM images.

grown nanorod films show the same characteristic diffraction peak from α-CuPc phase as that of planar CuPc, suggesting the crystalline structure of the nanorods.80 Creation of interdigitated bulk HJ based on the OAD grown nanorods requires infiltration of the acceptor material into the CuPc nanorod arrays. Here, acceptor materials

Figure 24. [(a) and (b)] Cross-sectional scanning electron microscope (SEM) images of CuPc nanorods grown on a (a) stationary or (b) rotational substrate with speed of ∼ 5 rpm. [(c) and (d)] CuPc nanorod/PCBM composited films with PCBM concentration of (c) 15 mg/mL and (d) 30 mg/mL in the chlorobenzene solution. The scale bar is 100 nm for all four images.

446

Y. Zheng and J. Xue

Figure 25. Current-density—voltage (J—V) characteristics of four CuPc/PCBM photovoltaic cells under 1 sun AM 1.5 illumination: a bilayer cell (labeled as “Planar”) with a 30 nm thick flat CuPc film, two devices with CuPc nanorods grown on ITO with a stationary (“NR-S”) or rotational (“NR-R”) substrate, and a device with a 20 nm thick flat CuPc film followed by a nanorod film (“Planar-NR”). The inset schematically illustrates the device structure of the NR-R device.

of [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) were infiltrated into the gaps by simply spin-coating a chlorobenzene solution of PCBM onto the nanorod arrays. Figures 24(c) and (d) show cross-section SEM images of two CuPc nanorods/PCBM composite films with different PCBM loading. No obvious voids or pin-holes are observed in the composite films. With a high PCBM loading (Fig. 24(d)), the gaps between CuPc nanorods are completely filled, resulting in a smooth surface. When a lower PCBM loading solution was used, the amount of PCBM deposited was not sufficient to completely fill the gaps, leading to a corrugated surface as shown in Fig. 24(c). This also suggests that the CuPc nanorods still orient upright on the substrate and the spin-coating process does not damage the contact of the nanorods with the underlying ITO electrode. This is important for efficient hole collection in OPV devices. The performance of interdigitated bulk HJ OPV cells based on CuPc nanorod/PCBM is compared with that of the planar HJ device. Figure 25 shows the J—V characteristics for three different CuPc nanorods/PCBM devices along with that for a planar HJ device.80 The Table 3 Comparison of the open-circuit voltage (VOC ), short-circuit current density (JSC ), fill factor (FF), and power conversion efficiency (ηP ) of Planar, NR-S, NR-R and Planar-NR devices under 1 sun illumination Device Planar NR-S NR-R Planar-NR

VOC (V)

JSC (mA/cm2)

FF

ηP (%)

0.57 0.57 0.57 0.60

3.4 ± 0.2 4.4 ± 0.2 4.4 ± 0.2 5.6 ± 0.3

0.47 0.40 0.55 0.53

0.85 ± 0.05 0.95 ± 0.05 1.4 ± 0.1 1.8 ± 0.1

Organic Photovoltaic Cells

447

40–60 nm long CuPc nanorods in the BHJ devices were grown on ITO substrates with α = 75◦ . Here the CuPc nanorods obtained with stationary or rotational substrate are labeled as “NR-S” and “NR-R,” respectively. With the same VOC , both NR-S and NR-R device show higher JSC than the planar HJ device. The fill factor of the NR-S device, however, is much lower than that of the NR-R device, 0.40 versus 0.50. This could be attributed to the difference in the morphology of nanorod arrays (see Fig. 24). The tilted nanorods obtained using stationary substrate may present more difficulties for the complete infiltration with PCBM molecules than the mostly upright nanorods grown with a rotational substrate. Further enhancement of the photovoltaic performance can be achieved by inserting a thin planar CuPc layer between the nanorods and the ITO to further enhance the hole transport across the film and also provide additional absorption of the incident photons. This is similar to the PM-HJ device structure used in the CuPc:C60 cell as described earlier. This “Planar-NR” device with 20 nm thick planar layer and 40 nm long CuPc nanorod as donor layer exhibit the highest efficiency. Table 3 compares the photovoltaic performance of all four devices.80 The highest short-circuit current density of JSC = 5.6 mA/cm2 and reasonable good FF and VOC were achieved in the Planar-NR device. The maximum ηP = 1.8% achieved in the Planar-NR device is approximately twice that of the planar HJ device. The ηEQE as a function of the wavelength for the Planar and Planar-NR devices is compared in Fig. 26. The Planar-NR device shows a 60–70% higher ηEQE than the planar device in both the CuPc and PCBM absorption regions. This is attributed to the increased interfacial area between the donor and acceptor species due to the interdigitated structure. Interdigitated bulk HJ can be achieved by growing a nanostructure of one active material first and filling in a second active material into the gaps. Both OVPD and OAD demonstrated that such structure can be achieved. The demonstrated interdigitated bulk HJ devices show enhanced OPV performance than the corresponding planar or mixed HJ devices. Further improvement based on these devices will require better control of the growth process of the nanostructure to achieve feature sizes closer to the exciton diffusion length. Moreover, a careful selection of the appropriate processing conditions for infiltration of the second material into the nanostructure is also important for securing good device performance.

Figure 26. External quantum efficiency (EQE) as function of wavelength λ for Planar and Planar-NR devices.

448

Y. Zheng and J. Xue

5. Conclusions and Outlook In this review, we have described the operating principles of three different types of donoracceptor heterojunction structures and discussed the advantages and limitations for each structure. While in general the planar-mixed HJ structure combines the advantages of the planar HJ and mixed HJ structures, the effectiveness of the PM-HJ structure strongly depends on the charge transport properties of the mixed layer. We also show that the molecular properties as well as processing conditions could have strong effects on the phase separation and percolation in the donor-acceptor mixture, which in turn have strong influences on the exciton diffusion and charge transport properties of the mixed layer. Furthermore, we show that, compared with randomly formed mixed HJs, employing a twostep approach to form a controlled, nanostructured bulk HJ could be more advantageous for charge transport with pure domains of donor or acceptor materials oriented vertically (in the thickness direction). The control of the lateral domain size is, however, critical to the high performance of such bulk HJ structures. To further improve the efficiency of molecular OPV cells, several critical issues need to be addressed, particularly the low open-circuit voltage and insufficient coverage of the solar spectrum. Using the example of the CuPc/C60 DA system whose photoresponse is in the range of 400–750 nm (see Fig. 1) we see that an open-circuit voltage of 0.5–0.6 V is typically achieved at 1 sun illumination (see Fig. 8(c)). This suggests that the conversion of an absorbed photon (with an energy of 1.7 to 3.1 eV) to an electron-hole pair collected at the electrodes (with an energy of 0.5–0.6 eV) suffers on average 75% loss in energy, which is certainly very substantial. A significant part of the energy loss may come from the energy level offsets at the DA HJ, which is necessary to induce efficient exciton dissociation. The HOMO and LUMO offsets at the CuPc/C60 interface are both approximately 1 eV.46 It appears that lowering the HOMO level of the donor molecule, thus reducing the HOMO offset with the acceptor (C60 ), could lead to a reduction of the energy loss. Indeed, it has been shown that using a subphthalocyanine (subPc) to replace CuPc, a VOC value as high as 1 V could be obtained, which was attributed to the 0.4 eV lower HOMO level of the subPc than that of the CuPc (therefore 0.4 eV smaller HOMO level offset with C60 ).86 Similarly, electron withdrawing dicyanovinyl groups have been used to cap a terthiophene molecule to lower its HOMO level, which resulted in VOC ≈ 1 V in a bulk HJ cell with C60 as the acceptor.87 However, energy level alignment is not the only parameter affecting VOC . As shown in Fig. 8(c), even with the same DA system (CuPc/C60 ), we see that VOC could range anywhere from 0.25 to 0.6 V, depending on the specific HJ structure and the illumination intensity. With VOC being the voltage at which the photocurrent and the dark current cancel each other, clearly an increase of VOC could be achieved by reducing the dark current without compromising the photocurrent, which is the case for mixed HJs and PM-HJs as compared to planar HJ cells. Li et al. reported a significant enhancement in VOC for a phthalocyanine/C60 cell by inserting electron blocking layers between the active layer and the anode to reduce the electron leakage current to the anode.88 However, in general the exact origin of the dark current in these molecular HJ cells is not very well understood. More work in this direction certainly is necessary, and so are the means to reduce the dark current with new materials and/or device structures. The other major limiting factor to the performance of these OPV cells is the insufficient coverage of the solar spectrum. The photoresponse of CuPc/C60 cells is in the spectral range of 400 to 750 nm, which means that most of the IR photons abundantly available in the solar spectrum are not harvested. Some near IR absorbing materials have been

Organic Photovoltaic Cells

449

applied to molecular PV cells, including certain phthalocyanines molecules such as tin and chloroaluminum phthalocyanines.89,90 However, cells containing these near IR absorbing materials have not outperformed CuPc-based devices. This is attributed to the limited absorption bandwidth of these organic materials, typically 200 nm (see Fig. 1), which means that the gain in the absorption of IR photons by the IR-absorbing materials is negated by the loss of absorption in the visible portion of the solar spectrum. A possible solution is to use the tandem cell architecture to stack two or more subcells vertically with each subcell absorbing in a specific but limited spectral range.10,13,91 Relying on the mostly complimentary absorption by CuPc and C60 (see Fig. 1), Xue et al. demonstrated asymmetric tandem cells in which both subcells consist of CuPc-C60 planar-mixed HJs but the front subcell contains more longer wavelength-absorbing CuPc whereas the back subcell contains more shorter wavelength-absorbing C60 .13 An example of using different DA material systems for the subcells was also provided recently by Cheyns et al., in which one subcell employs the subPc:C60 HJ (whose response is in the visible) and another subcell employs a HJ between subnaphthalocyanine:C60 with a photoresponse up to 750 nm.92 Finally, in addition to the obvious need to improve the device efficiency, achieving a long lifetime is also necessary for the commercialization of the OPV technology.1 Many of the molecules researchers have used in OPV cells have shown to have good thermal and photo-stability and there have been encouraging results on the lifetime of molecular OPV devices.93 However, significant future research is still needed to understand the degradation mechanisms in these devices and to eventually demonstrate long-lived devices with new materials and/or device structures. The manufacturability of these devices is another important point of study in the field.93 The premise for interests in the organic PV technologies lies in the potential of low-cost solar energy conversion. How to produce solar panels from these organic molecules in high throughput and with low cost, by vacuum thermal evaporation, vapor phase deposition, or some other methods, will ultimately dictate whether this technology can eventually reach the commercial market.

Acknowledgments The authors gratefully acknowledge financial support from the National Science Foundation, the US Department of Energy Solar Energy Technologies Program (SETP), and the Florida Energy Systems Consortium (FESC) for research on organic-based photovoltaics.

References 1. Shaheen, S. E.; Ginley, D. S.; Jabbour, G. E. “Organic-based photovoltaics toward low-cost power generation,” MRS Bull., 2005, 30, 10–19. 2. Peumans, P.; Yakimov, A.; Forrest, S. R. “Small molecular weight organic thin-film photodetectors and solar cells,” J. Appl. Phys., 2003, 93, 3693–3723. 3. Xue, J. “Perspectives on organic photovoltaics,” Polym. Rev., 2010, 50, 411–419. 4. Forrest, S. R. “The path to ubiquitous and low-cost organic electronic appliances on plastic,” Nature, 2004, 428, 911–918. 5. Tang, C. W. “2-layer organic photovoltaic cell,” Appl. Phys. Lett., 1986, 48, 183–185. 6. Halls, J. J. M.; Walsh, C. A.; Greenham, N. C.; Marseglia, E. A.; Friend, R. H.; Moratti, S. C.; Holmes, A. B. “Efficient photodiodes from interpenetrating polymer networks,” Nature, 1995, 376, 498–500. 7. Yu, G.; Gao, J.; Hummelen, J. C.; Wudl, F.; Heeger, A. J. “Polymer photovoltaic cells—enhanced efficiencies via a network of internal donor-acceptor heterojunctions,” Science, 1995, 270, 1789–1791.

450

Y. Zheng and J. Xue

8. Shaheen, S. E.; Brabec, C. J.; Sariciftci, N. S.; Padinger, F.; Fromherz, T.; Hummelen, J. C. “2.5% efficient organic plastic solar cells,” Appl. Phys. Lett., 2001, 78, 841–843. 9. Brabec, C. J.; Sariciftci, N. S.; Hummelen, J. C. “Plastic solar cells,” Adv. Funct. Mater., 2001, 11, 15–26. 10. Yakimov, A.; Forrest, S. R. “High photovoltage multiple-heterojunction organic solar cells incorporating interfacial metallic nanoclusters,” Appl. Phys. Lett., 2002, 80, 1667–1669. 11. Peumans, P.; Uchida, S.; Forrest, S. R. “Efficient bulk heterojunction photovoltaic cells using small-molecular-weight organic thin films,” Nature, 2003, 425, 158–162. 12. Gregg, B. A. “Excitonic solar cells,” J. Phys. Chem. B, 2003, 107, 4688–4698. 13. Xue, J.; Uchida, S.; Rand, B. P.; Forrest, S. R. “Asymmetric tandem organic photovoltaic cells with hybrid planar-mixed molecular heterojunctions,” Appl. Phys. Lett., 2004, 85, 5757–5759. 14. Yang, F.; Shtein, M.; Forrest, S. R. “Controlled growth of a molecular bulk heterojunction photovoltaic cell,” Nat. Mater., 2005, 4, 37–41. 15. Kim, J. Y.; Lee, K.; Coates, N. E.; Moses, D.; Nguyen, T. Q.; Dante, M.; Heeger, A. J. “Efficient tandem polymer solar cells fabricated by all-solution processing,” Science, 2007, 317, 222– 225. 16. Li, G.; Yao, Y.; Yang, H.; Shrotriya, V.; Yang, G.; Yang, Y. ““Solvent annealing” effect in polymer solar cells based on poly(3-hexylthiophene) and methanofullerenes,” Adv. Funct. Mater., 2007, 17, 1636–1644. 17. Kroon, R.; Lenes, M.; Hummelen, J. C.; Blom, P. W. M.; De Boer, B. “Small bandgap polymers for organic solar cells (polymer material development in the last 5 years),” Polym. Rev., 2008, 48, 531–582. 18. Hau, S. K.; Yip, H. L.; Baek, N. S.; Zou, J. Y.; O’Malley, K.; Jen, A. K. Y. “Air-stable inverted flexible polymer solar cells using zinc oxide nanoparticles as an electron selective layer,” Appl. Phys. Lett., 2008, 92, 253–301. 19. Matsuo, Y.; Sato, Y.; Niinomi, T.; Soga, I.; Tanaka, H.; Nakamura, E. “Columnar structure in bulk heterojunction in solution-processable three-layered p-i-n organic photovoltaic devices using tetrabenzoporphyrin precursor and silylmethyl[60]fullerene,” J. Am. Chem. Soc., 2009, 131, 16048–16050. 20. Liang, Y. Y.; Wu, Y.; Feng, D. Q.; Tsai, S. T.; Son, H. J.; Li, G.; Yu, L. P. “Development of new semiconducting polymers for high performance solar cells,” J. Am. Chem. Soc., 2009, 131, 56–57. 21. Liang, Y. Y.; Feng, D. Q.; Wu, Y.; Tsai, S. T.; Li, G.; Ray, C.; Yu, L. P. “Highly efficient solar cell polymers developed via fine-tuning of structural and electronic properties,” J. Am. Chem. Soc., 2009, 131, 7792–7799. 22. Chen, H. Y.; Hou, J. H.; Zhang, S. Q.; Liang, Y. Y.; Yang, G. W.; Yang, Y.; Yu, L. P.; Wu, Y.; Li, G. “Polymer solar cells with enhanced open-circuit voltage and efficiency,” Nat. Photon., 2009, 3, 649–653. 23. Forrest, S. R. “Ultrathin organic films grown by organic molecular beam deposition and related techniques,” Chem. Rev., 1997, 97, 1793–1896. 24. Shtein, M.; Gossenberger, H. F.; Benziger, J. B.; Forrest, S. R. “Material transport regimes and mechanisms for growth of molecular organic thin films using low-pressure organic vapor phase deposition,” J. Appl. Phys., 2001, 89, 1470–1476. 25. Shtein, M.; Peumans, P.; Benziger, J. B.; Forrest, S. R. “Micropatterning of small molecular weight organic semiconductor thin films using organic vapor phase deposition,” J. Appl. Phys., 2003, 93, 4005–4016. 26. Xue, J.; Forrest, S. R. “Carrier transport in multilayer organic photodetectors: I. Effects of layer structure on dark current and photoresponse,” J. Appl. Phys., 2004, 95, 1859–1868. 27. Eom, S. H.; Zheng, Y.; Chopra, N.; Lee, J.; So, F.; Xue, J. “Low voltage and very high efficiency deep-blue phosphorescent organic light-emitting devices,” Appl. Phys. Lett., 2008, 93, 133–309. 28. Forrest, S. R. “The limits to organic photovoltaic cell efficiency,” MRS Bull., 2005, 30, 28–32. 29. Rand, B. P.; Genoe, J.; Heremans, P.; Poortmans, J. “Solar cells utilizing small molecular weight organic semiconductors,” Prog. Photovolt., 2007, 15, 659–676.

Organic Photovoltaic Cells

451

30. Armstrong, N. R.; Wang, W. N.; Alloway, D. M.; Placencia, D.; Ratcliff, E.; Brumbach, M. “Organic/organic’ heterojunctions: organic light emitting diodes and organic photovoltaic devices,” Macromol. Rapid Comm., 2009, 30, 717–731. 31. Heremans, P.; Cheyns, D.; Rand, B. P. “Strategies for increasing the efficiency of heterojunction organic solar cells: material selection and device architecture,” Accounts Chem. Res., 2009, 42, 1740–1747. 32. Liang, Y.; Yu, L. “Development of semiconducting polymers for solar energy harvesting,” Polym. Rev., 2010, 50, 454–473. 33. Hau, S. K.; Yip, H.-L.; Jen, A. K.-Y. “Development of the inverted polymer solar cells,” Polym. Rev., 2010, 50, 474–510. 34. Xue, J.; Rand, B. P.; Uchida, S.; Forrest, S. R. “A hybrid planar-mixed molecular heterojunction photovoltaic cell,” Adv. Mater., 2005, 17, 66–71. 35. Xue, J.; Rand, B. P.; Uchida, S.; Forrest, S. R. “Mixed donor-acceptor molecular heterojunctions for photovoltaic applications. II. Device performance,” J. Appl. Phys., 2005, 98, 124903. 36. Pope, M.; Swenberg, C. E. Electronic Processes in Organic Crystals and Polymers; Oxford University Press: New York, 1999. 37. Hill, I. G.; Kahn, A.; Soos, Z. G.; Pascal, R. A. “Charge-separation energy in films of piconjugated organic molecules,” Chem. Phys. Lett., 2000, 327, 181–188. 38. Chandross, M.; Mazumdar, S.; Jeglinski, S.; Wei, X.; Vardeny, Z. V.; Kwock, E. W.; Miller, T. M. “Excitons in Poly(Para-Phenylenevinylene),” Phys. Rev. B, 1994, 50, 14702–14705. 39. Peumans, P.; Bulovic, V.; Forrest, S. R. “Efficient photon harvesting at high optical intensities in ultrathin organic double-heterostructure photovoltaic diodes,” Appl. Phys. Lett., 2000, 76, 2650–2652. 40. Rand, B. P.; Li, J.; Xue, J.; Holmes, R. J.; Thompson, M. E.; Forrest, S. R. “Organic doubleheterostructure photovoltaic cells employing thick tris(acetylacetonato)ruthenium(III) excitonblocking layers,” Adv. Mater., 2005, 17, 2714–2718. 41. Rim, S. B.; Zhao, S.; Scully, S. R.; McGehee, M. D.; Peumans, P. “An effective light trapping configuration for thin-film solar cells,” Appl. Phys. Lett., 2007, 91, 243501. 42. Agrawal, M.; Peumans, P. “Broadband optical absorption enhancement through coherent light trapping in thin-film photovoltaic cells,” Opt. Express, 2008, 16, 5385–5396. 43. Tvingstedt, K.; Dal Zilio, S.; Inganas, O.; Tormen, M. “Trapping light with micro lenses in thin film organic photovoltaic cells,” Opt. Express, 2008, 16, 21608–21615. 44. Hiramoto, M.; Suezaki, M.; Yokoyama, M. “Effect of thin gold interstitial-layer on the photovoltaic properties of tandem organic solar-cell,” Chem. Lett., 1990, 19, 327–330. 45. Rand, B. P.; Peumans, P.; Forrest, S. R. “Long-range absorption enhancement in organic tandem thin-film solar cells containing silver nanoclusters,” J. Appl. Phys., 2004, 96, 7519– 7526. 46. Peumans, P.; Forrest, S. R. “Very-high-efficiency double-heterostructure copper phthalocyanine/ C-60 photovoltaic cells,” Appl. Phys. Lett., 2001, 79, 126–128. 47. Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices; 3rd ed.; Wiley: New Jersey, 2007. 48. Xue, J.; Uchida, S.; Rand, B. P.; Forrest, S. R. “4.2% efficient organic photovoltaic cells with low series resistances,” Appl. Phys. Lett., 2004, 84, 3013–3015. 49. Shtein, M.; Mapel, J.; Benziger, J. B.; Forrest, S. R. “Effects of film morphology and gate dielectric surface preparation on the electrical characteristics of organic-vapor-phase-deposited pentacene thin-film transistors,” Appl. Phys. Lett., 2002, 81, 268–270. 50. Dimitrakopoulos, C. D.; Brown, A. R.; Pomp, A. “Molecular beam deposited thin films of pentacene for organic field effect transistor applications,” J. Appl. Phys., 1996, 80, 2501– 2508. 51. Butko, V. Y.; Chi, X.; Lang, D. V.; Ramirez, A. P. “Field-effect transistor on pentacene single crystal,” Appl. Phys. Lett., 2003, 83, 4773–4775. 52. Briseno, A. L.; Mannsfeld, S. C. B.; Ling, M. M.; Liu, S. H.; Tseng, R. J.; Reese, C.; Roberts, M. E.; Yang, Y.; Wudl, F.; Bao, Z. N. “Patterning organic single-crystal transistor arrays,” Nature, 2006, 444, 913–917.

452

Y. Zheng and J. Xue

53. Yoo, S.; Domercq, B.; Kippelen, B. “Efficient thin-film organic solar cells based on pentacene/C60 heterojunctions,” Appl. Phys. Lett., 2004, 85, 5427–5429. 54. Potscavage, W. J.; Yoo, S.; Kippelen, B. “Origin of the open-circuit voltage in multilayer heterojunction organic solar cells,” Appl. Phys. Lett., 2008, 93, 193–308. 55. Ma, W. L.; Yang, C. Y.; Gong, X.; Lee, K.; Heeger, A. J. “Thermally stable, efficient polymer solar cells with nanoscale control of the interpenetrating network morphology,” Adv. Funct. Mater., 2005, 15, 1617–1622. 56. Li, G.; Shrotriya, V.; Huang, J. S.; Yao, Y.; Moriarty, T.; Emery, K.; Yang, Y. “High-efficiency solution processable polymer photovoltaic cells by self-organization of polymer blends,” Nat. Mater., 2005, 4, 864–868. 57. Peet, J.; Kim, J. Y.; Coates, N. E.; Ma, W. L.; Moses, D.; Heeger, A. J.; Bazan, G. C. “Efficiency enhancement in low-bandgap polymer solar cells by processing with alkane dithiols,” Nat. Mater., 2007, 6, 497–500. 58. Tsuzuki, T.; Shirota, Y.; Rostalski, J.; Meissner, D. “The effect of fullerene doping on photoelectric conversion using titanyl phthalocyanine and a perylene pigment,” Sol. Energy Mater. Sol. Cells, 2000, 61, 1–8. 59. Uchida, S.; Xue, J.; Rand, B. P.; Forrest, S. R. “Organic small molecule solar cells with a homogeneously mixed copper phthalocyanine: C-60 active layer,” Appl. Phys. Lett., 2004, 84, 4218–4220. 60. Gebeyehu, D.; Maennig, B.; Drechsel, J.; Leo, K.; Pfeiffer, M. “Bulk-heterojunction photovoltaic devices based on donor-acceptor organic small molecule blends,” Sol. Energy Mater. Sol. Cells, 2003, 79, 81–92. 61. Sullivan, P.; Heutz, S.; Schultes, S. M.; Jones, T. S. “Influence of codeposition on the performance of CuPc-C-60 heterojunction photovoltaic devices,” Appl. Phys. Lett., 2004, 84, 1210–1212. 62. Myers, J. D.; Tseng, T.-K.; Xue, J. “Photocarrier behavior in organic heterojunction photovoltaic cells,” Org. Electron., 2009, 10, 1182–1186. 63. Rand, B. P.; Xue, J.; Uchida, S.; Forrest, S. R. “Mixed donor-acceptor molecular heterojunctions for photovoltaic applications. I. Material properties,” J. Appl. Phys., 2005, 98, 124902. 64. Sze, S. M. Physics of Semiconductor Devices; 2nd ed., Wiley: New York, 1981. 65. Paasch, G.; Lindner, T.; Scheinert, S. “Variable range hopping as possible origin of a universal relation between conductivity and mobility in disordered organic semiconductors,” Synth. Met., 2002, 132, 97–104. 66. Crandall, R. S. “Modeling of thin-film solar-cells—uniform-field approximation,” J. Appl. Phys., 1983, 54, 7176–7186. 67. Hiramoto, M.; Fujiwara, H.; Yokoyama, M. “3-layered organic solar-cell with a photoactive interlayer of codeposited pigments,” Appl. Phys. Lett., 1991, 58, 1062–1064. 68. Park, S. H.; Roy, A.; Beaupre, S.; Cho, S.; Coates, N.; Moon, J. S.; Moses, D.; Leclerc, M.; Lee, K.; Heeger, A. J. “Bulk heterojunction solar cells with internal quantum efficiency approaching 100%,” Nat. Photon., 2009, 3, 297–303. 69. Prabakaran, R.; Kesavamoorthy, R.; Reddy, G. L. N.; Xavier, F. P. “Structural investigation of copper phthalocyanine thin films using X-ray diffraction, Raman scattering and optical absorption measurements,” Phys. Stat. Solidi B, 2002, 229, 1175–1186. 70. Tse, S. C.; Cheung, C. H.; So, S. K. “Charge transport and injection in amorphous organic semiconductors,” In Organic Electronics: Materials, Processing, Devices and Applications; So, F., Ed.; Taylor and Francis: Boca Raton, FL, 2010. 71. Yasuda, T.; Tsutsui, T. “Organic field-effect transistors based on high electron and ambipolar carrier transport properties of copper-phthalocyanine,” Chem. Phys. Lett., 2005, 402, 395–398. 72. Mayer, A. C.; Lloyd, M. T.; Herman, D. J.; Kasen, T. G.; Malliaras, G. G. “Postfabrication annealing of pentacene-based photovoltaic cells,” Appl. Phys. Lett., 2004, 85, 6272–6274. 73. Zheng, Y.; Pregler, S. K.; Myers, J. D.; Ouyang, J. M.; Sinnott, S. B.; Xue, J. “Computational and experimental studies of phase separation in pentacene:C-60 mixtures,” J. Vac. Sci. Technol. B, 2009, 27, 169–179.

Organic Photovoltaic Cells

453

74. Dimitrakopoulos, C. D.; Malenfant, P. R. L. “Organic thin film transistors for large area electronics,” Adv. Mater., 2002, 14, 99–117. 75. Veenstra, S. C.; Malliaras, G. G.; Brouwer, H. J.; Esselink, F. J.; Krasnikov, V. V.; vanHutten, P. F.; Wildeman, J.; Jonkman, H. T.; Sawatzky, G. A.; Hadziioannou, G. “Sexithiophene-C-60 blends as model systems for photovoltaic devices,” Synth. Met., 1997, 84, 971–972. 76. Yang, F.; Shtein, M.; Forrest, S. R. “Morphology control and material mixing by high-temperature organic vapor-phase deposition and its application to thin-film solar cells,” J. Appl. Phys., 2005, 98, 014906. 77. Rolin, C.; Steudel, S.; Myny, K.; Cheyns, D.; Verlaak, S.; Genoe, J.; Heremans, P. “Pentacene devices and logic gates fabricated by organic vapor phase deposition,” Appl. Phys. Lett., 2006, 89, 203502. 78. Zheng, Y.; Bekele, R.; Ouyang, J.; Xue, J. “Interdigitated bulk heterojunction organic photovoltaic cells with aligned copper phthalocyanine nanorods,” IEEE J. Sel. Top. Quant. Electron., 2010 (in press). 79. Li, N.; Forrest, S. R. “Tilted bulk heterojunction organic photovoltaic cells grown by oblique angle deposition,” Appl. Phys. Lett., 2009, 95, 123309. 80. Zheng, Y.; Bekele, R.; Ouyang, J. M.; Xue, J. “Organic photovoltaic cells with vertically aligned crystalline molecular nanorods,” Org. Electron., 2009, 10, 1621–1625. 81. Hawkeye, M. M.; Brett, M. J. “Glancing angle deposition: Fabrication, properties, and applications of micro- and nanostructured thin films,” J. Vac. Sci. Technol. A, 2007, 25, 1317–1335. 82. Robbie, K.; Sit, J. C.; Brett, M. J. “Advanced techniques for glancing angle deposition,” J. Vac. Sci. Technol. B, 1998, 16, 1115–1122. 83. Xi, J. Q.; Schubert, M. F.; Kim, J. K.; Schubert, E. F.; Chen, M. F.; Lin, S. Y.; Liu, W.; Smart, J. A. “Optical thin-film materials with low refractive index for broadband elimination of Fresnel reflection,” Nat. Photon., 2007, 1, 176–179. 84. Kolmakov, A.; Moskovits, M. “Chemical sensing and catalysis by one-dimensional metal-oxide nanostructures,” Ann. Rev. Mater. Res., 2004, 34, 151–180. 85. Zhao, Y. P.; Ye, D. X.; Wang, G. C.; Lu, T. M. “Novel nano-column and nano-flower arrays by glancing angle deposition,” Nano Lett., 2002, 2, 351–354. 86. Mutolo, K. L.; Mayo, E. I.; Rand, B. P.; Forrest, S. R.; Thompson, M. E. “Enhanced open-circuit voltage in subphthalocyanine/C-60 organic photovoltaic cells,” J. Am. Chem. Soc., 2006, 128, 8108–8109. 87. Uhrich, C.; Schueppel, R.; Petrich, A.; Pfeiffer, M.; Leo, K.; Brier, E.; Kilickiran, P.; Baeuerle, P. “Organic thin-film photovoltaic cells based on oligothiophenes with reduced bandgap,” Adv. Funct. Mater., 2007, 17, 2991–2999. 88. Li, N.; Lassiter, B. E.; Lunt, R. R.; Wei, G.; Forrest, S. R. “Open circuit voltage enhancement due to reduced dark current in small molecule photovoltaic cells,” Appl. Phys. Lett., 2009, 94, 023307. 89. Bailey-Salzman, R. F.; Rand, B. P.; Forrest, S. R. “Near-infrared sensitive small molecule organic photovoltaic cells based on chloroaluminum phthalocyanine,” Appl. Phys. Lett., 2007, 91, 013508. 90. Rand, B. P.; Xue, J.; Yang, F.; Forrest, S. R. “Organic solar cells with sensitivity extending into the near infrared,” Appl. Phys. Lett., 2005, 87, 233508. 91. Maennig, B.; Drechsel, J.; Gebeyehu, D.; Simon, P.; Kozlowski, F.; Werner, A.; Li, F.; Grundmann, S.; Sonntag, S.; Koch, M.; Leo, K.; Pfeiffer, M.; Hoppe, H.; Meissner, D.; Sariciftci, N. S.; Riedel, I.; Dyakonov, V.; Parisi, J. “Organic p-i-n solar cells,” Appl. Phys. A, 2004, 79, 1–14. 92. Cheyns, D.; Rand, B. P.; Heremans, P. “Organic tandem solar cells with complementary absorbing layers and a high open-circuit voltage,” Appl. Phys. Lett., 2010, 97, 243–310. 93. Krebs, F. Polymeric Solar Cells: Materials, Design, Manufacture; DEStech Publications, Lancaster, PA, 2010.

Polymer Reviews, 50:454–473, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.515765

Development of Semiconducting Polymers for Solar Energy Harvesting YONGYE LIANG AND LUPING YU Department of Chemistry and James Franck Institute, The University of Chicago, 929 E 57th Street, Chicago, IL 60637 Semiconducting polymer solar cells are an attracting class of devices for low-cost solar energy harvesting. The bulk hetero-junction structure based on composite materials of semiconducting polymer donor and fullerene acceptor is an effective form of active layers for polymer solar cells. So far, the limiting factors for widespread, practical applications in polymers solar cell is their low power conversion efficiency (PCE) and potential instability under light exposure. Thus new polymeric materials with desired properties and stability are crucial for improving the solar cell performance. Numerous conjugated polymers, such as poly[phenylene vinylene]s (PPVs) and polythiophenes, have been explored for this purpose, which lead to PCE as high as 5%. To improve the performance, low bandgap polymers and polymers with low lying HOMO energy levels have been the subject of recent focus. Efficiencies close to 8% have been achieved in the polymer system composed of thieno[3,4-b]thiophene and benzodithiophene alternating units (PTB). The high efficiency is due to the synergistic combinations of desired properties in the polymer system through detailed fine-tuning of the polymer structure. The recent results reaffirmed the notion that better solar cell polymers could be further developed for vital applications in real devices. Keywords semiconducting polymer, solar cell, bulk hetero-junction, fine-tuning

1. Introduction In the recognition of energy demand increase and fossil fuel depletion, as well as the environmental pollution from fuel combustion, the search and utilization of clean and renewable energies is becoming one of the greatest challenges for our society. Solar energy is the largest renewable energy source, which can potentially provide about 124 PW (PW = 1015 Watts) energy globally, more than 8000 times of the total worldwide energy consumption in 2004 (15 TW, 1TW = 1012 Watts).1 The direct conversion of solar radiation into electricity, called photovoltaic, is a simple and clean way to harness such a vast energy source. The photovoltaic effect requires active semiconducting materials, and currently inorganic solar cell based on silicon is the dominant technology. It exhibits good performance in PCE and lifetime. However, the high production and installation cost of silicon solar cells limits its widespread use to provide a large fraction of our electricity.2,3 New photovoltaic (PV)

Received May 7, 2010; accepted August 5, 2010. Address correspondence to Professor Luping Yu, 929 E 57th Street, Chicago, IL 60637, USA. E-mail: [email protected]

454

Developing Semiconductor Polymers for Solar Energy Harvesting

455

systems exhibiting low cost and high efficiency for solar energy conversion are actively pursued worldwide.4 An alternative material is the semiconducting polymers. Solar cells based on organic semiconductors enjoy several unique advantages compared to their inorganic counterparts.5 First of all, polymer semiconductors are soluble in organic solvents and the composite active layer of solar cells can be fabricated from solution in a single step by a variety of simple techniques such as spin coating, inkjet printing, and roller casting, and all are cost-effective and easier than the inorganic ones. The physical properties of organic solar cells, such as absorption spectrum and charge transport property, can be tailored by chemical synthesis. Meanwhile, the thickness of active layer can be very thin (about 100 nm) due to their high absorption coefficient within their absorption range so that organic solar cell can be semitransparent and the color can be tuned. Furthermore, the lightweight and mechanical flexibility of polymer materials enable some specific application of organic solar cells, such as in portable devices. The donor/acceptor concept pioneered by Tang in 19866 and bulk-heterojunction (BHJ) developed by Yu et al. in 19957 led to the active pursuit of polymer solar cells. The solution processed BHJ solar cells are easy to prepare and present a high density of heterojunction interfaces that enables the efficient exciton dissociation and charge generation over the whole active layer, which favors high PCE. These two advantages establish the solution processed BHJ solar cells as the most promising technology for organic solar cells. Gradual progress in the solar cell performance is underway based on this architecture. PCE evolved from 1% in 1995, to 5% in 2005 and close to 8% most recently.8 Despite the envisioned advantages and recent technology advances, so far polymer solar cells are still inferior to inorganic counterparts in terms of PCE and life times.9 Solar cells with PCE over 10% is necessary for widespread application.10 Theoretical calculations have shown that the polymer/fullerene BHJ solar cell in a single layer configuration can achieve more than 10% in PCE.11 After an exhaustive effort in morphology control, electrode modification, the introduction of dielectric layers, and the utilization of the plasmonic effect, it is clear that the most important limiting factor for further improvement on the photovoltaic performance is the materials in active layers, especially the polymer donor material.12,13 In this article, we will briefly review research progress in the development of new semiconducting polymers for solar energy harvesting.

2. Materials for BHJ Organic Solar Cells There are two components in the BHJ structure, the donor and the acceptor. The selection of the materials for both components is very important for solar cell performance. Photovoltaic properties of several representative solar cell systems with different donor/acceptor are outlined in Table 1. Owing to their strong absorbing ability, good film-forming ability and unique electronic properties from the long pi-conjugated system, semiconducting polymers are good candidates as donor component. A potential drawback is that conjugated polymers exhibit a broad distribution in molecular weight, which is easy to cause purification difficulty and poor reproducibility due to batch-to-batch difference. Small conjugated molecules offer the advantages of mono-dispersity and easy purification, so they are actively searched for solar cell applications. However, the film forming ability from solution and the charge mobility in the blend structure of small molecules are not as good as polymers. As a result, so far, the PCE of the solution processed BHJ solar cells using small molecules donors is much lower than the polymer counterpart.14,15

456

Y. Liang and L. Yu

Table 1 Photovolatic properties of representative BHJ solar cell systems with different donor/ acceptor materials. Some structures are showed in Fig. 1 Type Polymer/Fulleride Small molecule/ Fulleride Polymer/Polymer Polymer/Inorganic nanocrystals

Donor materials

Acceptor materials

Voc (V)

PTB7 Squaraine 1 DH6TDPP P3HT P3HT MDMOPPV

PC71 BM PC61 BM PC61 BM F8TBT CdSe ZnO

0.74 0.62 0.67 1.15 0.7 0.84

Jsc FF PCE (mA/cm2) (%) (%) Reference 14.5 5.7 8.42 3.6 5.7 2.40

69 35 45 34 40 59

7.40 1.24 2.33 1.20 1.70 1.60

74 12b 12a 14c 17 18

Note: Voc = open-circuit voltage, Jsc = short-circuit current density, FF = fill factor.

Thus far, fullerene derivatives are found to be the best candidates as acceptor component due to their high electron affinity and superior electron mobility. Especially, the threedimensional structure of fullerene offers unique packing ability in blend, which can form the electron transport channels efficiently. The original fullerenes do not have enough solubility in organic solvents, so fullerene derivatives with solublizing groups are usually used.16 Introduction of the functional group usually improves the miscibility with the donor components, but with small impact on the electronic properties of fullerene. As a result, fullerene derivatives BHJ structure can provide the advantages of efficient exciton dissociation by intermixing the donor/acceptor phases well enough and efficient extraction of the separated charges from the interface by high electron mobility. The weakness of fullerene derivatives as acceptor materials in BHJ solar cells lies in their weak absorption in visible region (like C60 ) and expensive materials cost (like C70 ). N-type polymers as acceptor materials offer the advantages of low cost and high optical absorption. The absorption of the acceptor polymers can be tuned so that both the donor and the acceptor can cover complementary parts of the solar spectrum. However, the efficiency obtained from polymer acceptor solar cells is much lower than that from the fullerene derivatives.17–19 It has been reported recently that the low efficiency in polymer/polymer BHJ is due to the limited transport of the separated charges.20 To overcome this problem, polymers with high local electron mobility are needed. N-type semiconducting inorganic nanocrystals were also used as the acceptor components. Such nanocrystals provide high electron mobility and high optical absorption coefficient.21 CdSe nanocrystals22 and ZnO23 nanocrystals have been used to form polymer/inorganic hybrid BHJ solar cells. However, the achieved efficiencies are rather low. Problems associated with the inorganic nanocrystals are their poor solubility in organic solvents and the lack of connectivity between nanocrystals due to the existence of stabilizing bulky ligands .

3. Conjugated Polymer/Fulleride based BHJ Solar Cells The most effective composite structure in solution processed BHJ organic solar cells is based on conjugated polymer/fullerene BHJ structure, in which electron rich semiconducting polymer plays as donor and fullerene plays as acceptor. PPV/PC61 BM and P3HT/PC61 BM (Fig. 1) are two well-studied systems, which lead to the current understanding of the factors affecting the device performance of BHJ solar cells.

Developing Semiconductor Polymers for Solar Energy Harvesting

457

Figure 1. Structures of widely used conjugated polymers and fullerene derivative and some structures mentioned in Table 1 for BHJ solar cells.

3.1 PPV/PC61 BM BHJ Solar Cells The first BHJ solar cell was reported by Yu et al. in 1995 based on MEH-PPV//[6,6]-phenylC61 -butyric acid methyl ester (PC61 BM) composite.7 After several years of optimization, the efficiency of devices based on this system was still around 1% under AM 1.5 illumination. In 2001, Shaheen et al. reported that a 2.5% efficiency was obtained from a better processable composite of poly[2-methoxy-5-(3 ,7 -dimethyl- octyloxy)-1,4-phenylene vinylene] (MDMO-PPV) and PC61 BM by controlling the casting condition.24 The active layer is prepared by spincoating MDMO-PPV/PCBM in 1:4 weight ratio from chlorobenzene. The device has a Voc of 0.82 V, a Jsc of 5.25 mA/cm2, and an FF of 0.61, which leads to a PCE of 2.5%. There is about a three-fold enhancement in efficiency compared to the one spincoating from toluene. The performance difference is attributed to the change of active layer morphology. PCBM has better solubility in chlorobenzene than toluene, which increases the miscibility of the two components. A more homogenous and bicontinuous composite was observed in the chlorobenzene-casting film compared to the toluene-casting film. Therefore, the heterojunction interface increases for excitons dissociation and more biocontinuous network forms for charge transport, which attributes to the performance enhancement. The performance of MDMO-PPV/PC61 BM system can be further optimized to about 3% by using LiF as interlayer.25 Another performance improvement was achieved by substituting PC61 BM to PC71 BM. PC71 BM has better absorption in the visible region than PC61 BM, and as a result, the MDMO-PPV//[6,6]-phenyl-C71 -butyric acid methyl ester (PC71 BM) system gives a larger Jsc, and about 3% in efficiency was achieved.26 However, PPV materials can only harvest light below 600 nm due to their relatively large bandgap (2.2 eV) and they usually have low hole mobility, which limits the further improvement of such solar cells.

458

Y. Liang and L. Yu

3.2 Poly(3-alkylthiophene)/PC61 BM BHJ Solar Cells Poly(3-alkylthiophenes) are good alternatives to PPV for photovoltaic application. They have a bandgap around 1.9 eV, which can harvest sunlight from 350 nm to about 650 nm in film. The alkyl side chains make them very soluble in organic solvents. The regioregular poly(3-alkylthiophenes) have good interchain packing ability, offering hole mobility up to 0.1 cm2/Vs.27 Poly (3-hexylthiophene) (P3HT) is a generally used polymer in poly(3alkylthiophenes). However, low efficiency was obtained from P3HT/PC61 BM simple solar cells without any post-treatment in the early years. In 2003, Padinger et al. reported that about 3.5% can be achieved by P3HT/PC61 BM system by post thermal annealing with external bias, which is about ten fold increase in efficiency compared to the untreated system (0.4%).28 This thermal annealing effect attracted great attention and raised further extensive studies on thermal annealing for optimization of the P3HT/PC61 BM system, and about 4–5% device efficiencies were obtained in subsequent years.29,30,31 Generally, the optimized ratio of P3HT/PC61 BM is about 1:1 by weight and chlorobenzene or dichlorobenzene is used as solvent. Thermal annealing is usually taken at between 100 to 150∞C for about 5–60 mins. The detailed conditions on thermal annealing of P3HT/PCBM solar cells vary from report to report, which may be due to other different experimental parameters used. Similar changes on device characteristics upon annealing are observed: Voc decreases slightly to about 0.6 V, while Jsc and FF significantly increases to about 10 mA/cm2 and over 0.6, respectively. The better device performance of the annealed P3HT/PC61 BM is attributed to the increase of P3HT crystallinity in blend. When P3HT/PC61 BM active layer film forms from solution, both components are intimately mixed32 and the configuration and planarity of the P3HT chain are distorted by PC61 BM. Thermal annealing allows the growth of P3HT crystalline, which leads to the enhancement of long wavelength absorption and hole mobility. The growth of P3HT crystalline is accompanied by the aggregation of PC61 BM, and to a certain extent, a better bicontinuous interpenetrating network is formed for the charge transport.33 Such morphology change enhances Jsc and FF, which accounts for the efficiency increase. Besides thermal annealing, other methods have been developed to control the morphology of P3HT/PC61 BM in order to improve the efficiency. Li et al. proposed a “solventannealing” method.34 Through controlling the solvent evaporation time during film forming of the P3HT/PC61 BM, the ordering of P3HT is controlled.35 The performance of the device prepared from slow drying without thermal annealing is comparable to the device prepared from thermal annealing. Complying both the solvent annealing and thermal annealing, an efficiency of 4.4% can be achieved in P3HT/PC61 BM systems. Moule and Meerholz proposed a solvent mixing method.36 By adding a small ratio of dipolar solvent (nitrobenzene) in the P3HT/PC61 BM solution in chlorobenzne, the aggregation of P3HT increases in solution and it can be brought to film. The P3HT/PC61 BM devices made from such mixed solvents can achieve up to 4% efficiency without thermal annealing. Other methods, like the addition of PC61 BM soluble addictive37,38 or the growth of P3HT nanofibers,39 are also reported to improve P3HT/PC61 BM efficiency by controlling the morphology. P3HT/PC61 BM device performance is highly dependent on the processing condition, an external factor. In another side, internal factors, like P3HT regioregularity (RR), molecular weight, also have great influence. RR is defined by the percentage of head-to-tail linkages in the polymer chain. Kim et al. reported the study of P3HT/PC61 BM solar cells with different P3HT RR and showed that increasing the RR of P3HT results in device performance improvement due to the higher crystalline in polymer with higher RR.40 Sivula et al. reported that certain degree of regioirregularity in P3HT can improve the stability of solar

Developing Semiconductor Polymers for Solar Energy Harvesting

459

Table 2 Photovoltaic properties of some PPV/fulleride or P3HT/fulleride BHJ solar cells Entry 1 2 3 4 5 6 7 8 9

Materials PPV/PC61 BM PPV/PC71 BM P3HT/PC61 BM P3HT/PC61 BM P3HT/PC61 BM P3HT/PC61 BM P3HT/PC61 BM P3HT/PC61 BM (TiO2 ) P3HT/PC61 BM (NiO)

Voc (V) Jsc (mA/cm2) FF (%) PCE (%) Reference 0.82 0.77 0.55 0.61 0.63 0.65 0.61 0.61 0.64

5.25 7.6 8.5 10.6 9.5 11.1 10.6 11.1 11.3

61 51 60 62 68 54 67 66 69

2.5 3.0 3.5 4.0 5.0 4.9 4.4 5.0 5.0

19 21 23 24 (a) 24 (b) 24 (c) 27 35 36

cell devices.41 As far as molecular weight is concerned, a good system requires a rather broad polydispersity with a proper combination of low molecular weight and high molecular weight portion, so that a highly crystalline region formed by low molecular weight can be interconnected by the high molecular weight polymer chains.42 Some device techniques have also been explored to improve the performance of P3HT/PC61 BM systems. Kim et al. introduced a solution processed TiO2 layer between the active layer and aluminum electrode.43 Such TiO2 can act as an optical spacer, which increases the absorption of reflective light from Al electrode in the active layer, and at the same time act as a hole-block layer, which facilitates the electron injection to the Al electrode. About 5% can be achieved in this system. Irwin et al. reported the replacement of PEDOT:PSS layer into a thin NiO layer. Such replacement improves the Voc and Jsc of the P3HT/PC61 BM device, leading to a PCE of about 5.0% .44 The P3HT/PC61 BM system was the dominant system for BHJ organic solar cells in the last 6 years. After an exhaustive research effort, the maximum EQE of P3HT/PC61 BM solar cells can reach to over 70%, which is comparable to inorganic counterparts. However, the highest overall power conversion efficiencies reported so far based on this system is just around 5%, still low for commercial application. The low efficiency is mainly due to the limited spectral absorption range of polymer active layer (up to 650 nm) and too large energy offset between P3HT and PC61 BM LUMO energy level, which can cause energy waste. As a result, new materials exhibiting better performances are needed in order to achieve the desired performance in polymersolar cells for practical application.12,13

4. Development of New Polymer Materials for High Performance Polymer/Fullerene Solar Cells There are many factors limiting the performance of the BHJ solar cells. Among them, materials of active layer is the most determining factor in the overall solar cell performance as most of the fundamental processes, like light harvesting, charge generation and charge transport happen in the active layer. Based on the lessons learned from PPV/PC61 BM and P3HT/PC61 BM systems, several approaches have been pursued to optimize the polymer materials for solar cell performance improvement.12,13

460

Y. Liang and L. Yu

Figure 2. Solar photon flux spectra (AM 1.5).

4.1 Low Bandgap Polymers for Polymer/fulleride BHJ Solar Cells The spectrum of solar photon flux reaching the surface of the earth is shown in Fig. 2.45 The flux distributes over a large wavelength range from 280 nm to 4000 nm with a maximum at about 700 nm. The absorption of polymer/fulleride BHJ solar cells is mainly contributed from the polymer. However, due to the relatively large bandgaps, both PPV (∼2.2 eV) and P3HT (∼1.9 eV) can only harvest a very small fraction of the solar spectrum. Theoretical calculation showed that P3HT/PC61 BM system with absorption onset at about 650 nm, can absorb, at best, 23% of the available solar photons.46 Therefore, extending the polymer absorption to a near-infrared region is attractive in improving the efficiency of polymer/fullerene based BHJ solar cells by efficiently harvesting the energy in the entire solar spectrum. It can be achieved by the use of low bandgap polymers in BHJ solar cells. Herein, the low bandgap polymers are defined as polymers with a bandgap less than 1.8 eV. Low bandgap polymers can be synthesized by a variety of methods. One of the most used methods is the donor-accepter approach, in which a electron-rich unit (donor, D) and a electron-poor unit (acceptor, A) are alternatively coupled in the polymer backbone. Such alternation results in a mesomerism, D-A < = > D+ = A−.47 Increasing the strength of the donor and the acceptor can enhance the double bond character in the single bonds between these two units on the polymer chain, which reduces the bond-length alteration energy, leading to the reduction of the bandgap.48 Use of fused rings to enhance the quinoid character in polyaromatic system is another efficient way to lower the bandgap. It is first illustrated in polyisothianaphthene (PITN) reported by Wudl et al. in 1984.49 (Fig. 3) As the benzene has higher resonance energy than thiophene, the fused six-membered benzene ring gains aromaticity and the thiophene ring tends to dearomatize into a quinoid structure. The quinoid structure can effectively lower the bond alternation energy and reduce the polymer bandgap. Besides these two approaches, the polymer bandgap can be reduced by

Figure 3. Aromatic and quinoid structures of PITN.

Developing Semiconductor Polymers for Solar Energy Harvesting

461

Figure 4. Structures of benzothiadiazole based low bandgap polymers for BHJ solar cells.

increasing the planarity of the conjugated backbone, increasing the pi-conjugation length and the control of intermolecular interactions.50 Benzothiadiazole is often used to construct low bandgap polymers for BHJ solar cells. One of the first low bandgap polymers designed for BHJ solar cells was reported by Dhanabalan et al. in 2001.51 The polymer PTPTB (Fig. 4) is composed of dithienylpyrrole donor block and benzothiadiazole acceptor block, synthesized by Stille coupling reaction. The optical bandgap of the polymer is about 1.46 eV–1.60 eV, depending on the molecular weight. Initial results gave an efficiency of about 0.34% from the polymer solar cell fabricated by PTPTB/PC61 BM in 1:1 weight ratio. After optimization, a better efficiency of about 1% was achieved from PTPTB/PC61 BM in 1:3 weight ratio with a Voc of 0.72 V, a Jsc of 3.10 mA/cm2 and an FF of 0.37.52 The photocurrent of the PTPTB devices is extended to about 770 nm. However, the EQE is relatively low, almost smaller than 20% over the whole spectrum, which accounts for the low PCE. A variety of low bandgap polymers have been developed for OPV since53,54; however, few of them can exhibit comparable performance as P3HT until Muhlbacher et al. reported a copolymer of dialkyl-cyclopentabithiophene and benzothiadiazole, PCPDTBT in 2006 (Fig. 4).55 Cyclopentadithiophene is a rigid and strong electron-donating unit, which makes polymer PCPDTBT have very low bandgap and high hole mobility. PCPDTBT has an optical bandgap of about 1.40 eV, with a board absorption of about 890 nm. The primary BHJ solar cells of PCPDTBT/PC71 BM in 1:3 weight ratio showed a PCE of 3.2%, a Voc of 0.65 V, and a Jsc of 11 mA/cm2. The peak EQE value (38%) and FF (47%) are relatively low, which is due to the unoptimized morphology. In 2008 Peet and Lee et al. reported that the use of additives, like alkanedithiols or diiodoalkane, can ameliorate such a morphology problem.56,57 One of the best BHJ solar cell devices was achieved from PCPDTBT/PC71 BM (1:3.6 weight ratio) in chlorobenzene solution with 2.5% (volume ratio) diiodooctane. It delivers a Voc of 0.61 V, a Jsc of 15.3 mA/cm2 and an FF of 0.53, which results in an efficiency of 5.12%. A closely related polymer, PSBTBT, was reported by Hou et al.58 (Fig. 4) Replacing the bridging carbon atom in cyclopentadithiophene to the silicone atom has a positive impact on the hole transport but a slight change on absorption. The best BHJ solar cells were achieved from PSBTBT/PC71 BM blend in 1:1 weight ratio. It gives a Voc of 0.68 V and a Jsc of 12.7 mA/cm2 together with an FF of 0.55 leading to an efficiency of 5.1%. Recently, Coffin et al. optimized the polymerization condition by use of microwave heating and screening of comonomer reactant ratios to obtained a higher molecular weight polymer P2 with same backbone.59 Such high molecular weight P2 gives enhanced efficiency of 5.9% in P2/PC71 BM solar cells.

462

Y. Liang and L. Yu

Figure 5. Structures of diketo-pyrrolo-pyrrole based low bandgap polymers and acceptor side chain polymer for BHJ solar cells.

Recently, the use of diketo-pyrrolo-pyrrole as acceptor to achieve low band polymers for BHJ solar cell was reported by Wienk et al.60 (Fig. 5) The diketo-pyrrolo-pyrrole unit is very electron poor and has strong packing ability. These types of moieties were first incorporated into conjugated polymers as efficient light absorber in photorefractive polymers by Chan et al. in 1993.61 The polymer reported by Janssen et al.,60 PBBTDPP2, is composed of quaterthiophene and diketo pyrrolo-pyrrole and has a low optical bandgap of 1.40 eV in film. The photovoltaic performance of PBBTDPP2 is highly related to the processing solvent. The optimized PBBTDPP2/PC71 BM solar cells were achieved in 1:2 weight ratio prepared from chloroform/o-dichlorobenzene mixed solvent. The device gives a Voc of 0.61 eV and a Jsc of 11.3 mA/cm2, as well as an FF of 0.58, yielding an efficiency of 4.0%. Another interesting diketo-pyrrolo-pyrrole polymer (PDPP-BDP) is reported by Huo et al.62 The polymer is composed of benzo[2,1-b:3,4-b ]dithiophene as another building unit. It has a similar bandgap (1.34 eV) as PBBTDPP2, and PDPP-BDP/PC71BM solar cell gives an efficiency of 4.4% . Another method to make a low bandgap polymer for OPV is proposed by Huang et al. through making polymers with D-pi-bridge-A side chains structure.63 The polymers are composed of fluorene and triarylamine alternating units with acceptor groups connecting Table 3 Photovoltaic properties of some low bandgappolymer BHJ solar cells Materials PTPTB/PC61 BM PCPDTBT/PC71 BM PCPDTBT/PC71 BM PSBTBT/PC71 BM P2/PC71 BM PBBTDPP2/PC71 BM PDPP-BPD2/PC71 BM PFPDT/PC71 BM

Polymer bandgap (eV)

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

Reference

1.46 1.40 1.40 1.45 1.37 1.40 1.34 1.76

0.72 0.65 0.61 0.68 0.57 0.61 0.72 0.99

3.10 11.0 15.3 12.7 17.3 11.3 10.0 9.6

37 47 53 55 61 58 62 50

1.0 3.2 5.1 5.1 5.9 4.0 4.4 4.5

42 46 47 (b) 48 49 50 52 53

Developing Semiconductor Polymers for Solar Energy Harvesting

463

to one arm of the triarylamine as side chains. The polymer PFPDT has a bandgap of 1.76 eV. The solar cell made by PFPDT/PC71 BM gives an efficiency of 4.5% with a Voc of 0.99 V, a Jsc of 9.62, and an FF of 50%. The polymer has a relative high hole mobility (1.16 × 10−3 cm2 V−1S−1 from the space charge limiting current (SCLC) method), which is due to better isotropic charge transport from the two-dimensional structure. 4.2 Large Bandgap Polymers with Low Lying HOMO Energy Level for BHJ Solar Cells In P3HT/PC61 BM solar cell, there is about 1 eV energy offset between P3HT LUMO level and PC61 BM LUMO level. It is reported that an energy offset of about 0.3 eV is sufficient for effective charge separation.64 There is about 0.6 eV–0.7 eV energy offset wasted. The Voc is determined by the energy difference of the polymer HOMO level and the PC61 BM LUMO level. As a result, compared to the P3HT/PC61 BM system, improved performance can be achieved in large bandgap polymers by lowering the HOMO and LUMO energy levels simultaneously to provide larger Voc and less wasted LUMO offset energy.65 This concept was first illustrated in fluorine-based polymer solar cells reported by Svensson et al. 66 and Zhou et al. 67 (Fig. 6) The weak strength of the fluorene donor and the benzothiadiazole acceptor results in a rather large bandgap in PFDTBT, which is about 1.9 eV. The PFDTBT/PC61 BM (1:4 weight ratio) solar cells provided a Voc of about 1.0 V, much larger than that of P3HT. Combining with a Jsc of 4.66 mA/cm2 and an FF of 0.46, it gives an efficiency of 2.2%.66 Further optimization of the side-chain patterns and fabrication procedures leads to an efficiency of about 4.2%, mainly by the increase of Jsc to 7.7 mA/cm2.68 Substitution of the carbon atom in 9-position of fluorene to a silicon atom leads to a new polymer PSiF-DBT.69 (Fig. 6) The substitution significantly improved the hole mobility of PSiF-DBT. Still the polymer has rather low HOMO energy level, which was measured

Figure 6. Structures of large bandgap polymers with low lying HOMO energy levels for BHJ solar cells.

464

Y. Liang and L. Yu

Table 4 Photovoltaic properties of some large bandgappolymer BHJ solar cells with low large Voc Materials PFDTBT/PC61 BM PSiF-DBT/PC61 BM PCDTBT/PC61 BM PCDTBT/PC71 BM HXS-1/PC71 BM APFO-15/PC61 BM

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

Reference

1.00 0.90 0.89 0.88 0.81 1.00

4.66 9.50 6.92 10.6 9.8 6.00

46 51 63 66 69 63

2.2 5.4 3.6 6.1 5.4 3.7

56 59 60 61 62 63

to be −5.39 eV. PSiF-DBT/PC61 BM solar cell showed a PCE of 5.4% under 80 mW/cm2 illumination, which is mainly due to the high Voc (0.90 V) and good Jsc (9.5 mA/cm2). Blouin et al. reported a similar polymer based on carbazole, PCDTBT.70 (Fig. 6) Carbazole has good p-type transport properties. Substitution of fluorene in APGF-3 into carbazole in PCDTBT is expected to improve the hole mobility of the polymer. The polymer has an absorption onset at 660 nm, which corresponds to a bandgap of 1.88 eV. The cyclovoltametry (CV) measurement showed a low lying HOMO energy level, which is about −5.50 eV (LUMO of PC61 BM is −4.30 eV). The best device performance is about 3.6% (under 90 mW/cm2, AM 1.5) from PCDTBT/PC61 BM in 1:4 weight ratio. The good performance is attributed to the large Voc of 0.89 V (due to low lying HOMO) and the high FF of 0.63 (due to good hole mobility). Further optimization of this polymer is recently reported by Park et al.71 Three ways are used for the further improvement: 1) replacing the PC61 BM into PC71 BM to improve light harvesting; 2) inserting a TiO2 optical layer to improve absorption of reflective light and electron injection to cathode; 3) searching good solvent to improve nanomorphology. The best device from PCDTBT/PC71 BM in 1:4 ratio gives an efficiency of 6.1%, with Voc = 0.88 V, Jsc = 10.6 mA/cm2 and FF = 0.66. A similar polymer (HXS-1) was recently reported by Qin et al., in which two octyloxy chains were introduced to benzothiazole ring and an octyl chain was connected to carbazole instead of a branched chain.72 Such modification leads to planar conformation of the polymer chains in solid state. The polymer solar cell prepared by HXS-1/PC71BM from DCB/DIO gives a PCE of 5.4%, with Voc = 0.81 V and Jsc = 9.8 mA/cm2 and FF = 0.69. The high FF indicates the balanced change transports in the solar cell device. Solar cells with large Voc and good PCE can also be achieved in quinoxaline based polymers. Gadisa et al. reported a polymer, APFO-15 composed of quinoxaline and fluorene alternating units.73 The best polymer solar cells from APFO-15/PC61 BM in 1:3 ratio gave a PCE of 3.7% with Voc = 1.0 V, Jsc = 6.0 mA/cm2, and FF = 0.63. The high PCE is related to the rather high hole mobility and balanced charge carriers transport, as well as the low lying HOMO energy level. However, similar to other large bandgap polymers, the limited absorption results in rather small Jsc . 4.3 New Polymer Structure for Enhanced Photovoltaic Performance The above discussions have emphasized the importance of control in polymer bandgap and energy level to improve the polymer’s photovoltaic performance. Besides, the high hole mobility and suitable miscibility with fulleride to form interpenetrating network are also required for the polymer to achieve highly efficient polymer solar cells. However, it is very

Developing Semiconductor Polymers for Solar Energy Harvesting

465

Figure 7. Structures of thieno[3,4-b]thiophene ester based polymers.

difficult to design a polymer to fulfill all these requirements so far. The power of organic synthesis enables the tailoring of polymer properties by fine-tuning its structure, which can lead to enhanced photovoltaic performance in the polymer solar cell. Examples will be given below on a series of thieno[3,4-b]thiophene polymers developed for photovoltaic applications. Thieno[3,4-b]thiophene polymer is one type of low bandgap polymer in which a fused thiophene ring can stabilize the quinoidal structure of the polymer to lower the bandgap.74 Several thieno[3,4-b]thiophene polymer derivatives have been reported; however, the high HOMO energy levels of these polymers limit their photovoltaic application.75,76 To lower the HOMO, an electron withdrawing ester group was introduced to thienothiophene ring.77 Meanwhile, the side chain on ester can also improve the polymer solubility. A polymer (PTT) composed of such thienothiophene ester and thiophene alternating units was synthesized by Stille coupling reaction (Figure 7). Although PTT shows a bandgap of 1.30 eV and wide absorption in visible and near infrared range, the PTT/PC61 BM solar cell only yields a PCE of 0.6%.77 The low PCE is due to the high HOMO energy level and relative low hole mobility caused by the bulky side chain used. To adjust the energy level of the resulting polymers, a series of regio-random copolymers based on thieno[3,4-b]thiophene and alkyl thiophene unit have been synthesized. It is found that by controlling the ratio of thieno [3,4-b]thiophene to alkyl thiophene in the copolymer composition, the electro-optic properties of the copolymers can be fine-tuned.78 When these copolymers were blended with PC61 BM to form an active layer of the solar cell, an optimized copolymer composition (PTT-C) was found to give the highest efficiency of 1.9%.78 To further enhance the polymer mobility, a regioregular copolymer (PF) was made by the incorporation of well-defined oligothiophene on the polymer backbone. PCE of 2.4% has been achieved by PF/PC61 BM solar cell.79 The above thienothiophene based polymer have better light absorption than P3HT, however, the efficiencies of such polymer solar cells are much lower than P3HT. One limiting factor is the rather low hole mobility in these polymers. To solve the problem, a thieno[3,4b]thiophene ester and benzo[1,2-b:4,5-b ]dithiophene alternating polymer (PTB1) was developed (Fig. 8).80 Benzo[1,2-b:4,5-b ]dithiophene has been recently reported as a building

Figure 8. Structure of PTB1.

466

Y. Liang and L. Yu

Figure 9. Structures of PTB polymers. (Figure available in color online)

block for semiconducting polymers with high hole mobility.81 The dithiophene rings are fused by a benzene ring, which forms a rigid structure. The incorporation of the benzodithiophene unit is expected to improve the polymer hole mobility. Meanwhile, there are two side chains on the benzodithiophene ring, which can enable good solubility of the polymer without introducing too bulky a side chain on thienothiophene ester. The side chain pattern also provides good packing ability and space for fulleride intercalation. The optical bandgap of PTB1 is 1.58 eV with peak absorption at 690 nm, corresponding to the maximum photon flux of the solar spectrum. PTB1 has higher SCLC mobility (4.5 × 10−4 cm2 V−1s−1) than P3HT (2.7 × 10−4 cm2 V−1s−1) measured in the same condition. PTB1 has good miscibility with fulleride and fine interpenetrating networks are formed in PTB1/PC61 BM blend film. These superior properties lead to good photovoltaic performance of PTB1 solar cells. PTB1/PC61 BM solar cell gives a PCE of 4.8% with Voc = 0.58 V, Jsc = 12.5 mA/cm2 and FF = 65.4%; while PTB1/PC71 BM solar cell gives a PCE of 5.6% with Voc = 0.56 V, Jsc = 15.8 mA/cm2 and FF = 63.3%. The enhanced efficiency in PC71 BM solar cell is due to the higher absorption of PC71 BM in the visible region compared to PC61 BM. The Jsc and FF obtained from PTB1 solar cells are among the highest values reported for solar cell system based on low bandgap polymers. However, the Voc of the polymer solar cells is relatively small, just about 0.56–0.58 V. The encouraging results of polymer PTB1 lead us to select the polymer backbone as the structural platform to further investigate the structure/property relationship and to search for new polymers with improved solar cell performance. As mentioned before, a way to increase Voc is lowering the polymer HOMO energy level65 which can be achieved by introducing an electron withdrawing group to the polymer backbone or replacing the electron rich group on the backbone to a less electron rich group. Side chains on polymer backbone can be fine-tuned to further optimize the miscibility with fullerene. A series of new semiconducting polymers with alternating thieno[3,4-b]thiophene and benzodithiophene units have been developed.82 (Fig. 9) Two methods are applied to lower the polymer HOMO. In PTB3, the electron rich alkoxy side chains on benzodithiophene in PTB1 are replaced with a less electron rich alkyl chain. In PTB4, electron withdrawing fluorine is introduced to the thienothiophene of the polymer backbone. The length and pattern of the side chains on both the thienothiophene ester and benzodithiophene are varied in these polymers to tune the polymer solubility and miscibility with fullerene. The electro-optic properties of these polymers are very similar; for example, they have similar bandgap and absorption spectra. (Fig. 10) As expected, PTB3 and PTB4 have lower HOMO energy level (∼0.1 eV) compared to the analogues with the same side-chain patterns. The hole mobility of these polymers varied slightly according to structures: the polymer with bulkier side chains have lower mobility and the alkyl grafted PTB3 and fluorinated PTB4 exhibit larger hole mobility than other polymers.

Developing Semiconductor Polymers for Solar Energy Harvesting

467

Figure 10. (a) Absorption spectra and (b) energy levels of PTB polymers. (Figure available in color online)

The photovoltaic performance of these polymers was studied in the PTB/PC61 BM solar cells. The alkyl substituted PTB3 and the fluorinated polymer PTB4 devices showed enhanced Voc (>0.72 V). The changes in Voc are well correlated with the HOMO energy levels of polymers. In the optimized conditions, polymer solar cells from PTB3 (5.85%) and PTB4 (6.10%) showed obvious enhancement in PCE compared to PTB1 (4.76%). The PCE enhancement is due to the increase of Voc without sacrificing of Jsc and FF. It shows the success of structure fine-tuning to improve the solar cell performance. Side-chain patterns also play an important role in the polymer performance. When the long dodecyl chain on PTB1 changes to a shorter 2-ethylhexyl side chain on PTB2, the PCE of PTB2 slightly increases (5.10%), which may be due to the increase of miscibility. In another side, the too bulky side chains can be detrimental to the polymer performance as they reduce the polymer miscibility with fullerides (like PTB6). The polymer photovoltaic performance is related to the morphology of composite films, which can be affected by the film preparation conditions. In some of the PTB polymers, the active layer needs to be prepared from mixed solvents (like dichlorobenzene/diiodooctance) to achieve optimized morphology . As it has been demonstrated in PTB1 solar cells, the use of PC71 BM instead of PC61 BM as acceptor can lead to further enhancement of the PCE. A power conversion efficiency of 7.4% has been achieved from PTB7/PC71 BM solar cell devices with Voc = 0.74 V, Jsc = 14.5 mA/cm2 and FF = 69.0%.83 It is the first polymer solar cell system showing power conversion efficiency over 7%. A similar polymer with ketone replacing to ester in thieno[3,4-b]thiophene unit showed a similar PCE of 7.7%.84 Table 5 Characteristic properties of polymers and their solar cells in PTBx/PC61 BM composite Polymers PTB1 PTB2 PTB3 PTB4 PTB5 PTB6 PTB3a PTB4a PTB5a

EHOMO (eV)

ELUMO (eV)

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

−4.90 −4.94 −5.04 −5.12 −5.01 −5.01

−3.20 −3.22 −3.29 −3.31 −3.24 −3.17

0.58 0.60 0.74 0.76 0.68 0.62 0.72 0.74 0.66

12.5 12.8 13.1 9..20 10.3 7.74 13.9 13.4 10.7

65.4 66.3 56.8 44.5 43.1 47.0 58.5 61.4 58.0

4.76 5.10 5.53 3.10 3.02 2.26 5.85 6.10 4.10

Note: a). Devices prepared from mixed solvents dichlorobenzene/diiodooctance (97/3, v/v).

468

Y. Liang and L. Yu

Figure 11. Structure of thieno[3,4-c]pyrrole-4,6-dione polymers.

Inspired by the design of PTB polymers, several similar polymers, in which thieno[3,4b]thiophene units are replaced to thieno[3,4-c]pyrrole-4,6-dione units, have been reported most recently for efficient solar cells.85–87 The use of thieno[3,4-c]pyrrole-4,6-dione unit lowers the polymer HOMO energy level, which leads to a larger Voc (>0.80). However, the lowering of HOMO energy level also causes the increase of polymer bandgap (∼1.80 eV), resulting in a smaller Jsc. PCE of 5.5% has been demonstrated in the polymer solar cells with PC71 BM as acceptor.

5. Conclusions Since their discovery in 1995, polymer/fullerene BHJ solar cells constitute a very popular research area, and a better understanding of the operating mechanism and factors that affect the solar cell performance have been obtained. However, the observed efficiency in polymer solar cells, especially in large solar panels is still lower than the corresponding commercial inorganic systems, which limits their vital applications. It has been realized that the ideal polymer in BHJ structure should exhibit a broad absorption with a high coefficient in the solar spectrum to ensure effective harvesting of solar photons. In addition, the polymer should have high hole mobility for charge transport. To maximize the effective charge carriers extraction, the hole mobility of the polymer should be balanced with the electron mobility of the fulleride acceptor. It also requires for the polymer to have suitable energy levels matching the fulleride. The polymer should have a low-lying HOMO energy level to provide a large Voc and suitable LUMO energy level to provide enough offset for charge separation. Besides these optical and electronic properties, the polymer should have appropriate compatibility with fullerene to form effective bicontinuous interpenetrating network in nanoscale for charge separation and transport. Significant progress has been made in developing various polymer donor materials that exhibit properties that satisfy the above requirements. A good example is the PTB polymer system with thieno[3,4-b]thiophene and benzodithiophene alternating units, based on which the BHJ solar cells with PC71 BM exhibited a PCE close to 8%. However, several challenges still exist in this area. First, new donor materials with better properties are still needed. Second, acceptor polymers that can rival fullerenes are essential to lower the cost of polymer solar cells. Third, a detailed understanding of the photochemical stability of the organic solar cells must be gained. With the current success, we believe that solutions can be found and polymer solar cells will have a bright future.

Acknowledgements We would like to acknowledge the support from NSF, AFOSR, DOE, NSF MRSEC (University of Chicago) and Solarmer Energy Inc. for the preparation of this article and the works described here.

Developing Semiconductor Polymers for Solar Energy Harvesting

469

References 1. Krebs, F. C. Polymer Photovoltaics A Practical Approach; SPIE Press, Bellingham, 2008. 2. Shah, A.; Torres, P.; Tscharner, R.; Wyrsch, N.; Keppner, H. “Photovoltaic technology: The case for thin-film solar cells,” Science, 1999, 285, 692–698. 3. Pagliaro, M.; Ciriminna, R.; Palmisano, G. “Flexible solar cells,” Chem Sus Chem., 2008, 1, 880–981. 4. Lewis, N. S. “Toward cost-effective solar energy use,” Science, 2007, 315, 798–801. 5. Brabec, C. J. “Organic photovoltaics: Technology and market,” Sol. Energy Mater. Sol. Cells., 2004, 83, 273–292. 6. Tang, C.W. “Two-layer organic photovoltaic cell,” Appl. Phys. Lett., 1986, 48, 183–185. 7. Yu, G.; Gao, J.; Hummelen, J. C.; Wudl, F.; Heeger, A. J. “Polymer photovoltaic cells: Enhanced efficiencies via a network of internal donor-acceptor heterojunctions,” Science, 1995, 270, 1789–1791. 8. Press release, Solarmer Energy, Inc.: El Monte, CA, December 1, 2009. 9. Kippelen, B.; Bredas, J. L. “Organic photovoltaics,” Energy Environ. Sci., 2009, 2, 251– 261. 10. Dennler, G.; Scharber, M. C.; Brabec, C. J. “Polymer-fullerene bulk-heterojunction solar cells,” Adv. Mater., 2009, 21, 1323–1338. 11. Servaites, J. D.; Ratner, M. A.; Marks, T. J. “Practical efficiency limits in organic photovoltaic cells: Functional dependence of fill factor and external quantum efficiency,” Appl. Phys. Lett., 2009, 95, 163–302. 12. Chen, J. W.; Cao, Y. “Development of novel conjugated donor polymers for high-efficienty bulk-heterojunction photovoltaic devies,” Acc. Chem. Res., 2009, 42, 1709–1718. 13. Cheng, Y. J.; Yang, S. H.; Hsu, C. S. “Synthesis of conjugated polymers for organic solar cell applications,” Chem. Rev., 2009, 109, 5868–5923. 14. Tamayo, A. B.; Walker, B.; Nguyen, T. Q. “A low band gap, solution processable oligothiophene with a diketopyrrolopyrrole core for use in organic solar cells,” J. Phys. Chem. C, 2008, 112, 11545–11551. 15. Silvestri, F.; Irwin, M. D.; Beverina, L.; Facchetti, A.; Pagani, G. A.; Marks, T. J. “Efficient squaraine-based solution processable bulk-heterojunction solar cells,” J. Am. Chem. Soc., 2008, 130, 17640–17641. 16. Hummelen, J. C.; Knight, B. W.; LePeq, F.; Wudl, F. “Preparation and characterization of fulleroid and methanofullerene derivatives,” J. Org. Chem., 1995, 60, 532–538. 17. Halls, J. J. M.; Walsh, C. A.; Greenham, N. C.; Marseglia, E. A.; Friend, R. H.; Moratti, S. C.; Holmes, A. B. “Efficient photodiodes from interpenetrating polymer networks,” Nature, 1995, 376, 498–500. 18. McNeill, C. R.; Abrusci, A.; Zaumseil, J.; Wilson, R.; McKiernan, M. J.; Halls, J. J. M.; Greenham, N. C.; Friend, R. H. “Dual electron donor/electron acceptor character of a conjugated polymer in efficient photovoltaic diodes,” Appl. Phys. Lett., 2007, 90, 193506. 19. McNeill, C. R.; Halls, J. J. M.; Wilson, R.; Whiting, G. L.; Berkebile, S.; Ramsey, M. G.; Friend, R. H.; Greenham, N. “Efficient polythiophene/polyfluorene copolymer bulk heterojunction photovoltaic devices: Device physics and annealing effects,” Adv. Funct. Mater., 2008, 18, 2309–2321. 20. Westenhoff, S.; Howard, I. A.; Hodgkiss, J. M.; Kirov, K. R.; Bronstein, H. A.; Williams, C. K.; Greenham, N. C.; Friend, R. H. “Charge recombination in organic photovoltaic devices with high open-circuit voltages,” J. Am. Chem. Soc., 2008, 130, 13653–13658. 21. Coakley, K. M.; McGehee, M. D. “Conjugated polymer photovoltaic cells,” Chem. Mater., 2004, 16, 4533–4542. 22. Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. “Hybrid nanorod-polymer solar cells,” Science, 2002, 295, 2425–2427. 23. Beek, W. J. E.; Wienk, M. M.; Janssen, R. A. J. “Efficient hybrid solar cells from zinc oxide nanoparticles and a conjugated polymer,” Adv. Mater., 2004, 16, 1009–1013.

470

Y. Liang and L. Yu

24. Shaheen, S. E.; brabec, C. J.; Sariciftci, N. S.; Padinger, F.; Fromherz, T.; Hummelen, J. C. “2.5% efficient organic plastic solar cells,” Appl. Phys. Lett., 2001, 78, 841–843. 25. Brabec, C. J.; Shaheen, S. E.; Winder, C.; Sariciftci, N. S.; Denk, P. “Effect of LiF/metal electrodes on the performance of plastic solar cells,” Appl. Phys. Lett., 2002, 80, 1288–1290. 26. Wienk, M. M.; Kroon, J. M.; Verhees, W. J. H.; Krol, J.; Hummelen, J. C.; Van Haal, P.; Janssen, R. A. J. “Efficient methano[70]fullerene/MDMO-PPV bulk heterojunction photovoltaic cells,” Angew. Chem., Int. Ed., 2003, 42, 3371–3375. 27. Bao, Z. N.; Dodabalapur, A.; Lovinger, A. “Soluble and processable regioregular poly(3hexylthiophene) for thin film field-effect transistor applications with high mobility,” Appl. Phys. Lett., 1996, 69, 4108–4110. 28. Padinger, F.; Rittberger, R.; Sariciftci, N. S. “Effects of postproduction treatment on plastic solar cells,” Adv. Funct. Mater., 2003, 13, 85–88. 29. Li, G., Shrotriya, V.; Yao, Y.; Yang, Y. “Investigation of annealing effects and film thickness dependence of polymer solar cells based on poly(3-hexylthiophene),” J. Appl. Phys., 2005, 98, 043704. 30. Ma, W. L.; Yang, C. Y.; Gong, X.; Lee, K. H.; Heeger, A. J. “Thermally stable, efficient polymer solar cells with nanoscale control of the interpenetrating network morphology,” Adv. Funct. Mater., 2005, 15, 1617–1622. 31. Reyes-Reyes, M.; Kim, K.; Carroll, D. L. “High-efficiency photovoltaic devices based on annealed poly(3-hexylthiophene) and 1-(3-methoxycarbonyl)-propyl-1-phenyl-(6,6)C-61 blends,” Appl. Phys. Lett., 2005, 87, 083506. 32. Yang, C. Y.; Hu, J. G.; Heeger, A. J. “Molecular structure and dynamics at the interfaces within bulk heterojunction materials for solar cells,” J. Am. Chem. Soc., 2006, 128, 12007–12013. 33. Yang, X. N.; Loos, J.; Veenstra, S. C.; Verhees, W. J. H.; Wienk, M. M.; Kroon, J. M.; Michels, M. A. J.; Janssen, R. A. J. “Nanoscale morphology of high-performance polymer solar cells,” Nano Lett., 2005, 5, 579–583. 34. Li, G.; Shrotriya, V.; Huang, J. S.; Yao, Y.; Moriarty, T.; Emery, K.; Yang, Y. “High-efficiency solution processable polymer photovoltaic cells by self-organization of polymer blends,” Nature Mater., 2005, 4, 864–868. 35. Li, G.; Yao, Y.; Yang, H. C.; Shrotriya, V.; Yang, G. W.; Yang, Y. “‘Solvent annealing’ effect in polymer solar cells based on poly(3-hexylthiophene) and methanofullerenes,” Adv. Funct. Mater., 2007, 17, 1636–1644. 36. Moule, A.; Meerholz, K. “Controlling morphology in polymer-fullerene mixtures,” Adv. Mater., 2008, 20, 240–245. 37. Peet, J.; Soci, C.; Coffin, R. C.; Nguyen, T. Q.; Mihailovsky, A.; Moses, D.; Bazan, G. C. “Method for increasing the photoconductive response in conjugated polymer/fullerene composites,” Appl. Phys. Lett., 2006, 89, 252105. 38. Wang, W. L.; Wu, H. B.; Yang, C. Y.; Luo, C.; Zhang, Y.; Chen, J. W.; Cao, Y. “Highefficiency polymer photovoltaic devices from regioregular-poly(3-hexylthiophene-2,5-diyl) and [6,6]-phenyl-C-61-butyric acid methyl ester processed with oleic acid surfactant,” Appl. Phys. Lett., 2007, 90, 183512. 39. Berson, S.; De Bettingnies, R.; Bailly, S.; Guillerez, S. “Poly (3-hexylthiophene) fibers for photovoltaic applications,” Adv. Funct. Mater., 2007, 17, 1377–1384. 40. Kim, Y.; Cook, S.; Tuladhar, S. M.; Choulis, S. A.; Nelson, J.; Durrant, J. R.; Bradley, D. D. C.; Giles, M.; Mcculloch, I.; Ha, C. S.; Ree, M. “A strong regioregularity effect in selforganizing conjugated polymer films and high-efficiency polythiophene: Fullerene solar cells,” Nature Mater., 2006, 5, 197–203. 41. Sivula, K.; Luscombe, C. K.; Thompson, B. C.; Frechet, J. M. J. “Enhancing the thermal stability of polythiophene: Fullerene solar cells by decreasing effective polymer regioregularity,” J. Am. Chem. Soc., 2006, 128, 13988–13989. 42. Ma, W.; Kim, J. Y.; Lee, K.; Heeger, A. J. “Effect of the molecular weight of poly(3hexylthiophene) on the morphology and performance of polymer bulk heterojunction solar cells,” Macromol. Rapid Commun., 2007, 28, 1776–1780.

Developing Semiconductor Polymers for Solar Energy Harvesting

471

43. Kim, J. Y.; Kim, S. H.; Lee, H. H.; Lee, K.; Ma, W.; Gong, X.; Heeger, A. J. “New architecture for high-efficiency polymer photovoltaic cells using solution-based titanium oxide as an optical spacer,” Adv. Mater., 2006, 18, 572–576. 44. Irwin, M. D.; Buchholz, D. B.; Hains, A. W.; Chang, R. P. H.; Marks, T. J. “p-Type semiconducting nickel oxide as an efficiency-enhancing anode interfacial layer in polymer bulk-heterojunction solar cells,” Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 2783–2887. 45. The data of Solar Spectral Irradiance (air mass 1.5) was obtained from the Web site http: //rredc.nrel.gov/solar/spectra/am1.5/. 46. Bundgaard, E.; Krebs, F. C. “Low band gap polymers for organic photovoltaics,” Sol. Energy Mater. Sol. Cells, 2007, 91, 954–985. 47. Van Mullekom, H. A. M.; Vekemans, J. A. J. M.; Havinga, E. E.; Meijer, E. W. “Developments in the chemistry and band gap engineering of donor-acceptor substituted conjugated polymers,” Mater. Sci. Eng., 2001, 32, 1–40. 48. Roncali, J. “Synthetic principles for bandgap control in linear pi-conjugated systems,” Chem. Rev., 1997, 97, 173–205. 49. Wudl, F.; Kobayashi, M.; Heeger, A. J. “Poly(isothianaphthene),” J. Org. Chem., 1984, 49, 3382–3384. 50. Roncali, J. “Molecular engineering of the band gap of pi-conjugated systems: Facing technological applications,” Macromol. Rapid Commun., 2007, 28, 1761–1775. 51. Dhanabalan, A.; Van Duren, J. K. J.; Van Hal, P. A.; Van Dogen, J. L. J.; Janssen, R. A. J. “Synthesis and characterization of a low bandgap conjugated polymer for bulk heterojunction photovoltaic cells,” Adv. Funct. Mater., 2001, 11, 255–262. 52. Brabec, C. J.; Winder, C.; Sariciftci, N. S.; Hummelen, J. C.; Dhanabalan, A.; Van Hal, P. A.; Janssen, R. A. J. “A low-bandgap semiconducting polymer for photovoltaic devices and infrared emitting diodes,” Adv. Funct. Mater., 2002, 12, 709–712. 53. Kroon, R.; Lenes, M.; Hummelen, J. C.; Blom, P. W. M.; De Boer, B. “Small bandgap polymers for organic solar cells (polymer material development in the last 5 years),” Polym. Rev., 2008, 48, 531–582. 54. Bundgaard, E.; Krebs, F. C. “Low bandgap polymers for organic photovoltaics,” Sol. Energy Mater. Sol. Cells, 2007, 91, 954–985. 55. Muhlbacher, D.; Scharber, M.; Morana, M.; Zhu, Z. G.; Waller, D.; Gaudiana, R.; Brabec, C. “High photovoltaic performance of a low-bandgap polymer,” Adv. Mater., 2006, 18, 2884–2889. 56. Peet, J.; Kim, J. Y.; Coates, N. E.; Ma, W. L.; Moses, D.; Heeger, A. J.; Bazan, G. C. “Efficiency enhancement in low-bandgap polymer solar cells by processing with alkane dithiols,” Nat. Mater., 2007, 6, 497–500 57. Lee, J. K.; Ma, W. L.; Brabec, C. J.; Yuen, J.; Moon, J. S.; Kim, J. Y.; Lee, K.; Bazan, G. C.; Heeger, A. J. “Processing additives for improved efficiency from bulk heterojunction solar cells,” J. Am. Chem. Soc., 2008, 130, 3619–3623. 58. Hou, J. H.; Chen, H. Y.; Zhang, S. Q.; Li, G.; Yang, Y. “Synthesis, characterization, and photovoltaic properties of a low band gap polymer based on silole-containing polythiophenes and 2,1,3-benzothiadiazole,” J. Am. Chem. Soc., 2008, 130, 16144–16145. 59. Coffin, R. C.; Peet, J.; Rogers, J.; Bazan, G. C. “Streamlined microwave-assisted preparation of narrow-bandgap conjugated polymers for highperformance bulk heterojunction solar cells,” Nature Chem., 2009, 1, 657–661. 60. Wienk, M. M.; Turbiez, M.; Gilot, J.; Janssen, R. J. J. “Narrow-bandgap diketo-pyrrolo-pyrrole polymer solar cells: The effect of processing on the performance,” Adv. Mater., 2008, 20, 2556–2560. 61. Chan, W. K.; Chen, Y. M.; Peng, Z. H.; Yu, L. P. “Rational designs of multifunctional polymers,” J. Am. Chem. Soc., 1993, 115, 11735–11743. 62. Huo, L. J.; Hou, J. H.; Chen, H. Y.; Zhang, S. Q.; Jiang, Y.; Chen, T. L.; Yang, Y. “Bandgap and molecular level control of the low-bandgap polymers based on 3,6-dithiophen-2-yl-2,5dihydropyrrolo[3,4-c]pyrrole-1,4-dione toward highly efficient polymer solar cells,” Macromolecules, 2009, 42, 6564–6571.

472

Y. Liang and L. Yu

63. Huang, F.; Chen, K. S.; Yip, H. L.; Hau, S. K.; Acton, O.; Zhang, Y.; Luo, J. D.; Jen, A. K. Y. “Development of new conjugated polymers with donor-π -bridge-acceptor side chains for high performance solar cells,” J. Am. Chem. Soc., 2009, 131, 13886–13887. 64. Bredas, J. L.; Beljonne, D.; Coropceanu, V.; Cornil, J. “Charge-transfer and energy-transfer processes in pi-conjugated oligomers and polymers: A molecular picture,” Chem. Rev., 2004, 104, 4917–5003. 65. Scharber, M.; Muhlbacher, D.; Koppe, M.; Denk, P.; Waldauf, C.; Heeger, A. J.; Brabec, C. “Design rules for donors in bulk-heterojunction solar cells—towards 10% energy-conversion efficiency,” Adv. Mater., 2006, 18, 789–794. 66. Svensson, M.; Zhang, F. L.; Veenstra, S. C.; Verhees, W. J. H.; Hummelen, J. c.; Kroon, J. M.; Inganas, O.; Andersson, M. R. “High-performance polymer solar cells of an alternating polyfluorne copolymer and a fullerene derivative,” Adv. Mater., 2003, 15, 988–991. 67. Zhou, Q. M.; Hou, Q.; Zheng, L. P.; Deng, X. Y.; Yu, G.; Cao, Y. “Fluorene-based low bandgap copolymers for high performance photovoltaic devices,” Appl. Phys. Lett., 2004, 84, 1653– 1655. 68. Slooff, L. H.; Veenstra, S. C.; Kroon, J. M.; Moet, D. J. D.; Sweelssen, J.; Koetse, M. M. “Determining the internal quantum efficiency of highly efficient polymer solar cells through optical modeling,” Appl. Phys. Lett., 2007, 90, 143506. 69. Wang, E. G.; Wang, L.; Lan, L. F.; Luo, C.; Zhuang, W. L.; Peng, J. B.; Cao, Y. “Highperformance polymer heterojunction solar cells of a polysilafluorene derivative,” Appl. Phys. Lett., 2008, 92, 033307. 70. Blouin, N.; Michaud, A.; Leclerc, M. “A low-bandgap poly(2,7-carbazole) derivative for use in high-performance solar cells,” Adv. Mater., 2007, 19, 2295–2300. 71. Park, S. H.; Roy, A.; Beaupre, S.; Cho, S.; Coates, N.; Moon, J. S.; Moses, D.; Leclerc, M.; Lee, K.; Heeger, A. J. “Bulk heterojunction solar cells with internal quantum efficiency approaching 100%,” Nature Photonics, 2009, 3, 297–302. 72. Qin, R. P.; Li, W. W.; Li, C. H.; Du, C.; Veit, C.; Schleiermacher, H. F.; Andersson, M.; Bo, Z. S.; Liu, Z. P.; Inganas, O.; Wuerfel, U.; Zhang, F. L. “A planar copolymer for high efficiency polymer solar cells,” J. Am. Chem. Soc., 2009, 131, 14612–14613. 73. Gadisa, A.; Mammo, W.; Andersson, L. M.; Admassie, S.; Zhang, F. L.; Andersson, M. R.; Inganas, O. “A new donor-acceptor-donor polyfluorene copolymer with balanced electron and hole mobility,” Adv. Funct. Mater., 2007, 17, 3836–3842. 74. Lee, K. H.; Sotzing, G. A. “Poly(thieno[3,4-b]thiophene). A new stable low band gap conducting polymer” Macromolecules, 2001, 34, 5746–5747. 75. Neef, C. J.; Brotherston, I. D.; and Ferraris, J. P. “Synthesis and electronic properties of poly(2-phenylthieno[3,4-b]thiophene): A new low band gap polymer” Chem. Mater., 1999, 11, 1957–1958 76. Pomerantz, M.; Gu, X. M. “Poly(2-decylthieno[3,4-b]thiophene). A new soluble low-bandgap conducting polymer,” Synth. Met., 1997, 84, 243–244. 77. Yao, Y.; Liang, Y. Y.; Shrotriya, V.; Xiao, S. Q.; Yu, L. P.; Yang, Y. “Plastic near-infrared photodetectors utilizing low band gap polyme” Adv. Mater., 2007, 19, 3979–3983. 78. Liang, Y. Y.; Xiao, S. Q.; Feng, D. Q.; Yu, L. P. “Control in energy levels of conjugated polymers for photovoltaic application,” J. Phys. Chem. C, 2008, 112, 7866–7871. 79. Liang, Y. Y.; Feng, D. Q.; Guo, J. C.; Szarko, J. M.; Claire, R.; Chen, L. X.; Yu, L. P. “Regioregular oligomer and polymer containing thieno[3,4-b]thiophene moiety for efficient organic solar cells,” Macromolecules, 2009, 42, 1091–1098. 80. Liang, Y. Y.; Wu, Y.; Feng, D. Q.; Tsai, S.-T.; Li, G.; Son, H. J.; Yu, L. P. “Development of new semiconducting polymers for high performance solar cells,” J. Am. Chem. Soc., 2009, 131, 56–57. 81. Pan, H. L.; Li, Y. N.; Wu, Y. L.; Liu, P.; Ong, B. S.; Zhu, S. P.; Xu, G. “Low-temperature, solutionprocessed, high-mobility polymer semiconductors for thin-film transistors,” J. Am. Chem. Soc., 2007, 129, 4112–4113.

Developing Semiconductor Polymers for Solar Energy Harvesting

473

82. Liang, Y. Y.; Feng, D. Q.; Wu, Y.; Tsai, S.-T.; Li, G.; Ray, C.; Yu, L. P. “Highly efficient solar cell polymers developed via fine-tuning structural and electronic properties,” J. Am. Chem. Soc., 2009, 131, 7792–7799. 83. Liang, Y. Y.; Xu, Z.; Xia, J. B.; Tsai, S. T.; Wu, Y.; Li, G.; Ray, C.; Yu, L. P. For the bright futurebulk heterojunction polymer solar cells with power conversion efficiency of 7.4%. Adv. Mater., 2010, 22, E135–E138. 84. Chen, H. Y.; Hou, J. H.; Zhang, S. Q.; Liang, Y. Y.; Yang, G. W.; Yang, Y.; Yu, L. P.; Wu, Y.; Li, G. “Polymer solar cells with enhanced open-circuit voltage and efficiency,” Nature Photonics, 2009, 3, 649–653. 85. Zou, Y. P.; Najari, A.; Berrouard, P.; Beaupre, S.; RedaAch, B.; Tao, Y.; Leclerc, M. “A thieno[3,4c]pyrrole-4,6-dione-based copolymer for efficient solar cells,” J. Am. Chem. Soc., 2010, 132, 5330–5331. 86. Zhang, Y.; Hau, S. K.; Yip, H. L.; Sun, Y.; Acton, O.; Jen, A. K. Y. “Efficient polymer solar cells based on the copolymers of benzodithiophene and thienopyrroledione,” Chem. Mater., 2010, 22, 2696–2698. 87. Pilliego, C.; Holcombe, T. W.; Douglas, J. D.; Woo, C. H.; Beaujuge, P. M.; Frechet, J. M. J. “Synthetic control of structural order in N-alkylthieno[3,4-c]pyrrole-4,6-dione-based polymers for efficient solar cells,” J. Am. Soc. Chem., 2010, 132, 7595–7597.

Polymer Reviews, 50:474–510, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1558-3724 print / 1558-3716 online DOI: 10.1080/15583724.2010.515764

A Review on the Development of the Inverted Polymer Solar Cell Architecture STEVEN K. HAU, HIN-LAP YIP, AND ALEX K.-Y. JEN Department of Materials Science and Engineering, University of Washington, Seattle, WA 98195-2120, USA The increase in energy production costs for fossil fuels has led to a search for an economically viable alternative energy source. One alternative energy source of particular interest is solar energy. A promising alternative to inorganic materials is organic semiconductor polymer solar cells due to their advantages of being cheaper, light weight, flexible and made into large areas by roll-to-roll processing. However, the conventional architecture that is typically used for fabricating solar cells requires high vacuum to deposit the top metal electrode which is not suitable for roll-to-roll processing. Recently an inverted device architecture has been investigated as a suitable architecture for developing the ideal roll-to-roll type processing of polymer-based solar cells. This review will go over the recent advances and approaches in the development of this type of inverted device architecture. We will highlight some of the work that we have done to integrate materials, device, interface, and processing of the inverted device architecture platform to produce more idealized polymer-based solar cells. Keywords inverted device architecture, polymer solar cell, interface engineering, self assembled monolayers

1. Introduction Recently, organic polymer photovoltaic cells have attracted a significant amount of attention due to the need to develop an inexpensive clean and sustainable renewable energy source. Polymer-based photovoltaic cells have the advantage of being less expensive and solutionprocessible by roll-to-roll type processing techniques.1 The most widely studied polymer blend system is based on a solution processed p-type poly(3-hexyl-thiophene) (P3HT) polymer and an n-type [6,6]-phenyl C61 butyric acid methyl ester (PCBM) fullerene. This polymer blend system has led to efficiencies as high as 5%.2,3 Efficiencies as high as 6 to 7% have been achieved with polymer:fullerene bulk-heterojunction (BHJ) systems by developing lower band-gap polymer materials that can absorb a broader range of the solar spectrum.4–6 The BHJ device configuration is achieved by blending the n-type and p-type materials to increase the number of interfaces between the donor and acceptor phases in order to provide more exciton dissociation/charge separation sites to generate more charge carriers. Generally, phase segregation of the blend materials greater than ∼10–20 nm will lead to lower photocurrents from the high charge recombination rates due to the low mobility of the conjugated materials. Received April 17, 2010; accepted August 9, 2010. Address correspondence to Alex Jen, Dept of Materials Science and Engineering, Box 352120, University of Washington, Seattle, WA 98195. E-mail: [email protected]

474

Development of Inverted Polymer Solar Cell Architecture

475

The BHJ concept has generally been implemented into the conventional device architecture. This architecture consists of a transparent conducting metal oxide coated with a poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) hole-transporting layer followed by the active BHJ layer. To complete the device, a low work function metal electrode (Al, Ca/Al) is evaporated on top as an electron-collecting electrode. Although the majority of polymer-based solar cells are fabricated in the conventional architecture, there are concerns with the inherent device stability. The transparent conducting indium tin oxide (ITO) used as the hole collecting contact can be etched over time upon exposure to the acidic PEDOT:PSS hole-transporting buffer layer.7,8 Replacing the PEDOT:PSS layer with other hole-transporting layers have so far been unsuccessful in maintaining high conversion efficiencies due to the high resistance of the transporting layer.9 The low work function electrode used in this device architecture requires high vacuum deposition leading to increases in fabrication costs. Additionally, ambient exposure can lead to oxidation of these electrodes leading to device degradation and failure.10 Electron selective buffer layers (TiO2 , ZnO) have been successfully inserted between the top metal electrode and organic active layer to minimize oxygen diffusion leading to improved device stability.11–13 However, this still requires high vacuum deposition of the metal electrodes to complete the device. Device architectures that remove PEDOT:PSS at the ITO interface and use non-vacuum deposited high work function metal electrodes at the top interface are needed. Based on these considerations, an inverted device architecture where the nature of the charge collection is reversed was proposed as a good device alternative. This architecture has recently gained considerable research attention due to the device stability and processing advantages compared to the conventional architecture. In this inverted architecture, the polarity of charge collection is the opposite of the conventional architecture allowing the use of higher work function (Au, Ag, Cu) and less air-sensitive electrodes as the top electrode for hole collection. The use of higher work function metals offer better ambient interface device stability and the possibility for using non-vacuum coating techniques to deposit the top electrode helping to reduce fabrication complexity and costs.14–16 Most of the research attention for these inverted device architectures has been to understand how to improve the device efficiency, stability, and processing of the different interface layers in the device structure. The inverted architecture can be differentiated by the direction of the incoming light source illuminating the solar cells (top illuminated or bottom illuminated solar cell architecture). The top illuminated architecture utilizes a reflective buried bottom electrode and a semi-transparent top electrode whereas the bottom illuminated inverted architecture configuration utilizes a higher work function reflective electrode as the top hole collecting contact and a semi-transparent conducting electrode at the bottom to collect electrons. These two types of inverted solar cells architectures have been successfully demonstrated in literature, but most research has been focused on the bottom illuminated type inverted solar cell. This is due to the difficulty in finding a suitable transparent conducting top electrode material for top illuminated devices. In the top illuminated inverted solar cell configuration, the requirements are that the top electrode needs to be semi-transparent so that light can reach the active layer. This type of architecture was demonstrated by Glatthaar et al. In this architecture, a thin film of aluminum was deposited onto glass followed by a titanium film. An active layer of polymer was deposited on top followed by a PEDOT:PSS/Au grid layer.17 Efficiencies of 1.4% were demonstrated using the PEDOT:PSS/Au grid semi-transparent top electrode configuration. Chen et al. also demonstrated a top illuminated type inverted device with efficiency of 3%.18 The difference in this device configuration was that a flexible stainless

476

S. K. Hau et al.

steel metal foil was used as the substrate instead of a glass substrate. The sequence of layers deposited on top of the stainless steel foil were Ag, ITO, Cs2 CO3 , P3HT:PCBM active layer, MoO3 , Al metal grid and ITO. Since these devices were fabricated onto flexible foils, a flexural test was performed showing good mechanical stability with only a 12% decrease in efficiency after 450 bending cycles. The top illuminated inverted configuration was also demonstrated onto a polymer flexible substrate. Instead of a metal foil, a surface-nickelized polyimide film was used. A thin layer of titanium oxide precursor was cast onto the surfacenickelized polyimide film, followed by a P3HT:PCBM active layer, and a high conductivity PEDOT:PSS top electrode showing efficiency of up to 2.4%.19 In the bottom illuminated inverted solar cell configuration, a high work function electrode is utilized at the top to collect holes and an interface layer is used to modify the ITO interface to effectively collect electrons. Some of the interface layers that have been utilized to modify the ITO for effective electron collection are Cs2 CO3 ,20,21 Ca,22 ZnO,10,23,24 and TiO2 .25–27 Since light needs to pass through these layers to reach the active layer to generate a photocurrent, the layers are typically very thin for maintaining high transparency. Direct contact of the top high work function metal to the active layer can lead to degradation of the solar cell performance; therefore, hole transporting/electron blocking layers have been also deposited between these layers to improve charge selectivity and collection of holes. Some materials that have utilized at the active layer and metal interface are based on various transition metal oxides (MoO3 ,28–30 WO3 ,31,32 V2 O5 ,33–35) and solution processed conducting polymers (PEDOT:PSS,10,24–26 SPDPA36). Combining cesium carbonate (Cs2 CO3 ) and vanadium oxide (V2 O5 ), Li et al. demonstrated that by placing these two interface layers at different locations in the device layer stack, the polarity of the solar cell can be changed from a conventional device to an inverted device architecture.20 An inverted solar cell device was demonstrated by evaporating a thin Cs2 CO3 (∼1 nm) at the ITO interface and a thin V2 O5 (∼10 nm) layer between the active P3HT:PCBM layer and top metal electrode leading to an efficiency of 2.25%. Replacing the evaporated Cs2 CO3 with a solution processed Cs2 CO3 layer showed very similar conversion efficiency ∼2.1%. In another study by the same group, efficiency as high as 4.2% in inverted solar cells was achieved using the solution processed Cs2 CO3 based interface layer system.21 Upon annealing the Cs2 CO3 at 150◦ C, the Cs2 CO3 was found to decompose to doped cesium oxide (Cs2 O) helping to reduce the work function from -3.45 eV to -3.06 eV favoring improved electron collection. N-type metal oxides (ZnO and TiO2 ) are more commonly utilized as the interface modification layer at the ITO interface for inverted solar cells due to the high optical transparency in the visible and near infrared, high carrier mobility, and its solution processibility. The energy levels of these metal oxides (LUMO and HOMO) have been reported to be around −4.4 eV and −7.6 eV, respectively. The low LUMO and high HOMO levels allow these materials to be a good electron selective layer and hole-blocking layers. Many demonstrations of using these n-type metal oxide layers as the electron selective layer for inverted solar cells have been reported in literature. An efficient inverted solar cell from a high temperature processed sol-gel ZnO underlayer on ITO and an Ag electrode as the top hole collecting contact was demonstrated by White et al.23 The ZnO sol-gel was thermally annealed at 300◦ C for 5 min to crystallize the ZnO to improve its conductivity and mobility leading to conversion efficiencies of 2.97%. When these devices were exposed to air, it was found that the device performance improved. They attribute the improvement to the oxidation of Ag which caused a shift in the effective work function closer to the HOMO of P3HT leading to improved ohmic contact. It was found that devices stored in nitrogen with periodic exposure to air maintained a device efficiency of 2.32% even after

Development of Inverted Polymer Solar Cell Architecture

477

7 days of storage. A thin titanium oxide sol-gel layer (∼10 nm) at the ITO interface was also demonstrated as an effective electron selective layer for use in inverted solar cells.25 Devices were fabricated from a solution of P3HT:PCBM in o-xylene with a solution processed PEDOT:PSS interface layer between the active layer and Au top electrode showing efficiencies of 3.1%. The authors found using o-xylene as a solvent segregates more PCBM at the bottom (electron collection interface) which is better for inverted solar cells. Controlling the bulk blend vertical phase segregation of one of the materials towards a particular charge collecting interface is important to minimize losses in photocurrent and efficiency caused from charge recombination due to poorly distributed phases. The vertical phase segregation of the donor and acceptor blend materials in the bulk-heterojunction has already been shown by many research groups to have a dramatic effect on the performance of polymer-based solar cells.37–41 In another study by the same group, a thin polyoxyethylene tridecylether (PTE) (∼10 nm) was coated on the ITO surface prior to coating a thin titanium oxide layer in an inverted solar cell leading efficiencies of ∼3.6%.26 It is suggested that the improvement in the efficiency is that the passive PTE layer improves the surface quality for better titanium oxide wetting of the surface, therefore providing a more intimate interface with the active layer. In addition to modifying the ITO surface with interface layers for inverted solar cells, transition metal oxides (MoO3 , WO3 , and V2 O5 ) have also been used to modify the interface between the active layer and top metal electrode. These metal oxide layers are usually deposited by means of vacuum deposition. Inverted solar cell with 2.57% efficiency have been demonstrated using vacuum deposited MoO3 /Ag as the top metal contact and high temperature annealed TiO2 at the ITO interface.29 Further improvement in efficiency to 3.09% was demonstrated using high temperature annealed ZnO at the ITO interface and vacuum deposited MoO3 /Ag at the top electrode.28 Efficiency of 3.55% was achieved using a MoO3 as the top buffer layer and Ca as the electron collecting layer at the ITO interface in inverted solar cells.22 Semi-transparent inverted devices were demonstrated using MoO3 /sputtered ITO as the top electrode and atomic layer deposited titanium dioxide at the bottom ITO interface to collect electrons. The MoO3 buffer layer prevented damage to the active layer from the sputtering process of the ITO coating leading to 1.9% efficiencies.30 WO3 and high temperature processed TiO2 inverted solar cells have also been demonstrated with 2.58% efficiencies.31 Inverted cells using a solution processed V2 O5 and evaporated Ag metal contact on top with ZnO at the ITO interface have also been demonstrated showing efficiencies of 3.56%.33 The use of a solution processible method to cast the buffer layer on top has the advantage of minimizing the fabrication complexity which can lead to lowering of fabrication costs. Other solution processible materials that have been utilized as buffer layers between the active layer and top metal electrode have been based on solution processible polymers such as PEDOT:PSS and self-doped sulfonated poly(diphenylamine) (SPDPA). Li et al. demonstrated inverted solar cell efficiencies of 3.91% using SPDPA as the hole collecting buffer layer between the active layer and the top metal electrode.36 The ideal fabrication process for polymer-based solar cells would be to utilize solution processing techniques to deposit the different layers onto flexible substrates so that it allows for the potential to be used for simple roll-to-roll type processing. Additionally, the solar cells should maintain high device stability and efficiency over a certain period of time. The inverted device architecture has the potential to combine all of these requirements—device flexibility, stability, efficiency, and solution processibility into one system. However, in order to achieve this more ideal polymer-based solar cell system an integrated engineering approach to develop materials, devices and improve interfaces and processing are required. The following will review some of the work that we have done to integrate materials,

478

S. K. Hau et al.

device, interface, and processing onto the inverted device architecture platform to produce a more idealized polymer-based solar cell.

2. Inverted Polymer-Based Solar Cells 2.1 Inverted Polymer-Based Solar Cells on Flexible Substrates Many studies on the inverted device architecture have utilized n-type inorganic metal oxides (ZnO, TiO2 ) as the electron selective layer between the active layer and the ITO semi-transparent electrode. However, most of the electron selective layers require high temperature annealing processing conditions in order to improve the crystallinity of the material to minimize resistive losses in the solar cell devices. These high temperature processing conditions can reach as high as 500◦ C which is not compatible with the processing of solar cells on flexible polymer substrates. The use of flexible plastic based substrates requires that the electron selective layer to be processed at low temperatures in order to minimize degradation to the substrate. From this consideration, a method to fabricate and process a room temperature electron selective layer of n-type ZnO nanoparticles (ZnO-NPs) was demonstrated on a flexible inverted polymer solar cell. The feasibility of replacing high temperature processed sol-gel layers of ZnO with a room temperature processed ZnO-NPs layer was studied by X-ray diffraction of the ZnO layers and by fabrication of inverted solar cell devices. Devices on glass/ITO substrates using the high temperature ZnO sol-gel are compared to those fabricated using the ZnO-NPs layer. X-ray diffraction studies were performed on room temperature processed films of the ZnO NPs and the sol-gel processed ZnO after thermal annealing at 400◦ C (Fig. 1). Both the thermal annealed ZnO and ZnO NPs show three typical crystalline ZnO peaks at 31.8◦ (100), 34.3◦ (002), and 36.3◦ (101) indicating that both films have crystallinity regions. Devices were fabricated using these processing conditions for the two electron selective layers. Table 1 shows the average device performance of the inverted solar cells fabricated

Figure 1. X-ray diffraction of ZnO NPs at room temperature and sol-gel processed ZnO after thermal annealing at 400 ◦ C showing three distinct ZnO crystalline peaks at 31.8◦ (100), 34.3◦ (002), and 36.3◦ (101). Adapted with permission from.10 Copyright 2008 American Institute of Physics. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

479

Table 1 Average device performance of un-encapsulated inverted P3HT:PCBM bulk heterojunction solar cells using sol-gel processed or ZnO NPs on ITO/glass and ZnO NPs on ITO/plastic. Adapted with permission from.10 Copyright 2008 American Institute of Physics Substrate Electrode Glass/ITO Ag Glass/ITO Ag Plastic/ITO Ag

ZnO

Voc (V)

Jsc (mA/Cm2)

FF (%)

PCE (%)

ZnO Sol-Gel ZnO-NPs ZnO-NPs

0.62 0.62 0.62

11.1 10.7 9.8

51.3 54.2 53.9

3.5 3.6 3.3

onto ITO/glass substrates on the ZnO sol-gel and ZnO-NPs electron selective layer as well as devices fabricated from the ZnO-NPs onto flexible plastic substrates. The active layer used in these studies is based on a bulk-heterojunction blends of P3HT:PCBM at a weight ratio of 1:0.6 and thermally annealed at 160◦ C. The devices fabricated from the ZnO-NPs on ITO-coated glass show an average PCE of ∼3.6%. This value is very similar to that obtained from the high temperature processed ZnO sol-gel devices on glass/ITO which show an average efficiency of ∼3.5%. This demonstrates that both the sol-gel ZnO and ZnONPs derived layers both behave similarly as a good electron selective layer in the inverted device architecture. To demonstrate that the ZnO-NPs layer can be used onto flexible plastic substrates, inverted devices were fabricated onto ITO-coated plastic substrates showing an average PCE of ∼3.3%. However, a slightly lower efficiency is observed with these devices which is attributed to the lower transparency (∼80%) of the ITO-coated plastic in the region of 500–600 nm compared to ITO-coated glass (∼88%) which reduces the photon flux to the active layer as indicated from the lower Jsc . Nonetheless, the use of a low temperature process electron selective layer with high efficiency onto flexible plastic substrates was demonstrated which is important for the development of roll-to-roll type processing.

2.2 Stability of Inverted Polymer-Based Solar Cells 2.2.1 Air-Stablity of Inverted Polymer-Based Solar Cells. The architecture of the inverted device allows for the use of a more air-stable higher work function top electrode to collect holes such as Ag or Au. Because of the higher work function electrode utilized at the top, the device stability should to be improved. White et al. found that using an Ag electrode in the inverted device architecture did maintain improved stability under nitrogen atmosphere.23 However, unencapsulated inverted solar cells under ambient air exposure conditions have not been studied. To study the effect of the devices under ambient air exposure, unencapsulated inverted solar cells and conventional solar cells were fabricated and periodically tested and stored in air for 40 days (Fig. 2). The conventional device using LiF/Al as the electrode was extremely unstable as its PCE was reduced to less than half of its original value after 1 day of storage and totally degraded after 4 days. The FF and Voc decrease dramatically after 1 day of air exposure due to the interface instability of the electrode from oxidation of the aluminum. The J-V characteristics of the conventional device as a function of storage time in ambient are plotted in Fig. 3(a). The plots show a dramatic decrease in photocurrent in the first couple of days and show negligible photocurrent after 4 days of air exposure. The inset of Fig. 3(a) illustrates the dark current diode characteristics for which the device exposed to 4 days in ambient has a two orders of decrease in the current density at 2V compared to its initial state. The

480

S. K. Hau et al.

Figure 2. Device performance of un-encapsulated conventional and inverted solar cells stored 40 days in air under ambient conditions. (a) Normalized power conversion efficiency (PCE), (b) Shortcircuit current density (Jsc ), (c) Open-circuit voltage (Voc ), (d) Fill factor. Adapted with permission from.10 Copyright 2008 American Institute of Physics. (Figure available in color online)

Figure 3. (a) J-V characteristics of un-encapsulated conventional P3HT:PCBM bulk-heterojunction solar cells over a period of 4 days in air under ambient conditions. (b) J-V characteristics of unencapsulated inverted P3HT:PCBM bulk-heterojunction solar cells (ZnO NPs on ITO-coated plastic substrate) over a period of 40 days in air under ambient conditions. Inset: un-encapsulated dark current device characteristics at 0 days and 40 days in air under ambient conditions. Adapted with permission from.10 Copyright 2008 American Institute of Physics. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

481

inverted devices however, showed high stability to ambient exposure showing constant FF and Voc values over the period of 40 days while the Jsc slightly decreased over this time leading to normalized PCE of over 80%. Figure 3(b) shows a typical flexible inverted unencapsulated device exposed to ambient. The improved in-device stability is attributed to both the PEDOT:PSS layer and Ag electrode. The PEDOT:PSS can effectively prolong oxygen diffusion to the active layer. Additionally, the Ag electrode in air can form a layer of silver oxide thereby increasing its effective work function to −5.0 eV. This matches well with the PEDOT:PSS HOMO of −5.1 eV which improves its electrical coherence at the interface.23,42,43 By having devices stable in air, the encapsulation process can be performed under ambient conditions substantially reducing fabrication complexity leading to cost-effective solar cells. 2.2.2 Thermal Stability of Inverted Polymer-Based Solar Cells. In addition to the ambient device stability, another concern with polymer-based solar cells is the morphological stability of the BHJ blend phases. The phase-separated morphology of the blends are not very thermodynamically stable due the fact that the materials still have a certain degree of freedom to diffuse slowly or recrystallize over time. This is especially true under elevated temperatures which can lead to gradual changes in the nanostructure and microstructure.44,45 The two components will segregate into larger phases leading to reduction in the number of interfaces causing degradation in the device performance. There have been several strategies that have been developed to improve the thermal and morphological stability of P3HT:PCBM BHJ active layers. One is to change the polythiophene backbone to induce a controlled amount of disorder which helps to suppress the crystallization-driven phase segregation between the polymer and fullerene.46,47 Another method is to utilize diblock copolymers additives in the BHJ film to help stabilize the film from phase segregation.48 Another method found to be an effective one to stabilize the nanophase segregration is to utilize similar chemical motifs in both the polymer and fullerene used in the blend system.49 In our approach, amorphous fullerene derivatives are synthesized and utilized as the electron accepting material in P3HT:fullerene based BHJ inverted solar cells.50 The amorphous fullerenes studied are based on modifications to PCBM by replacing the planar phenylene ring with a bulky triphenylamine (TPA) or 9,9-dimethylfluorene (MF). The aim was to suppress the crystallization induced morphological changes that cause device degradation. 2.2.2.1 Electrochemical, Thermal, and Electrical Properties of TPA-PCBM and MFPCBM Derivatives. The electrochemical properties of the amorphous fullerene derivatives were studied by cyclic voltammetry as shown in Fig. 4. The fullerenes all show four quasireversible one-electron reduction waves, which are attributed to the fullerene core. The first reduction potential (E1 red) corresponding to the LUMO level of PCBM is shifted to a more negative value as compared to the parent C60 (Fig. 4) due to the release of strain energy after the introduction of the [6,6] methane substitute on C60 .51–53 The stronger electron-donating properties of the TPA and MF compared to benzene results in a reduction wave shift toward more negative potentials. Differential scanning calorimetry (DSC) of PCBM, TPA-PCBM, and MF-PCBM are shown in Fig. 5. The results show that PCBM has a crystallization peak at 295◦ C with no additional transitions between 20 and 350◦ C. Both the TPA-PCBM and MF-PCBM show a glass transition (Tg ) of 170◦ C and 180◦ C, respectively. One of the more important properties that can drastically influence the performance of polymer-based BHJ solar cell is the mobility of the materials. The electron mobilities of TPA-PCBM and MF-PCBM are compared to PCBM in an n-channel organic field effect transistor (OFET) device configuration. All of the PCBMs showed typical n-type OFET behavior with saturation field-effect electron mobilities of 1.6 × 10−2, 1.1 × 10−2, 5.4 ×

482

S. K. Hau et al.

Figure 4. Cyclic voltammograms of C60 , PCBM, TPA-PCBM and MF-PCBM in 1,2-dichlorobenzene solution. Adapted with permission from.50 Copyright 2009 American Chemical Society. (Figure available in color online)

10−3 cm2/V·s for PCBM, TPA-PCBM, and MF-PCBM respectively. The slight decrease in the mobilities of the two amorphous fullerenes compared to PCBM is attributed to the bulky substituent from the triphenylamine and dimethylfluorene. 2.2.2.2 Device Characteristics of Inverted Polymer-Based Solar Cells with Amorphous Fullerenes. These amorphous fullerenes are compared to PCBM as the electron accepting material in the active layer of inverted devices and the J-V characteristics of these devices are shown in Fig. 6. The conversion efficiencies of the TPA-PCBM and MF-PCBM devices were 4.0% and 3.8%, respectively, which is comparable to those fabricated with PCBM as the n-type material (4.2%). The open-circuit voltage (Voc ) of the TPA-PCBM and MFPCBM was 0.65 V, whereas the PCBM devices only had a Voc of 0.63 V. The 20 mV increase in Voc compared to PCBM is in agreement with what is observed in the shift in LUMO

Figure 5. DSC curves of PCBM, TPA-PCBM and MF-PCBM. Adapted with permission from.50 Copyright 2009 American Chemical Society. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

483

Figure 6. The J-V curves of PCBM, TPA-PCBM and MF-PCBM-based BHJ devices under AM1.5 illumination at 100 mW/cm2. Adapted with permission from.50 Copyright 2009 American Chemical Society. (Figure available in color online)

level as seen by cyclic voltammetry. The slight differences in short circuit current density (9.9 mA/cm2 for TPA-PCBM and 9.8 mA/cm2 for MF-PCBM) as compared to PCBM (10.4 mA/cm2) are attributed to the lower mobility of the two amorphous fullerenes. The thermal stability of these solar cells is examined by annealing the BHJ films at a typical post-treatment temperature of 150◦ C for a time period of 10 min to 10 hours. Figure 7 shows the dependence of the PCE on the annealing time of the three different systems. In the system with PCBM as the acceptor, a gradual degradation in the device performance is observed with prolonged annealing time. The PCE of the PCBM system was 4.2% after 10 min of annealing and gradually decreased to 1.8% after annealing for

Figure 7. Plot of PCE vs annealing time of PCBM, TPA-PCBM and MF-PCBM-based devices annealed at 150◦ C. Adapted with permission from.50 Copyright 2009 American Chemical Society. (Figure available in color online)

484

S. K. Hau et al.

Figure 8. Optical images of (a) P3HT:PCBM, (b) P3HT:TPA-PCBM and (c) P3HT:MF-PCBM films after annealing at 150◦ C for 600 mins. Adapted with permission from.50 Copyright 2009 American Chemical Society. (Figure available in color online)

10 hours. However, both the TPA-PCBM and MF-PCBM show a significantly higher thermal stability showing negligible PCE loss even after 10 hours of annealing. To understand the origin of the improved thermal stability, optical micrographs (Fig. 8) of the different fullerene active layer systems after annealing for 10 hours were taken. The optical image shows severe aggregation and microcrystallites in the P3HT:PCBM films annealed for 10 hours whereas both the TPA-PCBM and MF-PCBM films show no signs of severe phase segregation even after 10 hours of annealing. Atomic force microscopy (AFM) also show severe crystallization in the PCBM blend films whereas both the amorphous fullerene blend films have a surface roughness in the range of 1.3–1.5 nm rms (Fig. 9). These amorphous fullerene compounds show comparable electron mobilities to PCBM in OFETs and high power conversion efficiencies (∼4%) in P3HT:fullerene polymer solar cells. The thermal stability of the polymer solar cells are remarkably enhanced and show no significant degradation in morphology or solar cell performance after annealing at 150◦ C for 10 hours whereas PCBM devices degraded dramatically. These amorphous based fullerene acceptors are attractive candidates for improving the long-term thermal and morphological stability of inverted based BHJ polymer solar cells.

2.3 Interface Modification of Inverted Polymer-Based Solar Cell with Self-Assembled Monolayers Even though reasonable efficiencies have been reached with n-type metal oxides as the electron selective layer in inverted solar, it is known that the surface of metal oxides have hydroxyl groups that have been shown to cause charge trapping at the interface.54 These hydroxyl terminated surfaces lead to high charge carrier recombination due to poor charge

485

Figure 9. AFM images of (a) P3HT:PCBM, (b) P3HT:TPA-PCBM and (c) P3HT:MF-PCBM film annealed at 150◦ C for 600 min. Adapted with permission from.50 Copyright 2009 American Chemical Society. (Figure available in color online)

486

S. K. Hau et al.

transfer leading to losses in device performance. This increase in resistance across these interfaces needs to be minimized by appropriately controlling the electrical contacts (contact resistance, charge transfer, energy level alignment) to allow efficient collection of charges. Additionally, the poor morphological distribution of phases in the bulk-heterojunction active layer may also lead to high resistances in the solar cell due to high recombination of charge carriers. To improve the device parameters (PCE, Jsc , FF, and Voc ) and minimize resistance losses in the active layer, the size and distribution of phases needs to be appropriately controlled. One approach that can both reduce the resistance across the metal oxide interface and affect the active layer morphology is to utilize a self-assembled monolayer (SAM) between the inorganic and organic interface. SAMs can be utilized to significantly modify the interfaces of oxide and metallic surfaces to improve adhesion, compatibility, charge transfer properties, energy level alignment, and affect the upper layer growth of materials.13,24,27,55–58 We have demonstrated that modifying the metal oxide surfaces of TiO2 and ZnO based inverted solar cells with a fullerene-based self-assembled monolayer (C60 SAM) can improve the device performance. The C60 -SAM affects the photoinduced charge transfer at the interface to reduce the recombination of charges, passivate inorganic surface trap states, improve the exciton dissociation efficiency at the polymer/metal oxide interface as well as act as a template to influence the overlayer bulk-heterojunction distribution of phases and crystallinity leading to higher efficiency inverted solar cells.24,27 2.3.1 Inverted Solar Cells: Modification of TiO2 Electron Selective Layers with Self-Assembled Monolayers. 2.3.1.1 Effect of the Self-Assembled Monolayers on Bulk-Heterojunction Solar Cells. The device structure of the TiO2 based inverted polymer solar cells with the different types of surface modified SAMs are shown in (Fig. 10). The J-V characteristics for the inverted bulk-heterojunction cells modified with carboxylic acid SAMs are shown in Fig. 11. A summary of the average device performance modified with and without the monolayers is given in Table 2. The devices without SAM modification have an average power conversion efficiency (PCE) of 2.8% with a Jsc = 9.8 mA/cm2, a Voc = 0.61 V, and a FF = 46.9. When modified with the SAM, the devices show overall improvement in series resistance

Figure 10. Device architecture of the inverted polymer solar cell with the different carboxylic acid based self-assembled molecules used to modify the TiO2 surface. (Right side, top to bottom) C60 based SAM, terthiophene SAM, benzoic acid SAM, and lauric acid SAM.27—Reproduced by permission of The Royal Society of Chemistry. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

487

Figure 11. J-V characteristics of the inverted P3HT:PCBM bulk-heterojunction solar cells modified with and without self-assembled monolayers. (inset: plot of illuminated J-V curve in log scale).27— Reproduced by permission of The Royal Society of Chemistry. (Figure available in color online)

(Rs ) from 13 ·cm2 to 3–4 ·cm2 and shunt resistance (Rsh ) compared to devices without modification. The LA-SAM shows a slight improvement in PCE to 3.0%. The Jsc and Voc remain similar to devices without a monolayer; however, the FF increases slightly to 49.5% which is attributed to the reduced surface traps leading to better contact. Modification with BA-SAM shows further improvement in PCE to 3.2%. The FF is similar to that of devices modified with LA-SAM; however, an increase in Jsc is observed. The Jsc improvement is attributed to BA-SAM improving the interface electron transfer by removing the trap states at the interface of the TiO2 layer as described by Moser et al.59 Modifying the surface with an electroactive functional group like terthiophene or C60 shows efficiency improvements of 0.6% and 1.0%, respectively when compared to devices without modification. A dramatic increase in FF of 56.2% and 57.2% in the TT-SAM and C60 -SAM, respectively is observed indicating that the electroactive groups on the SAM help to reduce the charge recombination Table 2 Summary of the average device performance of inverted P3HT:PCBM bulk-heterojunction solar cells with and without carboxylic acid self-assembled monolayer modification.27—Reproduced by permission of The Royal Society of Chemistry SAM

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

Rs (·cm2)

Rsh (·cm2)

None C60 Terthiophene (TT) Benzoic Acid (BA) Lauric Acid (LA)

0.61 0.62 0.60 0.60 0.61

9.8 10.6 10.0 10.5 9.9

46.9 57.2 56.2 50.2 49.5

2.8 3.8 3.4 3.2 3.0

13 2.4 3.5 2.7 2.6

380 1010 880 580 440

488

S. K. Hau et al.

Figure 12. J-V characteristics of the inverted P3HT:TiO2 heterojunction solar cells modified with and without self-assembled monolayers under AM 1.5 illumination (100 mW/cm2) plotted in linear scale (a) and in the dark plotted in log scale (b). 27—Reproduced by permission of The Royal Society of Chemistry. (Figure available in color online)

losses at the interface and promotes photoinduced charge transfer at the interface. Devices fabricated with the C60 -SAM show a low Rs of 2.4 ·cm2 with average device efficiencies of 3.8%. 2.3.1.2 Effect of Self-Assembled Monolayer on Photoinduced Charge Transfer at the Interface. The photo-induced charge transfer property at the inorganic/SAM/organic interface is investigated by fabricating heterojunction TiO2 /P3HT devices. In this device configuration, the only exciton dissociation site is at the n-type TiO2 /p-type P3HT interface. Therefore, the effect of the SAM on the exciton dissociation efficiency can be independently studied with this type of device. The J-V plots and the device performance of the heterojunction devices are shown in Fig. 12 and Table 3. The data shows that both the LA-SAM and BA-SAM do not have a significant effect on the device performance whereas the TT-SAM and C60 -SAM shows an improvement in all the device characteristics. An improvement in the fill factor and the photocurrent density with the TT-SAM and C60 -SAM in this device configuration confirms that photoinduced charge transfer at the interface plays a role in preventing charge back recombination at the TiO2 interface. Interestingly, the C60 -SAM shows the largest improvement in the overall device performance paramters. The reason for this is that the C60 molecule is a very good electron acceptor (n-type) under Table 3 Summary of the average device performance of inverted P3HT:TiO2 heterojunction solar cells with and without carboxylic acid self-assembled monolayer modification.27— Reproduced by permission of The Royal Society of Chemistry SAM None C60 Terthiophene (TT) Benzoic Acid (BA) Lauric Acid (LA)

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

Rectification Ratio

0.08 0.37 0.19 0.12 0.07

0.45 0.65 0.56 0.43 0.49

31.5 49.1 44.1 35.7 30.6

0.01 0.12 0.05 0.02 0.01

4 × 10−2 3 × 10−4 2 × 10−3 2 × 10−2 4 × 10−2

Development of Inverted Polymer Solar Cell Architecture

489

illumination due to photoinduced electron transfer from polymer to C60 .60 The improvement of the BHJ blend system can also be partly explained by this photoinduced electron transfer from P3HT to the C60 -SAM, since both P3HT and PCBM can be located at this interface. Furthermore, the large rectification ratio ∼10−4 of the C60 -SAM modified device at ±3 V compared to the other SAMs indicates that the C60 -SAM serves as an effective hole blocking layer at the interface. 2.3.1.3 Self-Assembled Monolayer Effect on P3HT Polymer Crystallinity. The effect of the SAM modification on the overlayer crystallinity of the P3HT:PCBM BHJ blend was studied by XRD. The X-ray diffraction spectrum of the bulk-heterojunction blend prior to and after annealing at 150◦ C on devices with and without SAMs modification are shown in Fig. 13. The diffraction spectrum shows a typical P3HT (100) intensity peak in which the intensity differs depending on the SAM and whether the BHJ was annealed. An overall increase in diffraction intensity is observed in the BHJ films after thermal annealing which is a typical observation of better P3HT ordering. The P3HT diffraction intensity of devices modified with BA-SAM are lower than those without a monolayer showing a negative influence on the overlayer crystallinity implying that improved device performance from the BA-SAM is attributed mainly to the improved charge transfer and lower contact resistance at the interface. The TT-SAM devices show an increase in P3HT diffraction signal indicating that the improvement in crystallinity in addition to the better charge transfer properties at the interface lowers contact resistance and improves the device efficieny. Diffraction signals of the LA-SAM are higher, indicating better P3HT ordering; however, the device efficiencies are poor. The improved order is due to the long alkyl chain SAM interacting with P3HT to help order the P3HT chains.61 The slight improvement in performance is due to the improved crystallinity of the BHJ film which lowers the series resistance of the device. This improvement is not likely to be due to improved charge transfer at the interface since lauric acid is not electroactive and will act as a physical barrier that affects the electronic coupling between the polymer and TiO2 . C60 -SAM modified devices

Figure 13. X-ray diffraction spectrum of the P3HT (100) peaks before and after thermal annealing at 150◦ C on bulk-heterojunction devices modified with and without the SAM.27—Reproduced by permission of The Royal Society of Chemistry. (Figure available in color online)

490

S. K. Hau et al.

show the largest increase in the diffraction signal confirming that it can affect the overlayer crystallinity and morphology. The combined improvements in P3HT crystallinity, contact resistance by passivating surface traps, and photo-induced charge transfer to reduce back charge recombination are reasons for the high performance of these devices. 2.3.1.4 Capacitance-Voltage and Field Effect Transistor Measurements on BHJ Films. The effect of the C60 -SAM on the interface between the organic active and inorganic TiO2 layer can be studied with organic field effect transistors (OFET) and capacitancevoltage (C-V) measurements. C-V and OFET measurements of P3HT:PCBM films can give information regarding the distribution and vertical concentration gradients of the organic materials close to the interface.62 The transport properties in OFETs are generally determined by the first few nanometers of the active material close to the interface, therefore if one of the two components of the blend segregates towards the bottom interface, the charge carrier mobility should ideally reach its pristine value. OFETs were fabricated of the pristine films of P3HT and PCBM as well as P3HT:PCBM blend films with and without C60 -SAM modification on a TiO2 /SiO2 /doped-Si substrate. The transfer characteristics of the pristine films of P3HT and PCBM are given in (Fig. 14(a)) showing a hole mobility of 1.8 × 10−4 cm2/V·s for P3HT and an electron mobility of 1.2 × 10−4 cm2/V·s for PCBM. BHJ blend films without SAM modification had a hole mobility of 4.1 × 10−5 cm2/V·s and an electron mobility of 9.4 × 10−6 cm2/V·s. The higher hole mobility compared to the electron mobility indicates that a higher concentration of P3HT is accumulated at the interface which is in agreement with previous reports of blend films cast from di-chlorobenzene showing higher concentrations of P3HT near the bottom interface.63 However, BHJ films with C60 -SAM modification show a hole mobility of 1.8 × 10−5 cm2/V·s and an electron mobility of 1.0 × 10−4 cm2/V·s. The electron mobility in this case is higher than the hole mobility confirming that more PCBM is accumulated at the interface. To further confirm this, C-V measurements (Fig. 14(b)) also indicate that the C60 -SAM facilitates the accumulation of PCBM to the bottom TiO2 interface. The C60 -SAM modified devices show p-mode and n-mode accumulation that is in agreement with what was found from the OFET measurements. C-V measurements of devices without the SAM show higher capacitance under negative gate voltage implying a more p-mode accumulation at the interface and

Figure 14. (a) Square-root of the measured transfer characteristics of OFETs prepared from the pristine P3HT, PCBM and blend films of P3HT:PCBM (weight ratio 1:0.8) with and without C60 SAM modification at the TiO2 interface. (b) Capacitance-voltage measurement of blend films of P3HT:PCBM (weight ratio 1:0.8) with and without C60 -SAM modification at the TiO2 interface.27— Reproduced by permission of The Royal Society of Chemistry. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

491

show that no significant n-mode accumulates under positive gate bias. The C60 -SAM can help to nucleate PCBM to provide a better percolation conduction pathway of PCBM at the interface to improve the overall distribution of phases, morphology, and crystallinity of the blend leading to better charge selectivity. The C60 -SAM leads to a 35% improvement in device PCE over unmodified TiO2 inverted devices. Understanding how to reduce the interface and bulk-heterojunction resistance in a PV cell is critical in reducing losses in PV performance caused by a recombination of charge carriers. 2.3.2 Inverted Solar Cells: Modification of ZnO Electron Selective Layers with SelfAssembled Monolayers. The high electron mobility of ZnO compared to titanium oxide makes it a better electron selective contact layer for inverted solar cells due to the lower resistances. However, interface resistances caused by charge traps attributed to the hydroxylated ZnO surface can still lead to a high charge carrier recombination; therefore, modification of the surface with a SAM is still necessary. The ZnO inverted device architecture and structure of the C60 -SAM are shown in Fig. 15. The J-V characteristics of the bulk-P3HT/PCBM heterojunction (BHJ) and P3HT/ZnO heterojunction (HJ) inverted devices with and without C60 -SAM modification and device performance parameters are shown in Fig. 16 and Table 4. BHJ devices without SAM modification had an average power conversion efficiency (PCE) of ∼3.7%. The performance of the inverted solar cell is improved by modifying the ZnO surface with a C60 -SAM. The PCE is improved by over 20% compared to the unmodified devices giving an average PCE of 4.5% and the highest PCE of 4.9%. An improvement in both the FF from 55.4% to 60.6% and Jsc from 10.8 mA/cm2 to 12.0 mA/cm2 is observed after modification with the C60 -SAM. This improvement is attributed to better electronic coupling of the inorganic/organic interface from the C60 -SAM through the process of photo-induced charge transfer as similarly observed in the TiO2 based C60 -SAM modified inverted devices. This helps mediate forward charge transfer and reduces the back charge recombination at the interface leading to improved photocurrent and charge selectivity. Bilayer heterojunction devices of ZnO modified with and without the C60 -SAM and pristine P3HT were fabricated showing an almost two times improvement in power conversion efficiency in the C60 -SAM devices over unmodified HJ devices. An improvement in both the FF and Jsc from 46.7% to 54.2% and from 1.1 mA/cm2 to 1.8 mA/cm2 is observed. A lower Voc is observed with the C60 -SAM modified devices as compared to the unmodified device due to the fact that exciton dissociation happens at the C60 -SAM/P3HT

Figure 15. Device structure and chemical structure of ZnO-NPs based inverted solar cell with C60 SAM modification. Adapted with permission from.24 Copyright 2008 American Institute of Physics. (Figure available in color online)

492

S. K. Hau et al.

Figure 16. (a) J-V characteristics of inverted ZnO NP/P3HT:PCBM bulk-heterojunction solar cells with and without C60 -SAM modification. (b) J-V characteristics of inverted ZnO NP/P3HT heterojunction solar cells with and without C60 -SAM modification. Adapted with permission from.24 Copyright 2008 American Institute of Physics. (Figure available in color online)

interface and not the ZnO/P3HT interface. It is known that the Voc between fullerenes and P3HT is typically less than 0.7 V while between ZnO and P3HT can be greater than 0.7 V.64 Figure 17 shows the results of the external quantum efficiency (EQE) measurements on the BHJ and HJ devices modified with and without the C60 -SAM. The EQE for the BHJ devices without modification shows a maximum of ∼49% at 500 nm while after C60 -SAM modification, the EQE reaches ∼70%. A similar trend is observed with the HJ devices showing an increase in EQE from ∼11% to 17% after modification with the C60 -SAM. 2.4 Processing Optimization of the BHJ Active Layer for Inverted Polymer Solar Cells The processing conditions of the active layer film can have a dramatic effect on the final device efficiency. The processing conditions of the blend films that can affect the device efficiency are the blend ratio of donor and acceptor materials, the active layer film thickness, the thermal annealing temperature, and the thermal annealing time. By varying these processing conditions in C60 -SAM modified inverted solar cells, it is found that the optimum active layer efficiency is ∼4.5% using a P3HT:PCBM blend ratio of 1:0.7, an active layer thickness of ∼200 nm at an annealing temperature and time of 160◦ C for 10 min.65 Table 4 Summary of the average device performance of inverted ZnO-NPs/P3HT:PCBM bulkheterojunction solar cells and ZnO-NPs/P3HT heterojunction solar cells with and without C60 -SAM modification. Adapted with permission from.24 Copyright 2008 American Institute of Physics Active Material

SAM

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

P3HT:PCBM P3HT:PCBM P3HT P3HT

None C60 None C60

0.63 0.63 0.75 0.67

10.8 12.0 1.08 1.83

55.4 60.6 46.7 54.2

3.7 4.5 0.37 0.67

Development of Inverted Polymer Solar Cell Architecture

493

Figure 17. External quantum efficiencies of ZnO-NPs/P3HT:PCBM bulk-heterojunction and ZnONPs/P3HT heterojunction inverted solar cells with and without C60 -SAM modification. Adapted with permission from.24 Copyright 2008 American Institute of Physics. (Figure available in color online)

2.4.1 BHJ Donor/Acceptor Ratio Dependence on Performance of Inverted Solar Cells. The ratioFigure 17 of donor and acceptor materials in BHJ blends influence the size and distribution of phases leading to changes in the number of interfaces for exciton dissociation and charge carrier transport. Reduction in the donor and acceptor interfaces lowers the probability for exciton dissociation while increasing the probability for charge carrier recombination due to the lower mobilities of organic materials. A balanced electron and hole charge transport throughout the blends minimizes these recombination occurrences and improves the solar cell performance. Generally, the ratio of P3HT: PCBM that has been found to have high efficiencies in the conventional architecture in the range of 1:0.8 to 1:1. Table 5 and Fig. 18 summarize the results of the inverted devices with and without the C60 -SAM modification varying the blend ratios keeping the annealing temperature and time at 160◦ C and 10 min. Devices fabricated without PCBM (P3HT/ZnO-NPs heterojunctions) show a PCE of 0.2%, but when modified with a C60 -SAM, the PCE is improved to 0.3% due to the improvement in Jsc and FF. The improvement in FF and Jsc in the C60 -SAM devices is attributed to the improved charge transfer properties from P3HT to ZnO. The resistance at the ZnO interface dominates the device parameters since the only interface for exciton separation is at the polymer/ZnO-NPs interfaces. Devices with P3HT:PCBM ratios of 1:0.1, 1:0.2, and 1:0.3 show similar FF for both the C60 -SAM modified and unmodified devices. The similar FF of both modified and unmodified devices indicates that the active layer now dominates the resistances in the solar cell. The Voc for the unmodified devices are low (0.33 V, 0.38 V, 0.49 V) compared to ones modified with the C60 -SAM (0.41 V, 0.50 V, 0.58 V) in these blend ratios. The addition of the PCBM into the BHJ blend adds additional interfaces that also play a role in determining the Voc . The low concentration of PCBM in these blends may not provide a good morphology and percolation pathway thus leading to low Voc . At the blend ratio of 1:0.4 (∼29% PCBM content), the resistances in the bulk active layer (donor-acceptor) and across the interfaces (ZnO) are comparable, thus changes at ZnO interfaces can have a major effect on the solar cell performances. However, the Voc of devices are independent of ZnO modification indicating that the amount of

494

S. K. Hau et al.

Table 5 Average device performance of inverted ZnO-NPs/P3HT:PCBM bulk-heterojunction solar cells fabricated with different active layer blend weight ratios from 1:0 to 1:1 (w:w). Parentheses indicate average device performance of inverted solar cells modified with the C60 -SAM. Adapted with permission from.65 ©2010EEE Ratio

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

1:0 1:0.1 1:0.2 1:0.3 1:0.4 1:0.5 1:0.6 1:0.7 1:0.8 1:0.8 1:1

0.60 (0.43) 0.33 (0.41) 0.38 (0.50) 0.49 (0.58) 0.63 (0.64) 0.63 (0.64) 0.63 (0.63) 0.62 (0.62) 0.62 (0.62) 0.61 (0.61) 0.61 (0.62)

0.5 (1.2) 1.3 (1.8) 2.2 (3.4) 4.5 (6.2) 9.0 (10.3) 10.6 (10.8) 10.8 (11.0) 10.6 (11.2) 10.5 (10.8) 10.1 (10.2) 10.0 (10.3)

48.7 (52.6) 46.6 (47.2) 44.8 (44.4) 43.6 (43.9) 47.7 (52.2) 55.6 (60.0) 59.5 (61.8) 60.8 (64.2) 62.2 (64.8) 62.9 (65.2) 62.5 (62.6)

0.2 (0.3) 0.2 (0.4) 0.4 (0.8) 1.0 (1.6) 2.7 (3.5) 3.7 (4.1) 4.0 (4.3) 4.0 (4.5) 4.1 (4.3) 3.9 (4.1) 3.8 (4.0)

Figure 18. Plots of the four device parameters (a) Voc , (b) Jsc , (c) FF, and (d) PCE as a function of P3HT:PCBM blend ratio (w:w) (from 1:0 to 1:1) in inverted ZnO-NPs/P3HT:PCBM solar cells with and without C60 -SAM modification. Adapted with permission from.65 ©2010EEE. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

495

blend P3HT:PCBM interfaces are large enough to lead to good distribution of phases to minimize charge recombination. Increasing the ratio from 1:0.3 to 1:0.4, the efficiency increases from 1.0% to 2.7% in unmodified devices and from 1.6% to 3.5% in C60 -SAM modified devices. The unmodified devices still show low fill factor and Jsc (47.6%, 9.0 mA/cm2) indicating resistances in the solar cell, but after modification with a C60 -SAM, the fill factor and Jsc are improved to 52.2% and 10.3 mA/cm2. This shows that the SAM is influencing the blend active layer morphology and reducing the resistance losses caused from carrier recombination. Increasing the blend ratio from 1:0.5 to 1:1 show saturation in the device Voc in both the modified and unmodified devices. The C60 -SAM devices at higher blend ratios show improvements in Jsc and FF compared to devices without modification. The highest PCE reached ∼4.5% at blend ratios of 1:0.7. At these ratios, the C60 -SAM effect on the bulk morphology to improve solar cell efficiency is minimized. Therefore, the improvement from the C60 -SAM at these higher ratios is mainly due to the improved charge transfer properties at the interface. 2.4.2 Influence of BHJ Film Thickness on Inverted Solar Cell Performance. Another parameter that can influence the performance of the inverted solar cell is the BHJ film thickness. Inverted solar cells from solution concentrations of 10 mg/mL to 60 mg/mL were spun at 1500 rpm and its effect on the four device parameters are show in Fig. 19. The optimization of the BHJ film thickness in these inverted solar cells utilized a blend ratio of 1:0.7 and an annealing condition of 160◦ C for 10 min. The Voc using a 10 mg/mL concentration is low ∼0.53 V, but as the solution concentration increases (increasing film thickness), the Voc increases and remains similar ∼0.61–0.63 V. The lower Voc in the thin layer may be attributed to the formation of more current leakage pathways in the film causing higher recombination of charges. As the BHJ film thickness is increased, the Jsc also increases due to the higher optical absorbance of the film generating more charge carriers. With a thin BHJ film (10 mg/mL), the FF is very poor, but increasing the film thickness leads to fill factors greater than 60%. However, increasing the thickness further leads to a decrease in fill factor due to the high resistance of the active layer from the poor charge carrier transport of the organic materials. In the ideal case, the BHJ film thickness should both maximize photon absorbance and still maintain high carrier mobility to allow maximum collection of the charges. It is found that the BHJ film thickness of ∼200 nm (40 mg/mL) shows the optimum PCE in these inverted solar cells. 2.4.3 Influence of Thermal Annealing Temperature on the Performance of Inverted Solar Cells. Another processing parameter that can influence the performance of solar cells is the thermal annealing of the BHJ blend. Annealing can change the blend morphology and segregation of donor/acceptor phases which can influence the number of interfaces for exciton dissociation. The device performance of inverted solar cells with thermal annealing temperatures from 100–200◦ C using a BHJ film thickness of ∼200 nm, a blend ratio of 1:0.7 and an annealing time of 10 min were studied (Fig. 20). The annealing temperature showed very minimal effect on Voc until 200◦ C where it decreased to ∼0.57 V. The annealing temperature shows a maximum Jsc in devices with annealing temperatures ∼130◦ C to 160◦ C. Increasing the annealing temperature further leads to a lowering of the Jsc . The FF however shows a dramatic change with annealing temperature. Temperatures below 120◦ C lead to fill factors below 50%. Further increasing the temperature from 130◦ C to 170◦ C leads to fill factors as high as 65%. However, increasing temperatures beyond 170◦ C leads to a reduced fill factor. This drastic change in the fill factor is explained by the evolution of the blend morphology as characterized by AFM (Fig. 21). Low annealing temperatures are

496

S. K. Hau et al.

Figure 19. Plots of the four device parameters (a) Voc , (b) Jsc , (c) FF, and (d) PCE as a function of solution concentration of P3HT:PCBM blends spincoated at 1500 rpm (changes in thickness of active layer) in inverted ZnO-NPs/ C60 -SAM/P3HT:PCBM solar cells. Adapted with permission from.65 ©2010EEE.

insufficient to allow the P3HT and PCBM to interact strongly which causes poor mobility in the phases of the organic materials leading to high carrier recombination. At elevated temperatures (130◦ C to 160◦ C), the thermal energy required for both the P3HT and PCBM to pack and interact is sufficient to allow the phases to segregate into the optimum nanophase morphology. At higher temperatures (170◦ C to 200◦ C), the size of the phases begin to grow as indicated by the AFM images. These larger phases lower the number of interfaces for exciton dissociation causing higher charge carrier recombination therefore decreasing the Jsc and FF. The optimum annealing temperature leading to the PCE is at 160◦ C. 2.4.4 Influence of Thermal Annealing Time on the Performance of Inverted Solar Cells. In addition to the annealing temperature, the annealing time can also influence the final performance of polymer-based solar cells. Based on previously optimizing the BHJ blend ratio, BHJ film thickness, and annealing temperature device using a blend ratio of 1:0.7, thickness ∼200 nm, and annealing temperature ∼160◦ C were studied. The BHJ layers were varied from no annealing to 60 min of thermal annealing as shown in Fig. 22. Inverted devices without annealing and annealed for 60 mins show changes in Voc from ∼0.60 V to 0.62 V and Jsc from 9.5 mA/cm2 to 10.3 mA/cm2. The fill factor is found to be sensitive to the annealing time having fill factors lower than 45% when the active layer

Development of Inverted Polymer Solar Cell Architecture

497

Figure 20. Plots of the four device parameters (a) Voc , (b) Jsc , (c) FF, and (d) PCE as a function of thermal annealing temperatures of P3HT:PCBM blends from no annealing to 200◦ C in inverted ZnO-NPs/ C60 -SAM/P3HT:PCBM solar cells. Adapted with permission from.65 ©2010EEE.

is not annealed. After 1 min of annealing, the fill factor increases to ∼57% and further increases to over 65% after 5 mins of thermal annealing. With 10 mins of annealing, the PCE showed optimum performance with efficiencies over 4%. Annealing for longer times (60 min) show very little change in efficiency when compared to devices annealed at 10 min indicating that shorter annealing times at a annealing temperature of 160◦ C is sufficient to optimize the BHJ blend morphology. Optimizing the processing conditions for the BHJ active layer allows understanding of the processing control required to maintain high device performances for potential large-scale production of these inverted based solar cells.

2.5 Optimization of Electrodes in Inverted Polymer Solar Cells 2.5.1 Influence of the Top Metal Anode Electrode on the Performance of Inverted Polymer Solar Cells. The performance of polymer-based solar cells can also be influenced by the nature of charge collection at the interface of the active layer and electrode. For a single component diode, the metal-insulator metal (MIM) model suggests that the Voc is given by the difference between the work functions of the anode and cathode.66,67 This holds as long as one or both of the contacts is non-Ohmic in nature. If the work function of either electrode coincides with or is smaller than the LUMO or greater than the highest occupied

498

Figure 21. AFM images of thermally annealed inverted ZnO-NPs/ C60 -SAM/P3HT:PCBM solar cells blends from no annealing to 200◦ C with the PCBM phase removed by 1,8-octanedithiol. Images show an evolution of the P3HT phases into larger phases with higher annealing temperatures. The scan size for the AFM images are 0.5 µm × 0.5 µm. Adapted with permission from.65 ©2010EEE. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

499

Figure 22. Plots of the four device parameters (a) Voc , (b) Jsc , (c) FF, and (d) PCE as a function of thermal annealing time of P3HT:PCBM blends from no annealing to 60 min annealing in inverted ZnO-NPs/ C60 -SAM/P3HT:PCBM solar cells. Adapted with permission from.65 ©2010EEE.

molecular orbital (HOMO) of the semiconductor, the Fermi level of that electrode pins at the appropriate energy level. Under these conditions, the Voc cannot exceed the band-gap of the polymer. However, with the incorporation of an electron acceptor material in a BHJ system, Voc is limited to the difference between the HOMO of the donor and the LUMO of the acceptor. Different top metal contacts for inverted solar cells are examined and the device characteristics are summarized in Table 6. Device fabricated with the top anode electrode using Ag, Cu, Au, and Pd show very similar PCE and almost no change in Voc . However, with inverted devices fabricated with Al and Ca/Al show a lower PCE. The Jsc of the Al and Ca/Al devices were 6.7 and 0.2 mA/cm2, respectively. These values are much lower than the devices using higher work function electrodes which had average Jsc of ∼10 mA/cm2. X-ray photoelectron spectra show that Al reacts to form chemical bonds with the sulfonic acid moiety of PEDOT:PSS 68 resulting in a thin insulating layer. Calcium is also susceptible to moisture and likely oxides very rapidly in air leading to similar interfacial degradation. Higher work function electrodes such as Au and Pd show inverted solar cell efficiencies of 3.9% and 3.8%, respectively. The best device in this series of electrodes uses Ag as the anode with average efficiencies over 4%. Interestingly, Cu as an anode electrode also shows similar device performance having efficiencies of 3.8%. The data demonstrates

500

S. K. Hau et al.

Table 6 Average device performance of inverted ZnO-NPs/P3HT:PCBM bulk-heterojunction solar cells fabricated with different top metal electrodes. Adapted with permission from.65 ©2010EEE Anode Electrode Pd Au Cu Ag Al Ca/Al

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

0.63 0.62 0.62 0.62 0.61 0.61

9.7 10.2 10.0 10.3 6.7 0.2

61.9 61.7 61.6 62.3 60.5 46.2

3.8 3.9 3.8 4.0 2.5 0.1

that given the proper device architecture, even inexpensive electrodes can be utilized to generate high efficiency solar cells. This has implications for reducing fabrication cost for large-scale fabrication of polymer-based solar cells. 2.5.2 Solution Processed, Vacuum-Free Electrodes for Inverted Polymer Solar Cells. 2.5.2.1 Silver Nanoparticles as the Top Metal Anode in Inverted Polymer Solar Cells. Ideally, all the processing of the active layers and electrodes as well as encapsulation should be done under ambient conditions. However, this cannot be achieved with the typical conventional architecture which requires both high vacuum for metallization and an inert environment for encapsulation to produce the final solar cell. The inverted architecture has been shown previously to maintain high efficiency even with the use of higher work function electrodes. These higher work function metal electrodes like Ag can be done by printing and coating techniques which can be easily integrated into the roll-to-roll type processing. In our inverted device architecture, PEDOT:PSS is utilized between the active layer and the top metal electrode. Because PEDOT:PSS provides a good protective layer to the active layer, a method was developed to process solution processible Ag-nanoparticles as the top metal electrode in inverted solar cell devices. The top metal electrode in these inverted solar cell devices was processed by spraycoating a solution of the Ag-nanoparticles through a mask to define the device area. The spraycoating process and final coating can be controlled by changing variables including the solution viscosity, spray pressure, the spray distance from sample, spray time, and the number of spray coats. In this study, the variables other than the number of coats were held constant. The J-V characteristics and device performance parameters for the different number of Ag-nanoparticle coating layers is shown in Fig. 23 and Table 7. Devices with a coating of 20 layers show a low Jsc of 7.7 mA/cm2, FF of 44.5%, and PCE of 2.1%. Increasing the coating layer to 100 coats shows a saturation of Jsc to 8.3 mA/cm2. A significant improvement in the FF is observed from 20 coats (44.5%) to 100 coats (59.6%) which are attributed to the reduced resistance of the Ag-nanoparticle electrode with the thicker coating. Devices fabricated with the 100 layer coating of the Ag-nanoparticles show an average device efficiency of ∼3.0%. Devices using a vacuum deposited Ag electrode have a Jsc , FF, and PCE of 9.3 mA/cm2, 62.2% and 3.6% respectively. The Ag spraycoated electrode has a 0.6% difference in PCE compared to the device with the evaporated Ag electrode.

Development of Inverted Polymer Solar Cell Architecture

501

Figure 23. Illuminated J-V characteristics of inverted ZnO NP/C60 -SAM/P3HT:PCBM bulkheterojunction solar cells fabricated on glass/ITO by spraycoating Ag-NPs (from 20–100 coats) or evaporating Ag as the anode electrode. Adapted with permission from.14 Copyright 2009 Elsevier. (Figure available in color online)

Electrical (four-point probe) and optical (UV-Vis, SEM) characterization of the Ag spraycoated and Ag evaporated electrodes were compared to rationalize the reason for the differences in efficiency. Electrical four-point probe measurements and optical SEM images on a 20 layer coating show a poor Ag-nanoparticle interconnectivity which is the reason for the low efficiency from these devices (Fig. 24(b)). Increasing the layer coating to 40 gave a sheet resistance of 7.55 /sq while further increasing it to 100 lead to a sheet resistance of 0.71 /sq which is comparable to the evaporated electrode (0.51 /sq). The lower sheet resistance from the 100 coating layer of the Ag-nanoparticles is the reason for the similar fill factors in both the Ag spraycoated and Ag evaporated electrodes. This still does not explain the difference in efficiency between the 100 layer coating and Ag evaporated electrode. To Table 7 Average device performance of inverted ZnO NP/C60 -SAM/P3HT:PCBM bulk heterojunction solar cells fabricated on glass/ITO and plastic/ITO substrates by either vacuum deposition of Ag or spraycoating of Ag-NPs as the anode electrode. Sheet resistance of vacuum deposited Ag and different spraycoated Ag-NPs electrodes. Adapted with permission from.14 Copyright 2009 Elsevier Electrode Deposition

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

Ag Sheet Resistance (/)

Vacuum Vacuum (Plastic) Spraycoat (20 coats) Spraycoat (40 coats) Spraycoat (60 coats) Spraycoat (80 coats) Spraycoat (100 coats) Spraycoat (100 coats) (Plastic)

0.61 0.61 0.61 0.61 0.61 0.61 0.61 0.60

9.3 8.3 7.7 8.2 8.2 8.2 8.3 6.9

62.2 34.3 44.5 54.2 56.6 58.1 59.6 34.0

3.6 1.7 2.1 2.7 2.8 2.9 3.0 1.4

0.51 – – 7.55 3.47 1.84 0.71 –

502

S. K. Hau et al.

Figure 24. (a) Photograph of polymer solar cells using an Ag spraycoated electrode (left) and Ag evaporated electrode (right). SEM images of (b) spraycoated Ag-NPs (20 coats) after annealing at 120◦ C for 5 min; (c) spraycoated Ag-NPs (100 coats) after annealing 120◦ C for 5 min; (d) 100 nm evaporated Ag. Scale bar: 200 nm. Adapted with permission from.14 Copyright 2009 Elsevier. (Figure available in color online)

explain the reduced efficiency, UV-Vis measurements were performed on the electrodes in the range from 300–800 nm. The transmission near the P3HT absorbance peak (∼550 nm) of the different layer coatings show a transparency of ∼45% for a 20 layer coating (Fig. 25). After increasing the layer coating to 80 and 100, the transmission at 550 nm is reduced to ∼12% and ∼9%, respectively. These values are still higher than the evaporated Ag film at 550 nm which is ∼1%. SEM images of the Ag-NPs film with a layer coating of 100 compared to the Ag evaporated film show a less dense film morphology (Fig. 24(d)). The smoother and denser Ag evaporated film improves the reflectivity of Ag which improves the photocurrent density of the solar cell. This spraycoating technique was also demonstrated onto flexible ITO substrates showing efficiencies as high as 1.4%. This is comparable to Ag evaporate electrodes on these flexible ITO substrates which showed efficiencies of 1.7%. This approach removes the need to use high vacuum to deposit the top electrode allowing the possibility for roll-to-roll fabrication of solar cells. 2.5.2.2 Indium Tin Oxide-Free Inverted Solar Cells using Conducting Polymers as Electrodes. Indium tin oxide (ITO) has been utilized as the semi-transparent conducting electrode for development of organic solar cell. However, ITO is typically processed at elevated temperatures to improve crystallinity and conductivity, therefore the process is not ideal for flexible roll-to-roll solar cell fabrication. ITO with lower conductivity can be processed onto flexible plastic substrates, but the brittleness of ITO leads to significant degradation in conductivity and device performance due to the formation and propagation

Development of Inverted Polymer Solar Cell Architecture

503

Figure 25. UV-vis transmission spectrum of spraycoated Ag-NPs electrodes from 20 to 100 coats compared to a 100nm Ag evaporated electrode on glass substrates. Adapted with permission from.14 Copyright 2009 Elsevier. (Figure available in color online)

of cracks. The increasing costs of indium may also prevent large-scale usage of ITO as an electrode material for low-cost polymer-based solar cells. Therefore replacement of ITO as the semi-transparent electrode is needed. A particular organic based electrode material of interest is conducting PEDOT:PSS due to its solution processibility which makes them compatible with the concept of large-scale roll-to-roll processing. The use of PEDOT:PSS as a semi-transparent electrode to replace ITO is evaluated by measuring both the transparency and sheet resistance of different thicknesses of PEDOT:PSS processed with DMSO (Fig. 26).69 UV-Vis spectroscopy of various thicknesses

Figure 26. Transparency vs. wavelength of varying thicknesses of PEDOT:PSS −5% DMSO electrodes on glass as compared to transparency of ITO on glass as referenced against air. Legend also indicates corresponding sheet resistances obtained as measured by four-point probe. Adapted with permission from.69 Copyright 2009 Elsevier. (Figure available in color online)

504

S. K. Hau et al.

Figure 27. Device architectures of the inverted solar cells with ITO or PEDOT:PSS-DMSO electrodes. Adapted with permission from.69 Copyright 2009 Elsevier. (Figure available in color online)

of PEDOT:PSS show that a ∼130 nm thick PEDOT:PSS layer has comparable transparency to the ITO. The sheet resistance of the PEDOT:PSS layer however, is ∼25 times higher than that of ITO (15 /). Increasing the PEDOT:PSS thickness to ∼220 nm leads to a reduction in sheet resistance to as low as 160 /, but also shows a decrease in transparency due to the absorbance from the PEDOT:PSS film. Inverted solar cells fabricated with different electrode PEDOT:PSS thicknesses is compared to devices fabricated from ITO. The device configuration of these two architectures is shown in Fig. 27. The J-V characteristics and device performance parameters are shown in Fig 28(a) and Table 8. Devices using a thin PEDOT:PSS layer (∼40 nm) have an average Jsc of 9.6 mA/cm2, FF of 39.4% and a PCE of 2.3%. A PEDOT:PSS electrode thickness of ∼130 nm improved the overall PCE to ∼3.1%. However, a lower average Jsc of 9.4 mA/cm2 was observed while an increase in FF to 53.3% is observed. Further increasing the

Figure 28. (a) Illuminated J-V characteristics of inverted devices fabricated from various thicknesses of PEDOT:PSS −5% DMSO as cathode electrodes on glass as compared to ITO-based cathode electrodes. (b) Illuminated J-V characteristics of inverted devices fabricated from PEDOT:PSS −5% DMSO cathode electrodes on plastic substrates and ITO-based cathode electrodes on plastic substrates. Adapted with permission from.69 Copyright 2009 Elsevier. (Figure available in color online)

Development of Inverted Polymer Solar Cell Architecture

505

Table 8 Average device performance of inverted ZnO NP/C60 -SAM/P3HT:PCBM bulkheterojunction solar cells fabricated using glass/PEDOT:PSS, plastic/PEDOT:PSS, glass/ITO and plastic/ITO as the cathode electrode and vacuum deposited Ag as the anode electrode. Adapted with permission from.69 Copyright 2009 Elsevier Electode (cathode)

Electrode (anode)

Voc (V)

Jsc (mA/cm2)

FF (%)

PCE (%)

ITO/glass PEDOT:PSS (∼40 nm)/glass PEDOT:PSS (∼130 nm)/glass PEDOT:PSS (∼175 nm)/glass PEDOT:PSS (∼220 nm)/glass ITO/plastic PEDOT:PSS (∼175 nm)/glass

Ag Ag Ag Ag Ag Ag Ag

0.62 0.61 0.61 0.61 0.61 0.61 0.62

10.3 9.6 9.4 8.6 8.2 9.9 8.4

66.6 39.4 53.3 57.2 60.0 61.2 57.7

4.2 2.3 3.1 3.0 3.0 3.7 3.0

thickness of the PEDOT:PSS films lead to a saturation in PCE to ∼3.0%. Interestingly, the thicker PEDOT:PSS electrodes lead to higher FF, but a reduction in device Jsc . This trend can be correlated to the lower transparency and reduced sheet resistance with increasing layer thickness. The transparency and sheet resistance of the electrodes are an important consideration for the development of semi-transparent electrode materials in solar cells as they can affect both the photocurrent and fill factor. Lower transparency leads to a reduction of the potential photons that can be absorbed by the active layer therefore leads to lower photocurrent densities. Lower sheet resistance minimizes the resistive losses in the solar cells and improves fill factor. The charges lost from the lateral charge collection through the PEDOT:PSS electrode become much more apparent with higher sheet resistance. A dramatic improvement in the fill factor from 39.4% (∼40 nm PEDOT:PSS) to 60.0% (∼220 nm PEDOT:PSS) is observed by changing the sheet resistance from 2800 / to 160 /. ITO/glass based solar cells had an average Jsc = 10.3 mA/cm2, FF = 66.6%, and PCE of ∼4.2%. The higher efficiency of the ITO substrates is due to the higher transparency (∼10% higher) and lower sheet resistance (∼10 times less) when compared to the ∼220 nm thick PEDOT:PSS electrode. Devices using PEDOT:PSS as the electrode in flexible substrates show similar performance (PCE ∼3%) to that of solar cells fabricated onto glass substrates (Fig. 28(b)). The PCE is still lower in efficiency compared to ITO based plastic substrates which have an average PCE of ∼3.7%. An important consideration for the development of organic solar cells that is often ignored is the mechanical device stability. Flexible solar cell devices using ITO and PEDOT:PSS electrodes were subjected to a cyclic bending (bend radius ∼7.4 mm) test to evaluate the mechanical stability. Figure 29(a) shows the J-V characteristics of a flexible ITO based electrode device subjected to multiple bending cycles. A decrease in Jsc and FF is observed with devices using an ITO electrode and after 300 bending cycles showed only a ∼50% PCE retention. Devices fabricated from the PEDOT:PSS electrodes show negligible degradation in Jsc and Voc , but some slight degradation in FF after multiple bend cycles (Fig. 29(b)). In contrast to the ITO devices, the PEDOT:PSS electrodes retained ∼92% of its original PCE even after 300 bending cycles (Fig 29(c)). Improving the mechanical stability will become important for commercial realization of low-cost polymer solar cells.

506

S. K. Hau et al.

Figure 29. Plot of illuminated J-V characteristics of inverted devices fabricated from flexible ITO electrode substrates under multiple bending cycles. (b) Plot of illuminated J-V characteristics of inverted devices fabricated from flexible PEDOT:PSS −5% DMSO electrode substrates under multiple bending cycles. (c) Comparison of the normalized power conversion efficiency of devices fabricated from flexible ITO and flexible PEDOT:PSS electrodes as a function of bending cycles. Adapted with permission from.69 Copyright 2009 Elsevier. (Figure available in color online)

3. Conclusions In conclusion, the inverted device architecture is a promising architecture for the development of the more ideal polymer-based solar cell when compared to the conventional device architecture. This architecture allows for solution processing techniques to deposit the various layers onto flexible substrates allowing them to be potentially processed by industrial roll-to-roll type fabrication. In addition, the inverted device architecture is more stable to ambient due the use of higher work function metals as the top electrode. These higher work function electrodes also allow for the potential for coating the top metal electrode by a printing and coating processes that will help to minimize fabrication costs. An integrated engineering approach to develop materials, devices and improve interfaces and processing to improve the performance of inverted solar cell is described. Amorphous fullerenes were processed into the BHJ active layer of inverted solar cells showing improved thermal and morphological stability compared to devices processed with PCBM. The interfaces of the metal oxides (ZnO, TiO2 ) in the inverted architecture were improved by using selfassembled monolayers which help to improve the charge transfer properties at the interface

Development of Inverted Polymer Solar Cell Architecture

507

and help to improve the bulk-heterojunction morphology leading to high efficiency solar cells. Solution processed electrodes to replace vacuum deposited metal electrodes and expensive ITO electrodes in inverted solar cells are demonstrated showing the feasibility for developing roll-to-roll type processing of solar cells. More research in the development of the inverted solar cell device architecture to further improve the efficiency and roll-to-roll type processibility is necessary in order for polymer-based solar cell to be fabricated in large-scale production.

Acknowledgements The authors appreciate the financial support through the National Science Foundation STC program under DMR-0120967, the DOE “Future Generation Photovoltaic Devices and Process” program under DE-FC36-08GO18024/A000, the Office of Naval Research program under N00014-08-1-1129, the AFOSR “Interface Engineering” program under FA9550-09-1-0426, and the World Class University (WCU) program through the National Research Foundation of Korea under the Ministry of Education, Science and Technology (R31-10035). A. K. Y. Jen thanks the Boeing-Johnson Foundation for financial support.

References 1. Brabec, C. J.; Sariciftci, N. S.; Hummelen, J. C. “Plastic solar cells,” Adv. Funct. Mater., 2001, 11(15), 15–26. 2. Ma, W. L.; Yang, C. Y.; Gong, X.; Lee, K.; Heeger, A. J. “Thermally stable, efficient polymer solar cells with nanoscale control of the interpenetrating network morphology,” Adv. Funct. Mater., 2005, 15, 1617–1622. 3. Li, G.; Shrotyriya, V.; Huang, J. S.; Yao, Y.; Moriarty, T.; Emery, K.; Yang, Y. “High-efficiency solution processable polymer photovoltaic cells by self-organization of polymer blends,” Nature Mater., 2005, 4, 864–868. 4. Peet, J.; Kim, J. Y.; Coates, N. E.; Ma, W.; Moses, D.; Heeger, A. J.; Bazan, G. C. “Efficiency enhancement in low-bandgap polymer solar cells by processing with alkane dithiols,” Nature. Mater., 2007, 6, 497–500. 5. Liang, Y. Y.; Wu, Y.; Feng, D.; Tsai, S.-T.; Son, H.-J.; Li, G.; Yu, L. “Development of new semiconducting polymers for high performance solar cells,” J. Am. Chem. Soc., 2009, 131(1), 56–57. 6. Liang, Y. Y.; Feng, D.; Wu, Y.; Tsai, S.-T.; Li, G.; Ray, C.; Yu, L. “Highly efficient solar cell polymers developed via fine-tuning of structural and electronic properties,” J. Am. Chem. Soc., 2009, 131(22), 7792–7799. 7. Wong, K. W.; Yip, H. L.; Luo, Y.; Wong, K. Y.; Lau, W. M.; Low, K. H.; Chow, H. F.; Gao, Z. Q.; Yeung, W. L.; Chang, C. C. “Blocking reactions between indium-tin oxide and poly(3,4–3thylenedioxythiophene):poly(styrene sulphonate) with a self-assembly monolayer,” Appl. Phys. Lett., 2002, 80, 2788–2790. 8. de Jong, M. P.; van IJzendoorn, L. J.; de Voigt, M. J. A. “Stability of the interface between indium-tin oxide and poly(3,4-ethylenedioxythiophene)/poly(styrenesulphonate) in polymer light-emitting diodes,” Appl. Phys. Lett., 2000, 77, 2255–2257. 9. Hains, A. W.; Marks, T. J. “High-efficiency hole extraction/electron-blocking layer to replace poly(3,4–3thylenedioxythiophene):poly(styrene sulphonate) in bulk-heterojunction polymer solar cells,” Appl. Phys. Lett., 2008, 92, 023504–023506. 10. Hau, S. K.; Yip, H. L.; Baek, N. S.; Zou, J.; O’Malley, K.; Jen, A. K. Y. “Air-stable inverted flexible polymer solar cells using zinc oxide nanoparticles as an electron selective layer,” Appl. Phys. Lett., 2008, 92, 253301–253303.

508

S. K. Hau et al.

11. Lee, K.; Kim, J. Y.; Park, S. H.; Kim, S. H.; Cho, S.; Heeger, A. J. “Air-stable polymer electronic devices,” Adv. Mater., 2007, 19, 2445–2249. 12. Kim, J. Y.; Kim, S. H.; Lee, H. H.; Lee, K.; Ma, W.; Gong, X.; Heeger, A. J. “New architecture for high-efficiency polymer photovoltaic cells using solution-based titanium oxide as an optical spacer,” Adv. Mater., 2006, 18, 572–576. 13. Yip, H.-L.; Hau, S. K.; Baek, N. S.; Ma, H.; Jen, A. K.-Y. “Polymer solar cells that use self-assembled monolayer modified ZnO/Metals as cathodes,” Adv. Mater., 2008, 20, 2376– 2382. 14. Hau, S. K.; Yip, H.-L.; Leong, K.; Jen, A. K.-Y. “Spraycoating of silver nanoparticle electrodes for inverted polymer solar cells,” Org. Electron., 2009, 10, 719–723. 15. Girotto, C.; Rand, B. P.; Steudel, S.; Genoe, J.; Heremans, P. “Nanoparticle-based spray-coated silver top contacts for efficient polymer solar cells,” Org. Electron., 2009, 10, 735–740. 16. Krebs, F. C. “Air stable polymer photovoltaics based on process free from vacuum steps and fullerenes,” Solar Energy Mater. and Solar Cells., 2008, 92, 715–726. 17. Glatthaar, M.; Niggemann, M.; Zimmermann, B.; Lewer, P.; Riede, M.; Hinsch, A.; Luther, J. “Organic solar cells using inverted layer sequence,” Thin Solid Films., 2005, 491, 298–300. 18. Chen, F. C.; Wu, J. L.; Lee, C. L.; Huang, W. C.; Chen, H. M. P.; Chen, W. C. “Flexible polymer photovoltaic devices prepared with inverted structures on metal foils,” IEEE Electron. Device Lett., 2009, 30, 727–729. 19. Hsiao, Y. S.; Chen, C. P.; Chao, C. H.; Whang, W. T. “All-solution-processed inverted polymer solar cells on granular surface-nickelized polyimide,” Org. Electron., 2009, 10, 551–561. 20. Li, G.; Chu, C. W.; Shrotriya, V.; Huang, J.; Yang, Y. “Efficient inverted polymer solar cells,” Appl. Phys. Lett., 2006, 88, 253503–253505. 21. Liao, H. H.; Chen, L. M.; Xu, Z.; Li, G.; Yang, Y. “Highly efficient inverted polymer solar cell by low temperature annealing of Cs2 CO3 interlayer,” Appl. Phys. Lett., 2008, 92, 173303–173305. 22. Zhao, D. W.; Liu, P.; Sun, X. W.; Tan, S. T.; Ke, L.; Kyaw, A. K. K. “An inverted organic solar cell with an ultrathin Ca electron-transporting layer and MoO3 hole-transporting layer,” Appl. Phys. Lett., 2009, 95, 153304–153306. 23. White, M. S.; Olson, D. C.; Shaheen, S. E.; Kopidakis, N.; Ginley, D. S. “Inverted bulkheterojunction organic photovoltaic device using solution-derived ZnO underlayer,” Appl. Phys. Lett., 2006, 87, 143517–143519. 24. Hau, S. K.; Yip, H. L.; Ma, H.; Jen, A. K. Y. “High performance ambient processed inverted polymer solar cells through interfacial modification with a fullerene self-assembled monolayer,” Appl. Phys. Lett., 2008, 93, 233304–233306. 25. Waldauf, C.; Morana, M.; Denk, P.; Schilinsky, P.; Coakley, K.; Choulis, S. A.; Brabec, C. J. “Highly efficient inverted organic photovoltaics using solution based titanium oxide as electron selective contact,” Appl. Phys. Lett., 2006, 89, 233517–233519. 26. Steim, R.; Choulis, S. A.; Schilinsky, P.; Brabec, C. J. “Interface modification for highly efficient organic photovoltaics,” Appl. Phys. Lett., 2008, 92, 093303–093306. 27. Hau, S. K.; Yip, H. L.; Acton, O.; Baek, N. S.; Ma, H.; Jen, A. K. Y. “Interfacial modification to improve inverted polymer solar cells,” J. Mater. Chem., 2008, 18, 5113–5119. 28. Kyaw, A. K. K.; Sun, X. W.; Jiang, C. Y.; Lo, G. Q.; Zhao, D. W.; Kwong, D. L. “An inverted organic solar cell employing a sol-gel derived ZnO electron selective layer and thermal evaporated MoO3 hole selective layer,” Appl. Phys. Lett., 2008, 93, 221107–221109. 29. Tao, C.; Ruan, S. P.; Zhang, X. D.; Xie, G. H.; Shen, L.; Kong, X. Z.; Dong, W.; Liu, C. X.; Chen, W. Y. “Performance improvement of inverted polymer solar cells with different top electrodes by introducing a MoO3 buffer layer,” Appl. Phys. Lett., 2008, 93, 193307–193309. 30. Schmidt, H.; Flugge, H.; Winkler, T.; Bulow, T.; Riedl, T.; Kowalsky, W. “Efficient semitransparent inverted organic solar cells with indium tin oxide top electrode,” Appl. Phys. Lett., 2009, 94, 243302–243304. 31. Tao, C.; Ruan, S. P.; Xie, G. H.; Kong, X. Z.; Shen, L.; Meng, F. X.; Liu, C. X.; Zhang, X. D.; Dong, W.; Chen, W. Y. “Role of tungsten oxide in inverted polymer solar cells,” Appl. Phys. Lett., 2009, 94, 043311–043313.

Development of Inverted Polymer Solar Cell Architecture

509

32. Chan, M. Y.; Lee, C. S.; Lai, S. L.; Fund, M. K.; Wong, F. L.; Sun, H. Y.; Lau, K. M. Lau; Lee, S. T. “Efficient organic photovoltaic devices using a combination of exciton blocking layer and anodic buffer layer,” J. Appl. Phys., 2006, 100, 094506–094509. 33. Huang, J. S.; Chou, C. Y.; Liu, M. Y.; Tsai, K. H.; Lin, W. H.; Lin, C. F. “Solution-processed vanadium oxide as an anode interlayer for inverted polymer solar cells hybridized with ZnO nanorods,” Org. Electron., 2009, 10, 1060–1065. 34. Chu, C. W.; Chen, C. W.; Li, S. H.; Wu, E. H. E.; Yang, Y. “Integration of organic light-emitting diode and organic transistor via a tandem structure,” Appl. Phys. Lett., 2005, 86, 253503–253505. 35. Kido, J.; Matsumoto, T.; Nakada, T.; Endo, J.; Mori, K.; Kawamura, N. Kawamura; Yokoi, A., SID 03 Digest., 2003, 984, 984–979. 36. Li, C. Y.; Wen, T. C.; Guo, T. F. “Sulfonated poly(diphenylamine) as a novel hole-collecting layer in polymer photovoltaic cells,” J. Mater. Chem., 2008, 18, 4478–4482. 37. van Bavel, S.; Sourty, E.; de With, G.; Frolic, K.; Loos, J. “Relation between photoactive layer thickness, 3D morphology, and device performance in P3HT/PCBM bulk-heterojunction solar cells,” Macromolecules., 2009, 42, 7396–7403. 38. Kumar, A.; Li, G.; Hong, Z. R.; Yang, Y. “High efficiency polymer solar cells with vertically modulated nanoscale morphology,” Nanotechnology., 2009, 20, 165202–165205. 39. Campoy-Quiles, M.; Ferenczi, T.; Agostinelli, T.; Etchegoin, P. G.; Kim, Y.; Anthopoulos, T. D.; Stavrinou, P. N.; Bradley, D. D. C. Bradley; Nelson, J. “Morphology evolution via selforganization and lateral, and vertical diffusion in polymer: fullerene solar cell blends,” Nature Mater., 2008, 7, 158–164. 40. Xu, Z.; Chen, L. M.; Yang, G. W.; Huang, C. H.; Hou, J. H.; Wu, Y.; Li, G.; Hsu, C. S.; Yang, Y. “Vertical phase separation in poly(3-hexylthiophene):fullerene derivative blends and its advantage for inverted structure solar cells,” Adv. Funct. Mater., 2009, 19, 1227–1234. 41. Germack, D. S.; Chan, C. K.; Hamadani, B. H.; Richter, L. J.; Fischer, D. A.; Gundlach, D. J.; DeLongchamp, D. M. “Substrate-dependent interface composition and charge transport in films for organic photovoltaics,” Appl. Phys. Lett., 2009, 94, 233303–233305. 42. Lloyd, M. T.; Olson, D. C.; Lu, P.; Fang, E.; Moore, D. L.; White, M. S.; Reese, M. O.; Ginley, D. S.; Hsu, J. W. P. “Impact of contact evolution on the shelf life of organic solar cells,” J. Mater. Chem., 2009, 19, 7638–7642. 43. Kim, J. B.; Kim, C. S.; Kim, Y. S.; Loo, Y. L. “Oxidation of silver electrodes induces transition from conventional to inverted photovoltaic characteristics in polymer solar cells,” Appl. Phys. Lett., 2009, 95, 183301–183303. 44. Swinnen, A.; Haeldermans, I.; vande Ven, M.; D’Haen, J.; Vanhoyland, G.; Aresu, S.; D’Olieslaeger, M. D.; Manca, J. “Tuning the dimensions of C60 -based needlelike crystals in blended thin films,” Adv. Funct. Mater., 2005, 16, 760–765. 45. Muller, C.; Ferenczi, T. A. M.; Campoy-Quiles, M.; Frost, J. M.; Bradley, D. D. C.; Smith, P.; Stingelin-Stutzmann, N.; Nelson, J. “Binary organic photovoltaic blends: a simple rationale for optimum compositions,” Adv. Mater., 2008, 20, 3510–3515. 46. Woo, C. H.; Thompson, B. C.; Kim, B. J.; Toney, M. F.; Frechet, J. M. “The influence of poly(3hexylthiophene) regioregularity on fullerene-composite solar cell performance,” J. Am. Chem. Soc., 2008, 130, 16324–16329. 47. Sivula, K.; Luscombe, C. K.; Thompson, B. C.; Frechet, J. M. “Enhancing the thermal stability of polythiophene:fullerene solar cells by decreasing effective polymer regioregularity,” J. Am. Chem. Soc., 2006, 128, 13988–13989. 48. Sivula, K.; Ball, Z. T.; Watanabe, N.; Frechet, J. M. “Amphiphilic diblock copolymer compatibilizers and their effect on the morphology and performance of polythiophene:fullerene solar cells,” Adv. Mater., 2006, 18, 206–210. 49. Zhou, Z. Y.; Chen, X. W.; Holdcroft, S. “Stabilizing bicontinuous nanophase segregation in pi CP-C60 donor-acceptor blends,” J. Am. Chem. Soc., 2008, 130, 11711–11718. 50. Zhang, Y.; Yip, H. L.; Acton, O.; Hau, S. K.; Huang, F.; Jen, A. K.-Y. “A simple and effective way of achieving highly efficient and thermally stable bulk-heterojunction polymer solar cells using amorphous fullerene derivatives as electron acceptor,” Chem. Mater., 2009, 21, 2598–2600.

510

S. K. Hau et al.

51. Hummelen, J. C.; Knight, B. W.; LePeq, F.; Wudl, F.; Yao, J.; Wilkins, C. L. “Preparation and characterization of fulleroid and methanofullerene derivatives,” J. Org. Chem., 1995, 60, 532–538. 52. Zheng, L. P.; Zhou, Q. M.; Deng, X. Y.; Yuan, M.; Yu, G.; Cao, Y. “Methanofullerenes used as electron acceptors in polymer photovoltaic devices,” J. Phys. Chem. B., 2004, 108, 11921–11926. 53. Yang, C.; Kim, J. Y.; Cho, S.; Lee, J. K.; Heeger, A. J.; Wudl, F. “Functionalized methanofullerenes used as n-type materials in bulk-heterojunction polymer solar cells and in field-effect transistors,” J. Am. Chem. Soc., 2008, 130, 6444–6450. 54. Chua, L. L.; Zaumseil, J.; Chang, J.-F.; Ou, E. C.-W.; Ho, P. K.-H.; Sirringhaus, H.; Friend, R. H. “General observation of n-type field-effect behaviour in organic semiconductors,” Nature., 2005, 434, 194–199. 55. Khodabakhsh, S.; Sanderson, B. M.; Nelson, J.; Jones, T. S. “Using self-assembling dipole molecules to improve charge collection in molecular solar cells,” Adv. Func. Mater., 2006, 16, 95–100. 56. Armstrong, N. R.; Carter, C.; Donley, C.; Simmonds, A.; Lee, P.; Brumbach, M.; Kippelen, B.; Domercq, B.; Yoo, S. “Interface modification of ITO thin films: Organic photovoltaic cells,” Thin Solid Films., 2003, 445, 342–352. 57. Kim, J. S.; Park, J. H.; Lee, J. H.; Jo, J.; Kim, D. -Y.; Cho, K. “Control of the electrode work function and active layer morphology via surface modification of indium tin oxide for high efficiency organic photovoltaics,” Appl. Phys. Lett., 2007, 91, 112111–112113. 58. Goh, C.; Scully, S. R.; McGehee, M. D. “Effects of molecular interface modification in hybrid organic-inorganic photovoltaic cells,” J. Appl. Phys., 2007, 101, 114503–114514. 59. Moser, J.; Punchihewa, S.; Infelta, P. P.; Gratzel, M. “Surface complexation of colloidal semiconductors strongly enhances interfacial electron-transfer rates,” Langmuir., 1991, 7, 3012–3018. 60. Sariciftci, N. S.; Smilowitz, L.; Heeger, A. J.; Wudl, F. “Photoinduced electron-transfer from a conducting polymer to buckminsterfullerene,” Science., 1992, 258, 1474–1476. 61. Kim, D. H.; Jang, Y.; Park, Y. D.; Cho, K. “Surface-induced conformational changes in poly(3hexylthiophene) monolayer films,” Langmuir., 2005, 21, 3203–3206. 62. Morana, M.; Koers, P.; Waldauf, C.; Koppe, M.; Muehlbacher, D.; Denk, P.; Scharber, M.; Waller, D.; Brabec, C. “Organic field-effect devices as tool to characterize the bipolar transport in polymer-fullerene blends: the case of P3HT:PCBM,” Adv. Funct. Mater., 2007, 17, 3274–3283. 63. Kim, Y.; Choulis, S. A.; Nelson, J.; Bradley, D. D. C.; Cook, S.; Durrant, J. R. “Device annealing effect in organic solar cells with blends of regioregular poly(3-hexylthiophene) and soluble fullerene,” Appl. Phys. Lett., 2005, 86, 063502–063504. 64. Beek, W. J. E.; Wienk, M. M.; Kemerink, M.; Yang, X.; Janssen, R. A. J. “Hybrid zinc oxide conjugated polymer bulk heterojunction solar cells,” J. Phys. Chem. B. 2005, 109, 9505–9516. 65. Hau, S. K.; O’Malley, K. M.; Cheng, Y. J.; Yip, H. L.; Ma, H.; Jen, A. K.-Y. “Optimization of active layer and anode electrode for high-performance inverted bulk-heterojunction solar cells,” IEEE J. Selected Topics in Quantum Electronics., 2010, 99 (in press). 66. Mihailetchi, V. D.; Blom, P. W. M.; Hummelen, J. C.; Rispens, M. T. “Cathode dependence of the open-circuit voltage of polymer:fullerene bulk heterojunction solar cells,” J. Appl. Phys., 2003, 94, 6849–6854. 67. Reese, M. O.; White, M. S.; Rumbles, G.; Ginley, D. S.; Shaheen, S. E. “Optimal negative electrodes for poly(3-hexyltiophene):[6,6]-phenyl C61 -butyric methyl ester bulk heterojunction photovoltaic devices,” Appl. Phys. Lett., 2008, 92, 053307–053309. 68. J¨onsson, S. K. M.; Salaneck, W. R.; Fahlman, M. “X-ray photoelectron spectroscopy study of the metal/polymer contacts involving aluminum and poly(3,4-ethylenedioxythiophene)poly(styrenesulfonic acid) derivatives,” J. Mater. Res., 2003, 18, 1219–1226. 69. Hau, S. K.; Yip, H. L.; Zou, J.; Jen, A. K.-Y. “Indium tin oxide-free semi-transparent inverted polymer solar cells using conducting polymer as both bottom and top electrodes,” Org. Electron., 2009, 10, 1401–1407.